self unpacking zip, quick and dirty code

Cargo.toml

@@ -4,6 +4,5 @@ members = [
     "apk_frauder",
     "androscalpel_serializer",
     "androscalpel_serializer_derive",
-    "androscalpel", "androscalpel_platform_api_list",
+    "androscalpel",
 ]
-resolver = "2"

README.md

@@ -1,17 +0,0 @@
-# Androscalpel
-
-Androscalpel is a rust crate to manipulate Android application bytecode (dalvik).
-
-This developed between 2022 and 2025 by Jean-Marie Mineau for their PhD thesis, and release under the GPLv3 licence with the permission of CentraleSupelec.
-
-# Documentation
-
-The documentation can be generated with `cargo doc`.
-It will be generated at `target/doc/androscalpel/index.html`
-
-Right now, the recompilation process is a little complexe, an example can be found [here](./androscalpel/examples/count_ins_and_genapk.rs).
-
-# Note
-
-The Dalvik v41 format (concatenated DEX files) is not currently supported.
-Support is planned in the future.

TODO.md

@@ -1,12 +1,8 @@
 - sanity checks
-- SANITY CHECK (https://cs.android.com/android/platform/superproject/main/+/main:art/libdexfile/dex/dex_file_verifier.cc if check failed, .dex is not loaded but apk do not crash !!!)
-- V41 https://cs.android.com/android/platform/superproject/main/+/main:art/libdexfile/dex/dex_file_verifier.cc;drc=e8da7cd1d0e7d3535c82f8d05adcef3edd43cd40;l=634
 - tests
 - https://source.android.com/docs/core/runtime/dex-format#system-annotation
 - goto size computation
-- Check label duplication (maybe allow identical labels at the same address only?)
 - no nop when no payload
+- option to get label at every code addresses
 - name register / parameters
-- fix flake
-- fix python binding or remove
-- clean repo
+- ord in python

androscalpel/Cargo.toml

@@ -1,48 +1,22 @@
 [package]
 name = "androscalpel"
 version = "0.1.0"
-edition = "2024"
-license = "GPL-3.0-or-later"
+edition = "2021"
 
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 [lib]
 name = "androscalpel"
-crate-type = ["cdylib", "lib"]
+crate-type = ["cdylib"]
 
 [dependencies]
 adler = "1.0.2"
 androscalpel_serializer = { version = "0.1.0", path = "../androscalpel_serializer" }
-androscalpel_platform_api_list = { version = "0.1.0", path = "../androscalpel_platform_api_list", optional = true }
 anyhow = { version = "1.0.75", features = ["backtrace"] }
 apk_frauder = { version = "0.1.0", path = "../apk_frauder" }
 log = "0.4.20"
-pyo3 = { version = "0.23.4", features = ["anyhow", "abi3-py38", "extension-module"], optional = true}
-pyo3-log = { version = "0.12.1", optional = true}
+pyo3 = { version = "0.20.0", features = ["anyhow"] }
+pyo3-log = "0.8.3"
 rayon = "1.9.0"
 serde = { version = "1.0.195", features = ["derive"] }
 serde_json = "1.0.111"
 sha1 = "0.10.6"
-zip = {version = "2.2.2", optional = true}
-
-[dev-dependencies]
-pretty_assertions = "1.4.1"
-# For examples
-clap = { version = "4.5.27", features = ["derive"] }
-env_logger = "0.11.6"
-
-[features]
-default = ["code-analysis", "platform-list"]
-# TODO: need refactoring to https://github.com/PyO3/pyo3/issues/2935#issuecomment-2560930677 or cfg_eval https://github.com/rust-lang/rust/issues/82679
-python = ["pyo3", "pyo3-log"] # Currently not supported
-external-zip-reader = ["zip"]
-platform-list = ["androscalpel_platform_api_list"]
-code-analysis = []
-
-[[example]]
-name = "list-method"
-
-[[example]]
-name = "count_ins_and_genapk"
-
-[[example]]
-name = "dump_cfg"

androscalpel/examples/count_ins_and_genapk.rs

@@ -1,56 +0,0 @@
-use androscalpel::{Apk, Instruction, Result, VisitableMut, VisitorMut};
-use apk_frauder::replace_dex_unsigned;
-
-use std::collections::HashMap;
-use std::env;
-use std::io::Cursor;
-
-#[derive(Default)]
-struct InsCounter {
-    pub n: u64,
-}
-impl VisitorMut for InsCounter {
-    fn visit_instruction(&mut self, ins: Instruction) -> Result<Instruction> {
-        if !ins.is_pseudo_ins() {
-            self.n += 1;
-        }
-        ins.default_visit_mut(self)
-    }
-}
-
-fn main() {
-    let args: Vec<_> = env::args().collect();
-    assert!(args.len() > 2, "usage: {} input.apk output.apk", args[0]);
-
-    let mut cnt = InsCounter::default();
-    let apk = cnt
-        .visit_apk(Apk::load_apk_path((&args[1]).into(), false, false).unwrap())
-        .unwrap();
-    println!("Nb INS: {}", cnt.n);
-
-    // This should be streamlined
-    let mut dex_files = vec![];
-    let mut files = apk.gen_raw_dex().unwrap();
-    let mut i = 0;
-    loop {
-        let name = if i == 0 {
-            "classes.dex".into()
-        } else {
-            format!("classes{}.dex", i + 1)
-        };
-        if let Some(file) = files.remove(&name) {
-            dex_files.push(Cursor::new(file))
-        } else {
-            break;
-        }
-        i += 1;
-    }
-
-    replace_dex_unsigned(
-        &args[1],
-        &args[2],
-        &mut dex_files,
-        None::<HashMap<_, Option<Cursor<&[u8]>>>>,
-    )
-    .unwrap();
-}

androscalpel/examples/dump_cfg.rs

@@ -1,31 +0,0 @@
-use std::fs::File;
-use std::path::PathBuf;
-
-use androscalpel::{Apk, IdMethod, MethodCFG};
-
-use clap::Parser;
-
-#[derive(Parser, Debug)]
-#[command(version, about, long_about = None, arg_required_else_help = true)]
-struct Cli {
-    #[arg(short, long)]
-    apk: PathBuf,
-    #[arg(short, long)]
-    method: String,
-}
-
-fn main() {
-    env_logger::init();
-    let cli = Cli::parse();
-    let mut apk = Apk::load_apk(File::open(&cli.apk).unwrap(), |_, _, _| None, false).unwrap();
-    let mid = IdMethod::from_smali(&cli.method).unwrap();
-    let class = apk.get_class_mut(&mid.class_).unwrap();
-    let method = if let Some(method) = class.virtual_methods.get(&mid) {
-        method
-    } else {
-        class.direct_methods.get(&mid).unwrap()
-    };
-    let cfg = MethodCFG::new(method).unwrap();
-    print!("{}", cfg.to_dot(true));
-}
-

androscalpel/examples/list-method.rs

@@ -1,20 +0,0 @@
-use androscalpel::{Apk, Method, Result, SmaliName, Visitable, Visitor};
-use std::env;
-
-#[derive(Default)]
-struct MethodPrinter {}
-impl Visitor for MethodPrinter {
-    fn visit_method(&mut self, method: &Method) -> Result<()> {
-        println!("Method: {}", method.descriptor.try_to_smali()?);
-        method.default_visit(self)?;
-        Ok(())
-    }
-}
-
-fn main() {
-    let args: Vec<_> = env::args().collect();
-    assert!(args.len() > 1, "Need one argument");
-
-    let apk = Apk::load_apk_path((&args[1]).into(), false, false).unwrap();
-    MethodPrinter::default().visit_apk(&apk).unwrap();
-}

androscalpel/pyproject.toml

@@ -0,0 +1,16 @@
+[build-system]
+requires = ["maturin>=1.2,<2.0"]
+build-backend = "maturin"
+
+[project]
+name = "androscalpel"
+requires-python = ">=3.7"
+classifiers = [
+    "Programming Language :: Rust",
+    "Programming Language :: Python :: Implementation :: CPython",
+    "Programming Language :: Python :: Implementation :: PyPy",
+]
+
+
+[tool.maturin]
+features = ["pyo3/extension-module"]

androscalpel/src/annotation.rs

@@ -2,38 +2,41 @@
 use serde::{Deserialize, Serialize};
 use std::collections::{HashMap, HashSet};
 
-#[cfg(feature = "python")]
 use pyo3::prelude::*;
 
 use crate::hashmap_vectorize;
 use crate::{
-    DexString, IdField, IdMethod, IdMethodType, MethodHandle, Result, Visitable, VisitableMut,
-    Visitor, VisitorMut, dex_id::IdType, value::DexValue,
+    dex_id::IdType, value::DexValue, DexString, IdField, IdMethod, IdMethodType, MethodHandle,
+    Result,
 };
 
 /// Annotation with a visibility
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, PartialEq, Deserialize, Serialize)]
 pub struct DexAnnotationItem {
     // TODO: the get/set will probably be wonky on the python side when edditing
     /// The actual annotation
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub annotation: DexAnnotation,
     // TODO: enforce exclusivity
     /// If the annotation visibility is set to build
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub visibility_build: bool,
     /// If the annotation visibility is set to runtime
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub visibility_runtime: bool,
     /// If the annotation visibility is set to system
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub visibility_system: bool,
 }
 
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl DexAnnotationItem {
-    #[cfg_attr(feature = "python", new)]
+    pub fn __eq__(&self, other: &Self) -> bool {
+        self == other
+    }
+
+    #[new]
     pub fn new(annotation: DexAnnotation) -> Self {
         Self {
             annotation,
@@ -102,44 +105,34 @@ impl DexAnnotationItem {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 }
 
-impl<V: Visitor> Visitable<V> for DexAnnotationItem {
-    fn default_visit(&self, v: &mut V) -> Result<()> {
-        v.visit_annotation(&self.annotation)
-    }
-}
-impl<V: VisitorMut> VisitableMut<V> for DexAnnotationItem {
-    fn default_visit_mut(self, v: &mut V) -> Result<Self> {
-        Ok(Self {
-            annotation: v.visit_annotation(self.annotation)?,
-            ..self
-        })
-    }
-}
-
 /// An annotation.
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, PartialEq, Deserialize, Serialize)]
 pub struct DexAnnotation {
     // TODO: check the relation between type and encoded_value.
     /// The type of the annotation.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub type_: IdType,
     // TODO: the get/set will probably be wonky on the python side when edditing
     /// The annotation elements.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     #[serde(with = "hashmap_vectorize")]
     pub elements: HashMap<DexString, DexValue>, // TODO: check MemberName syntax?
 }
 
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl DexAnnotation {
-    #[cfg_attr(feature = "python", new)]
+    pub fn __eq__(&self, other: &Self) -> bool {
+        self == other
+    }
+
+    #[new]
     pub fn new(type_: IdType, elements: HashMap<DexString, DexValue>) -> Self {
         Self { type_, elements }
     }
@@ -219,32 +212,8 @@ impl DexAnnotation {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 }
-
-impl<V: Visitor> Visitable<V> for DexAnnotation {
-    fn default_visit(&self, v: &mut V) -> Result<()> {
-        v.visit_type(&self.type_)?;
-        for (id, val) in &self.elements {
-            v.visit_string(id)?;
-            v.visit_value(val)?;
-        }
-        Ok(())
-    }
-}
-
-impl<V: VisitorMut> VisitableMut<V> for DexAnnotation {
-    fn default_visit_mut(self, v: &mut V) -> Result<Self> {
-        let type_ = v.visit_type(self.type_)?;
-        let mut elements = HashMap::new();
-        for (id, val) in self.elements {
-            let id = v.visit_string(id)?;
-            let val = v.visit_value(val)?;
-            elements.insert(id, val);
-        }
-        Ok(Self { type_, elements })
-    }
-}

androscalpel/src/class.rs

@@ -3,90 +3,94 @@
 use serde::{Deserialize, Serialize};
 use std::collections::{HashMap, HashSet};
 
-#[cfg(feature = "python")]
 use pyo3::prelude::*;
 
+use crate::hashmap_vectorize;
 use crate::{
     DexAnnotationItem, DexString, Field, IdField, IdMethod, IdMethodType, IdType, Method,
-    MethodHandle, Result, Visitable, VisitableMut, Visitor, VisitorMut,
+    MethodHandle, Result,
 };
 use androscalpel_serializer::consts::*;
 
 /// Represent an apk
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, Deserialize, Serialize, PartialEq)]
 pub struct Class {
     /// Type, format described at
     /// <https://source.android.com/docs/core/runtime/dex-format#typedescriptor>
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub descriptor: IdType,
     /// If the class is visible everywhere
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_public: bool,
     /// If the class is subclassable
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_final: bool,
     /// If the class is a 'multipy-implementable abstract class' AKA an interface
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_interface: bool,
     /// If the class is instanciable
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_abstract: bool,
     /// If the class is not directly defined in the source code
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_synthetic: bool,
     /// If the class is an annotation
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_annotation: bool,
     /// If the class is an enum
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_enum: bool,
     /// Name of the superclass, format described at
     /// <https://source.android.com/docs/core/runtime/dex-format#typedescriptor>
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub superclass: Option<IdType>,
     /// List of the interfaces that class implement, format of the interfaces
     /// name is discribed at
     /// <https://source.android.com/docs/core/runtime/dex-format#typedescriptor>
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub interfaces: Vec<IdType>,
     /// Name of the source file where this class is defined.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub source_file: Option<DexString>,
 
     /// The static fields
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
+    #[serde(with = "hashmap_vectorize")]
     pub static_fields: HashMap<IdField, Field>,
     /// The instance fields
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
+    #[serde(with = "hashmap_vectorize")]
     pub instance_fields: HashMap<IdField, Field>,
     /// The direct (static, private or constructor) methods of the class
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
+    #[serde(with = "hashmap_vectorize")]
     pub direct_methods: HashMap<IdMethod, Method>,
     /// The virtual (ie non direct) methods of the class
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
+    #[serde(with = "hashmap_vectorize")]
     pub virtual_methods: HashMap<IdMethod, Method>,
     // Do we need to distinguish direct and virtual (all the other) methods?
     // Maybe overlapping descriptor (same name, class and proto?)
     /// The annotation related to this class (note: this does not include the
     /// methods/field/parameters annotations, they are stored in the methods and fields
     /// structutres)
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub annotations: Vec<DexAnnotationItem>,
 }
 
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl Class {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(name: DexString) -> Result<Self> {
         Ok(Self {
             descriptor: IdType::new(name)?,
@@ -109,15 +113,15 @@ impl Class {
     }
 
     pub fn __str__(&self) -> String {
-        let name: String = self.descriptor.get_name().__str__();
+        let name: String = (&self.descriptor.get_name()).into();
         let file = if let Some(file) = &self.source_file {
-            let file: String = file.__str__();
+            let file: String = file.into();
             format!("  defined in {file}\n")
         } else {
             "".into()
         };
         let superclass = if let Some(spcl) = &self.superclass {
-            let spcl: String = spcl.get_name().__str__();
+            let spcl: String = spcl.get_name().into();
             format!("    extends: {spcl}\n")
         } else {
             "".into()
@@ -127,7 +131,7 @@ impl Class {
         } else {
             let mut interfaces: String = "  implements:\n".into();
             for it in &self.interfaces {
-                let it: String = it.get_name().__str__();
+                let it: String = it.get_name().into();
                 interfaces += &format!("    {it}\n");
             }
             interfaces
@@ -137,7 +141,7 @@ impl Class {
     }
 
     pub fn __repr__(&self) -> String {
-        let name: String = self.descriptor.get_name().__str__();
+        let name: String = (&self.descriptor.get_name()).into();
         format!("Class({name})")
     }
 
@@ -317,7 +321,7 @@ impl Class {
             .any(|field| field.value.is_some())
     }
 
-    /// Check if the class or its fields/methods have annotations
+    /// If the class or its fields/methods have annotations
     pub fn has_annotations(&self) -> bool {
         !self.annotations.is_empty()
             || self
@@ -331,30 +335,11 @@ impl Class {
             || self
                 .direct_methods
                 .values()
-                .any(|method| method.has_annotations())
+                .any(|field| field.has_annotations())
             || self
                 .virtual_methods
                 .values()
-                .any(|method| method.has_annotations())
-    }
-
-    /// Check if the class have a field of method with hiddenapi information
-    pub fn has_hiddenapi(&self) -> bool {
-        self.static_fields
-            .values()
-            .any(|field| field.hiddenapi.is_some())
-            || self
-                .instance_fields
-                .values()
-                .any(|field| field.hiddenapi.is_some())
-            || self
-                .direct_methods
-                .values()
-                .any(|method| method.hiddenapi.is_some())
-            || self
-                .virtual_methods
-                .values()
-                .any(|method| method.hiddenapi.is_some())
+                .any(|field| field.has_annotations())
     }
 
     /// Return the binary representation of access flags.
@@ -384,98 +369,7 @@ impl Class {
         flags
     }
 
-    /// Check if the class is a platform class (ie in the android SDK or a hidden API).
-    #[cfg(feature = "platform-list")]
-    pub fn is_platform_class(&self) -> bool {
-        self.descriptor.is_platform_class()
-    }
-}
-
-impl<V: Visitor> Visitable<V> for Class {
-    fn default_visit(&self, v: &mut V) -> Result<()> {
-        v.visit_type(&self.descriptor)?;
-        if let Some(superclass) = &self.superclass {
-            v.visit_type(superclass)?;
-        }
-        for interface in &self.interfaces {
-            v.visit_type(interface)?;
-        }
-        if let Some(source_file) = &self.source_file {
-            v.visit_string(source_file)?;
-        }
-        for (id, field) in &self.static_fields {
-            v.visit_field_id(id)?;
-            v.visit_field(field)?;
-        }
-        for (id, field) in &self.instance_fields {
-            v.visit_field_id(id)?;
-            v.visit_field(field)?;
-        }
-        for (id, meth) in &self.direct_methods {
-            v.visit_method_id(id)?;
-            v.visit_method(meth)?;
-        }
-        for (id, meth) in &self.virtual_methods {
-            v.visit_method_id(id)?;
-            v.visit_method(meth)?;
-        }
-        for annot in &self.annotations {
-            v.visit_annotation_item(annot)?;
-        }
-        Ok(())
-    }
-}
-impl<V: VisitorMut> VisitableMut<V> for Class {
-    fn default_visit_mut(self, v: &mut V) -> Result<Self> {
-        Ok(Self {
-            descriptor: v.visit_type(self.descriptor)?,
-            superclass: self
-                .superclass
-                .map(|superclass| v.visit_type(superclass))
-                .transpose()?,
-            interfaces: self
-                .interfaces
-                .into_iter()
-                .map(|interface| v.visit_type(interface))
-                .collect::<Result<_>>()?,
-            source_file: self
-                .source_file
-                .map(|source_file| v.visit_string(source_file))
-                .transpose()?,
-            static_fields: {
-                let mut static_fields = HashMap::new();
-                for (id, field) in self.static_fields.into_iter() {
-                    static_fields.insert(v.visit_field_id(id)?, v.visit_field(field)?);
-                }
-                static_fields
-            },
-            instance_fields: {
-                let mut instance_fields = HashMap::new();
-                for (id, field) in self.instance_fields.into_iter() {
-                    instance_fields.insert(v.visit_field_id(id)?, v.visit_field(field)?);
-                }
-                instance_fields
-            },
-            direct_methods: {
-                let mut direct_methods = HashMap::new();
-                for (id, meth) in self.direct_methods.into_iter() {
-                    direct_methods.insert(v.visit_method_id(id)?, v.visit_method(meth)?);
-                }
-                direct_methods
-            },
-            virtual_methods: {
-                let mut virtual_methods = HashMap::new();
-                for (id, meth) in self.virtual_methods.into_iter() {
-                    virtual_methods.insert(v.visit_method_id(id)?, v.visit_method(meth)?);
-                }
-                virtual_methods
-            },
-            annotations: self
-                .annotations
-                .into_iter()
-                .map(|annotation| v.visit_annotation_item(annotation))
-                .collect::<Result<_>>()?,
-            ..self
-        })
+    pub fn __eq__(&self, other: &Self) -> bool {
+        self == other
     }
 }

androscalpel/src/code.rs

@@ -1,16 +1,13 @@
 //! Representation of a method.
 
 use anyhow::anyhow;
-use log::debug;
 use serde::{Deserialize, Serialize};
 use std::collections::{HashMap, HashSet};
 
-#[cfg(feature = "python")]
 use pyo3::prelude::*;
 
 use crate::{
-    DexString, IdField, IdMethod, IdMethodType, IdType, Method, MethodHandle, Result, Visitable,
-    VisitableMut, Visitor, VisitorMut, ins::Instruction,
+    ins, ins::Instruction, DexString, IdField, IdMethod, IdMethodType, IdType, MethodHandle, Result,
 };
 
 // TODO: make this easy to edit/manipulate, maybe move to Method
@@ -18,86 +15,76 @@ use crate::{
 // type TmpHandlerType = (Vec<(IdType, u32)>, Option<u32>);
 
 /// The code run by a method.
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, Deserialize, Serialize)]
 pub struct Code {
     // TODO: remove and compute this value from code?
     /// The number of registers used by the code
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub registers_size: u16,
+    // TODO: what does it means? is it computable?
+    /// The number of words of incoming arguments to the method
+    #[pyo3(get)]
+    pub ins_size: u16,
+    // TODO: what does it means? is it computable?
+    /// The number of words of outgoing argument space
+    #[pyo3(get)]
+    pub outs_size: u16,
+    // TODO: implement
+    /// The debug info
+    #[pyo3(get)]
+    pub debug_info: (u32, Vec<u8>), // Should be stripped, copying like this just don't work
     /// The names of the parameters if given
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub parameter_names: Option<Vec<Option<DexString>>>,
     /// The instructions.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub insns: Vec<Instruction>,
 }
 
-// TODO reimplement PartialEq: label should become address independant
 impl PartialEq for Code {
     fn eq(&self, other: &Self) -> bool {
         let comparable_self = self.semantic_comparable().unwrap();
         let comparable_other = other.semantic_comparable().unwrap();
         (comparable_self.registers_size == comparable_other.registers_size)
+            && (comparable_self.ins_size == comparable_other.ins_size)
+            && (comparable_self.outs_size == comparable_other.outs_size)
+            && (comparable_self.debug_info == comparable_other.debug_info)
             && (comparable_self.insns == comparable_other.insns)
     }
 }
 
-#[cfg_attr(feature = "python", pymethods)]
+// TODO reimplement PartialEq: label should become address independant
+
+#[pymethods]
 impl Code {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
-    #[cfg_attr(feature = "python", pyo3(signature = (registers_size, ins_size, outs_size, insns, parameter_names=None)))]
+    #[new]
     pub fn new(
         registers_size: u16,
+        ins_size: u16,
+        outs_size: u16,
         insns: Vec<Instruction>,
         parameter_names: Option<Vec<Option<DexString>>>,
     ) -> Self {
         Self {
             registers_size,
+            ins_size,
+            outs_size,
             insns,
             parameter_names,
+            debug_info: (0, vec![]),
         }
     }
 
-    /// Compute the `ins_size` field. This is the number of register parameters, including the
-    /// `this` parameter for non-static methods.
-    /// This information is stored in the code item in dex files, but is not computable from the code
-    /// (as opposed to `outs_size`). The [`Method`] struct is needed to compute it.
-    pub fn ins_size(&self, method: &Method) -> u16 {
-        method.ins_size()
-    }
-    /// Compute the `outs_size` field. This is the number of registers needed to call other
-    /// function.
-    pub fn outs_size(&self) -> u16 {
-        let mut outs = 0;
-        for ins in &self.insns {
-            match ins {
-                Instruction::InvokeVirtual { args, .. }
-                | Instruction::InvokeSuper { args, .. }
-                | Instruction::InvokeDirect { args, .. }
-                | Instruction::InvokeStatic { args, .. }
-                | Instruction::InvokeInterface { args, .. }
-                | Instruction::InvokePolymorphic { args, .. }
-                | Instruction::InvokeCustom { args, .. } => {
-                    if args.len() > outs {
-                        outs = args.len();
-                    }
-                }
-                _ => (),
-            }
-        }
-        outs as u16
-    }
-
     pub fn __str__(&self) -> String {
         self.__repr__()
     }
@@ -114,8 +101,10 @@ impl Code {
             strings.extend(ins.get_all_strings());
         }
         if let Some(names) = &self.parameter_names {
-            for name in names.iter().flatten() {
-                strings.insert(name.clone());
+            for name in names {
+                if let Some(name) = name {
+                    strings.insert(name.clone());
+                }
             }
         }
         strings
@@ -167,55 +156,59 @@ impl Code {
         handles
     }
 
+    pub fn __eq__(&self, other: &Self) -> bool {
+        self == other
+    }
+
     /// Return all the labels used by instrutions in the code.
     pub fn get_referenced_label(&self) -> HashSet<String> {
         let mut used_labels = HashSet::new();
         for ins in &self.insns {
             match ins {
-                Instruction::Goto { label } => {
+                Instruction::Goto(ins::Goto { label }) => {
                     used_labels.insert(label.clone());
                 }
-                Instruction::IfEq { label, .. } => {
+                Instruction::IfEq(ins::IfEq { label, .. }) => {
                     used_labels.insert(label.clone());
                 }
-                Instruction::IfNe { label, .. } => {
+                Instruction::IfNe(ins::IfNe { label, .. }) => {
                     used_labels.insert(label.clone());
                 }
-                Instruction::IfLt { label, .. } => {
+                Instruction::IfLt(ins::IfLt { label, .. }) => {
                     used_labels.insert(label.clone());
                 }
-                Instruction::IfGe { label, .. } => {
+                Instruction::IfGe(ins::IfGe { label, .. }) => {
                     used_labels.insert(label.clone());
                 }
-                Instruction::IfGt { label, .. } => {
+                Instruction::IfGt(ins::IfGt { label, .. }) => {
                     used_labels.insert(label.clone());
                 }
-                Instruction::IfLe { label, .. } => {
+                Instruction::IfLe(ins::IfLe { label, .. }) => {
                     used_labels.insert(label.clone());
                 }
-                Instruction::IfEqZ { label, .. } => {
+                Instruction::IfEqZ(ins::IfEqZ { label, .. }) => {
                     used_labels.insert(label.clone());
                 }
-                Instruction::IfNeZ { label, .. } => {
+                Instruction::IfNeZ(ins::IfNeZ { label, .. }) => {
                     used_labels.insert(label.clone());
                 }
-                Instruction::IfLtZ { label, .. } => {
+                Instruction::IfLtZ(ins::IfLtZ { label, .. }) => {
                     used_labels.insert(label.clone());
                 }
-                Instruction::IfGeZ { label, .. } => {
+                Instruction::IfGeZ(ins::IfGeZ { label, .. }) => {
                     used_labels.insert(label.clone());
                 }
-                Instruction::IfGtZ { label, .. } => {
+                Instruction::IfGtZ(ins::IfGtZ { label, .. }) => {
                     used_labels.insert(label.clone());
                 }
-                Instruction::IfLeZ { label, .. } => {
+                Instruction::IfLeZ(ins::IfLeZ { label, .. }) => {
                     used_labels.insert(label.clone());
                 }
-                Instruction::Try {
+                Instruction::Try(ins::Try {
                     end_label,
                     handlers,
                     default_handler,
-                } => {
+                }) => {
                     used_labels.insert(end_label.clone());
                     for (_, label) in handlers {
                         used_labels.insert(label.clone());
@@ -224,7 +217,7 @@ impl Code {
                         used_labels.insert(label.clone());
                     }
                 }
-                Instruction::Switch { branches, .. } => {
+                Instruction::Switch(ins::Switch { branches, .. }) => {
                     for label in branches.values() {
                         used_labels.insert(label.clone());
                     }
@@ -248,8 +241,8 @@ impl Code {
 
         for ins in &self.insns {
             match ins {
-                Instruction::Label { name } => {
-                    if !used_labels.contains(name) {
+                Instruction::Label(ins::Label { name }) => {
+                    if used_labels.get(name).is_none() {
                         continue;
                     }
                     let new_label_id = if last_ins_was_a_label {
@@ -267,8 +260,8 @@ impl Code {
         }
 
         for label in &used_labels {
-            if !new_labels.contains_key(label) {
-                debug!("{label} use but not in new_labels");
+            if new_labels.get(label).is_none() {
+                println!("{label} use but not in new_labels");
             }
         }
 
@@ -276,191 +269,183 @@ impl Code {
         last_ins_was_a_label = false;
         for ins in self.insns.iter().cloned() {
             match ins {
-                Instruction::Goto { label } => {
+                Instruction::Goto(mut instr) => {
                     last_ins_was_a_label = false;
-                    let label = new_labels
-                        .get(&label)
+                    instr.label = new_labels
+                        .get(&instr.label)
                         .ok_or(anyhow!(
                             "Internal error: {} not found in renamed label",
-                            label
+                            instr.label
                         ))?
                         .clone();
-                    new_insns.push(Instruction::Goto { label });
+                    new_insns.push(Instruction::Goto(instr));
                 }
-                Instruction::IfEq { label, a, b } => {
+                Instruction::IfEq(mut instr) => {
                     last_ins_was_a_label = false;
-                    let label = new_labels
-                        .get(&label)
+                    instr.label = new_labels
+                        .get(&instr.label)
                         .ok_or(anyhow!(
                             "Internal error: {} not found in renamed label",
-                            label
+                            instr.label
                         ))?
                         .clone();
-                    new_insns.push(Instruction::IfEq { label, a, b });
+                    new_insns.push(Instruction::IfEq(instr));
                 }
-                Instruction::IfNe { label, a, b } => {
+                Instruction::IfNe(mut instr) => {
                     last_ins_was_a_label = false;
-                    let label = new_labels
-                        .get(&label)
+                    instr.label = new_labels
+                        .get(&instr.label)
                         .ok_or(anyhow!(
                             "Internal error: {} not found in renamed label",
-                            label
+                            instr.label
                         ))?
                         .clone();
-                    new_insns.push(Instruction::IfNe { label, a, b });
+                    new_insns.push(Instruction::IfNe(instr));
                 }
-                Instruction::IfLt { label, a, b } => {
+                Instruction::IfLt(mut instr) => {
                     last_ins_was_a_label = false;
-                    let label = new_labels
-                        .get(&label)
+                    instr.label = new_labels
+                        .get(&instr.label)
                         .ok_or(anyhow!(
                             "Internal error: {} not found in renamed label",
-                            label
+                            instr.label
                         ))?
                         .clone();
-                    new_insns.push(Instruction::IfLt { label, a, b });
+                    new_insns.push(Instruction::IfLt(instr));
                 }
-                Instruction::IfGe { label, a, b } => {
+                Instruction::IfGe(mut instr) => {
                     last_ins_was_a_label = false;
-                    let label = new_labels
-                        .get(&label)
+                    instr.label = new_labels
+                        .get(&instr.label)
                         .ok_or(anyhow!(
                             "Internal error: {} not found in renamed label",
-                            label
+                            instr.label
                         ))?
                         .clone();
-                    new_insns.push(Instruction::IfGe { label, a, b });
+                    new_insns.push(Instruction::IfGe(instr));
                 }
-                Instruction::IfGt { label, a, b } => {
+                Instruction::IfGt(mut instr) => {
                     last_ins_was_a_label = false;
-                    let label = new_labels
-                        .get(&label)
+                    instr.label = new_labels
+                        .get(&instr.label)
                         .ok_or(anyhow!(
                             "Internal error: {} not found in renamed label",
-                            label
+                            instr.label
                         ))?
                         .clone();
-                    new_insns.push(Instruction::IfGt { label, a, b });
+                    new_insns.push(Instruction::IfGt(instr));
                 }
-                Instruction::IfLe { label, a, b } => {
+                Instruction::IfLe(mut instr) => {
                     last_ins_was_a_label = false;
-                    let label = new_labels
-                        .get(&label)
+                    instr.label = new_labels
+                        .get(&instr.label)
                         .ok_or(anyhow!(
                             "Internal error: {} not found in renamed label",
-                            label
+                            instr.label
                         ))?
                         .clone();
-                    new_insns.push(Instruction::IfLe { label, a, b });
+                    new_insns.push(Instruction::IfLe(instr));
                 }
-                Instruction::IfEqZ { label, a } => {
+                Instruction::IfEqZ(mut instr) => {
                     last_ins_was_a_label = false;
-                    let label = new_labels
-                        .get(&label)
+                    instr.label = new_labels
+                        .get(&instr.label)
                         .ok_or(anyhow!(
                             "Internal error: {} not found in renamed label",
-                            label
+                            instr.label
                         ))?
                         .clone();
-                    new_insns.push(Instruction::IfEqZ { label, a });
+                    new_insns.push(Instruction::IfEqZ(instr));
                 }
-                Instruction::IfNeZ { label, a } => {
+                Instruction::IfNeZ(mut instr) => {
                     last_ins_was_a_label = false;
-                    let label = new_labels
-                        .get(&label)
+                    instr.label = new_labels
+                        .get(&instr.label)
                         .ok_or(anyhow!(
                             "Internal error: {} not found in renamed label",
-                            label
+                            instr.label
                         ))?
                         .clone();
-                    new_insns.push(Instruction::IfNeZ { label, a });
+                    new_insns.push(Instruction::IfNeZ(instr));
                 }
-                Instruction::IfLtZ { label, a } => {
+                Instruction::IfLtZ(mut instr) => {
                     last_ins_was_a_label = false;
-                    let label = new_labels
-                        .get(&label)
+                    instr.label = new_labels
+                        .get(&instr.label)
                         .ok_or(anyhow!(
                             "Internal error: {} not found in renamed label",
-                            label
+                            instr.label
                         ))?
                         .clone();
-                    new_insns.push(Instruction::IfLtZ { label, a });
+                    new_insns.push(Instruction::IfLtZ(instr));
                 }
-                Instruction::IfGeZ { label, a } => {
+                Instruction::IfGeZ(mut instr) => {
                     last_ins_was_a_label = false;
-                    let label = new_labels
-                        .get(&label)
+                    instr.label = new_labels
+                        .get(&instr.label)
                         .ok_or(anyhow!(
                             "Internal error: {} not found in renamed label",
-                            label
+                            instr.label
                         ))?
                         .clone();
-                    new_insns.push(Instruction::IfGeZ { label, a });
+                    new_insns.push(Instruction::IfGeZ(instr));
                 }
-                Instruction::IfGtZ { label, a } => {
+                Instruction::IfGtZ(mut instr) => {
                     last_ins_was_a_label = false;
-                    let label = new_labels
-                        .get(&label)
+                    instr.label = new_labels
+                        .get(&instr.label)
                         .ok_or(anyhow!(
                             "Internal error: {} not found in renamed label",
-                            label
+                            instr.label
                         ))?
                         .clone();
-                    new_insns.push(Instruction::IfGtZ { label, a });
+                    new_insns.push(Instruction::IfGtZ(instr));
                 }
-                Instruction::IfLeZ { label, a } => {
+                Instruction::IfLeZ(mut instr) => {
                     last_ins_was_a_label = false;
-                    let label = new_labels
-                        .get(&label)
+                    instr.label = new_labels
+                        .get(&instr.label)
                         .ok_or(anyhow!(
                             "Internal error: {} not found in renamed label",
-                            label
+                            instr.label
                         ))?
                         .clone();
-                    new_insns.push(Instruction::IfLeZ { label, a });
+                    new_insns.push(Instruction::IfLeZ(instr));
                 }
-                Instruction::Try {
-                    end_label,
-                    mut handlers,
-                    default_handler,
-                } => {
+                Instruction::Try(mut instr) => {
                     last_ins_was_a_label = false;
-                    let end_label = new_labels
-                        .get(&end_label)
+                    instr.end_label = new_labels
+                        .get(&instr.end_label)
                         .ok_or(anyhow!(
                             "Internal error: {} not found in renamed label",
-                            end_label
+                            instr.end_label
                         ))?
                         .clone();
-                    for handler in &mut handlers {
-                        handler.1 = new_labels
-                            .get(&handler.1)
+                    for i in 0..instr.handlers.len() {
+                        instr.handlers[i].1 = new_labels
+                            .get(&instr.handlers[i].1)
                             .ok_or(anyhow!(
                                 "Internal error: {} not found in renamed label",
-                                handler.1
+                                instr.handlers[i].1
                             ))?
                             .clone();
                     }
-                    let default_handler = default_handler
-                        .map(|label| {
+                    if let Some(label) = instr.default_handler {
+                        instr.default_handler = Some(
                             new_labels
                                 .get(&label)
                                 .ok_or(anyhow!(
                                     "Internal error: {} not found in renamed label",
                                     label
-                                ))
-                                .cloned()
-                        })
-                        .transpose()?;
-                    new_insns.push(Instruction::Try {
-                        end_label,
-                        handlers,
-                        default_handler,
-                    });
+                                ))?
+                                .clone(),
+                        );
+                    }
+                    new_insns.push(Instruction::Try(instr));
                 }
-                Instruction::Switch { mut branches, reg } => {
+                Instruction::Switch(mut instr) => {
                     last_ins_was_a_label = false;
-                    for label in branches.values_mut() {
+                    for label in instr.branches.values_mut() {
                         *label = new_labels
                             .get(label)
                             .ok_or(anyhow!(
@@ -469,15 +454,15 @@ impl Code {
                             ))?
                             .clone();
                     }
-                    new_insns.push(Instruction::Switch { branches, reg });
+                    new_insns.push(Instruction::Switch(instr));
                 }
-                Instruction::Label { name } => {
-                    if !used_labels.contains(&name) {
+                Instruction::Label(ins::Label { name }) => {
+                    if used_labels.get(&name).is_none() {
                         //println!("{name} not used");
                         continue;
                     }
                     if !last_ins_was_a_label {
-                        new_insns.push(Instruction::Label {
+                        new_insns.push(Instruction::Label(ins::Label {
                             name: new_labels
                                 .get(&name)
                                 .ok_or(anyhow!(
@@ -485,11 +470,11 @@ impl Code {
                                     name
                                 ))?
                                 .clone(),
-                        });
+                        }));
                     }
                     last_ins_was_a_label = true;
                 }
-                Instruction::Nop {} => (),
+                Instruction::Nop(_) => (),
                 instr => {
                     last_ins_was_a_label = false;
                     new_insns.push(instr);
@@ -502,38 +487,3 @@ impl Code {
         })
     }
 }
-
-impl<V: Visitor> Visitable<V> for Code {
-    fn default_visit(&self, v: &mut V) -> Result<()> {
-        for ins in &self.insns {
-            v.visit_instruction(ins)?;
-        }
-        Ok(())
-    }
-}
-impl<V: VisitorMut> VisitableMut<V> for Code {
-    fn default_visit_mut(self, v: &mut V) -> Result<Self> {
-        let parameter_names = if let Some(parameter_names) = self.parameter_names {
-            let mut new_param = vec![];
-            for param in parameter_names {
-                if let Some(param) = param {
-                    new_param.push(Some(v.visit_string(param)?));
-                } else {
-                    new_param.push(None);
-                }
-            }
-            Some(new_param)
-        } else {
-            None
-        };
-        Ok(Self {
-            parameter_names,
-            insns: self
-                .insns
-                .into_iter()
-                .map(|ins| v.visit_instruction(ins))
-                .collect::<Result<_>>()?,
-            ..self
-        })
-    }
-}

androscalpel/src/code_analysis/method_cfg.rs

@@ -1,350 +0,0 @@
-//! The Control Flow Graph for a method.
-
-use crate::{Instruction, Method, Result};
-use anyhow::Context;
-use std::collections::HashMap;
-
-const EMPTY_INSNS_SLICE: &[Instruction] = &[];
-
-/// A basic block of code of a method.
-#[derive(Debug, PartialEq)]
-pub struct MethodCFGNode<'a> {
-    /// Code represented by the block
-    pub code_block: &'a [Instruction],
-    /// Labels at the begining of the node if they exists
-    pub labels: Vec<String>,
-    /// Indices in CodeGraph.nodes of the next nodes
-    pub next_nodes: Vec<usize>,
-    /// Indices in CodeGraph.nodes of the previous nodes
-    pub prev_nodes: Vec<usize>,
-}
-
-/// The CFG for a method, with potentially additionnal informations.
-#[derive(Debug, PartialEq)]
-pub struct MethodCFG<'a> {
-    pub method: &'a Method,
-    pub nodes: Vec<MethodCFGNode<'a>>,
-}
-
-impl Method {
-    pub fn get_cfg(&self) -> Result<MethodCFG> {
-        MethodCFG::new(self)
-    }
-}
-
-impl<'a> MethodCFG<'a> {
-    pub fn new(method: &'a Method) -> Result<Self> {
-        let insns: &'a [Instruction] = if let Some(code) = method.code.as_ref() {
-            &code.insns
-        } else {
-            EMPTY_INSNS_SLICE
-        };
-
-        let mut nodes = vec![MethodCFGNode {
-            code_block: &insns[0..0],
-            labels: vec![],
-            next_nodes: vec![],
-            prev_nodes: vec![],
-        }];
-        let mut nodes_next_label = vec![vec![]];
-        let nb_insns = insns.len();
-        if nb_insns != 0 {
-            nodes[0].next_nodes.push(1);
-        }
-        let mut start_last_block = 0;
-        let mut last_labels = vec![];
-        let mut block_started = false;
-        let mut try_block: Vec<(String, Vec<String>)> = vec![];
-        for (i, ins) in insns.iter().enumerate() {
-            match ins {
-                // TODO: handle error better: list ins that can throw exceptions better
-                Instruction::Throw { .. }
-                | Instruction::InvokeVirtual { .. }
-                | Instruction::InvokeSuper { .. }
-                | Instruction::InvokeDirect { .. }
-                | Instruction::InvokeDirect { .. }
-                | Instruction::InvokeInterface { .. }
-                | Instruction::InvokePolymorphic { .. }
-                | Instruction::InvokeCustom { .. }
-                    if !try_block.is_empty() =>
-                {
-                    nodes_next_label.push(try_block.last().unwrap().1.clone());
-                    let next_nodes =
-                        if i + 1 < nb_insns && !matches!(ins, Instruction::Throw { .. }) {
-                            vec![nodes.len() + 1] // If no exception, continue to next ins
-                        } else {
-                            vec![]
-                        };
-                    nodes.push(MethodCFGNode {
-                        code_block: &insns[start_last_block..i + 1],
-                        labels: last_labels,
-                        next_nodes,
-                        prev_nodes: vec![],
-                    });
-                    start_last_block = i + 1;
-                    last_labels = vec![];
-                    block_started = false;
-                }
-                Instruction::Goto { label } => {
-                    nodes_next_label.push(vec![label.clone()]);
-                    nodes.push(MethodCFGNode {
-                        code_block: &insns[start_last_block..i + 1],
-                        labels: last_labels,
-                        next_nodes: vec![], // Do not continue the execution at next ins
-                        prev_nodes: vec![],
-                    });
-                    start_last_block = i + 1;
-                    last_labels = vec![];
-                    block_started = false;
-                }
-                Instruction::Switch { branches, .. } => {
-                    nodes_next_label.push(branches.values().cloned().collect());
-                    let next_nodes = if i + 1 < nb_insns {
-                        vec![nodes.len() + 1] // If no branches match, continue execution
-                    } else {
-                        vec![]
-                    };
-                    nodes.push(MethodCFGNode {
-                        code_block: &insns[start_last_block..i + 1],
-                        labels: last_labels,
-                        next_nodes,
-                        prev_nodes: vec![],
-                    });
-                    start_last_block = i + 1;
-                    last_labels = vec![];
-                    block_started = false;
-                }
-                Instruction::IfEq { label, .. }
-                | Instruction::IfNe { label, .. }
-                | Instruction::IfLt { label, .. }
-                | Instruction::IfGe { label, .. }
-                | Instruction::IfGt { label, .. }
-                | Instruction::IfLe { label, .. }
-                | Instruction::IfEqZ { label, .. }
-                | Instruction::IfNeZ { label, .. }
-                | Instruction::IfLtZ { label, .. }
-                | Instruction::IfGeZ { label, .. }
-                | Instruction::IfGtZ { label, .. }
-                | Instruction::IfLeZ { label, .. } => {
-                    nodes_next_label.push(vec![label.clone()]);
-                    let next_nodes = if i + 1 < nb_insns {
-                        vec![nodes.len() + 1] // depending on test, continue execution
-                    } else {
-                        vec![]
-                    };
-                    nodes.push(MethodCFGNode {
-                        code_block: &insns[start_last_block..i + 1],
-                        labels: last_labels,
-                        next_nodes,
-                        prev_nodes: vec![],
-                    });
-                    start_last_block = i + 1;
-                    last_labels = vec![];
-                    block_started = false;
-                }
-                Instruction::Try {
-                    end_label,
-                    handlers,
-                    default_handler,
-                } => {
-                    let mut branches: Vec<_> =
-                        handlers.iter().map(|(_, label)| label.clone()).collect();
-                    if let Some(default_handler) = default_handler.as_ref().cloned() {
-                        branches.push(default_handler);
-                    }
-                    try_block.push((end_label.clone(), branches))
-                }
-                Instruction::Label { name } => {
-                    if !block_started {
-                        last_labels.push(name.clone());
-                    } else {
-                        nodes_next_label.push(vec![]);
-                        nodes.push(MethodCFGNode {
-                            code_block: &insns[start_last_block..i],
-                            labels: last_labels,
-                            next_nodes: vec![nodes.len() + 1],
-                            prev_nodes: vec![],
-                        });
-                        start_last_block = i;
-                        last_labels = vec![name.clone()];
-                    }
-                }
-                Instruction::ReturnVoid {}
-                | Instruction::Return { .. }
-                | Instruction::ReturnWide { .. }
-                | Instruction::ReturnObject { .. }
-                | Instruction::Throw { .. } => {
-                    nodes_next_label.push(vec![]);
-                    nodes.push(MethodCFGNode {
-                        code_block: &insns[start_last_block..i + 1],
-                        labels: last_labels,
-                        next_nodes: vec![], // Do not continue the execution at next ins
-                        prev_nodes: vec![],
-                    });
-                    start_last_block = i + 1;
-                    last_labels = vec![];
-                    block_started = false;
-                }
-                _ => {
-                    if !ins.is_pseudo_ins() {
-                        block_started = true;
-                    }
-                }
-            }
-        }
-        if start_last_block != nb_insns {
-            nodes_next_label.push(vec![]);
-            nodes.push(MethodCFGNode {
-                code_block: &insns[start_last_block..nb_insns],
-                labels: last_labels,
-                next_nodes: vec![],
-                prev_nodes: vec![],
-            });
-        }
-        let label_to_node: HashMap<String, usize> = nodes
-            .iter()
-            .enumerate()
-            .flat_map(|(i, node)| node.labels.clone().into_iter().map(move |lab| (lab, i)))
-            .collect();
-        for (node, labels) in nodes.iter_mut().zip(nodes_next_label) {
-            for label in labels {
-                node.next_nodes
-                    .push(*label_to_node.get(&label).with_context(|| {
-                        format!("found jumb to label '{}' but label not found", label)
-                    })?);
-            }
-        }
-
-        for i in 0..nodes.len() {
-            let next_nodes = nodes[i].next_nodes.clone();
-            for j in &next_nodes {
-                nodes[*j].prev_nodes.push(i);
-            }
-        }
-        Ok(Self { method, nodes })
-    }
-
-    /// Serialize the graph to dot format.
-    pub fn to_dot(&self, add_reg_ty: bool) -> String {
-        let mut dot_string: String = "digraph {\n".into();
-        dot_string += "overlap=false;\n";
-        dot_string += &self.to_dot_subgraph();
-        dot_string += "\n";
-        if add_reg_ty {
-            dot_string += &self.reg_types_dot();
-        }
-        dot_string += "}";
-        dot_string
-    }
-
-    /// Compute a sanitized version of the method name
-    pub(crate) fn dot_sanitized_method_dscr(&self) -> String {
-        self.method
-            .descriptor
-            .__str__()
-            .replace("/", "_")
-            .replace(";", "")
-            .replace(">", "")
-            .replace("-", "_")
-            .replace("(", "_")
-            .replace(")", "_")
-            .replace("[", "T")
-    }
-
-    // method call: cluster_{mth}:node_{i}:i{j}:e -> cluster_{mth2}:n ?
-
-    /// Serialize the graph to dot format.
-    pub fn to_dot_subgraph(&self) -> String {
-        let mut dot_string = format!(
-            "subgraph \"cluster_{}\" {{\n",
-            self.dot_sanitized_method_dscr()
-        );
-        dot_string += "    style=\"dashed\";\n";
-        dot_string += "    color=\"black\";\n";
-        dot_string += &format!("    label=\"{}\";\n", self.method.descriptor.__str__());
-        for (i, node) in self.nodes.iter().enumerate() {
-            let block_name = if i == 0 {
-                "ENTRY".into()
-            } else if !node.labels.is_empty() {
-                format!(
-                    "block '{}'",
-                    node.labels[0]
-                        .replace("&", "&amp;")
-                        .replace(">", "&gt;")
-                        .replace("<", "&lt;")
-                        .replace("\"", "&quot;")
-                )
-            } else {
-                format!("block {i}")
-            };
-            let label = if node.code_block.is_empty() {
-                //format!("{{\\< {block_name} \\>}}")
-                format!("<TABLE BORDER=\"0\" CELLBORDER=\"1\" CELLSPACING=\"0\"><TR><TD><B> {block_name} </B></TD></TR></TABLE>")
-            } else {
-                /*
-                let mut label = format!("{{\\< {block_name} \\>:\\l\\\n");
-                for (i, ins) in node.code_block.iter().enumerate() {
-                    label += &format!("|<i{i}>");
-                    label += ins
-                        .__str__()
-                        .replace(" ", "\\ ")
-                        .replace(">", "\\>")
-                        .replace("<", "\\<")
-                        .replace("\"", "\\\"")
-                        .replace("{", "\\{")
-                        .replace("}", "\\}")
-                        //.replace("[", "\\[")
-                        .as_str();
-                    label += "\\l\\\n";
-                }
-                label += "}";
-                */
-                let mut label = format!("<TABLE BORDER=\"0\" CELLBORDER=\"1\" CELLSPACING=\"0\">\n        <TR><TD PORT=\"blkname\"><B> {block_name} </B></TD></TR>\n");
-                for (i, ins) in node.code_block.iter().enumerate() {
-                    label += &format!(
-                        "        <TR><TD PORT=\"i{i}\">{}</TD></TR>\n",
-                        ins.__str__()
-                            //.replace(" ", "\\ ")
-                            .replace("&", "&amp;")
-                            .replace(">", "&gt;")
-                            .replace("<", "&lt;")
-                            .replace("\"", "&quot;")
-                            //.replace("{", "\\{")
-                            //.replace("}", "\\}")
-                            //.replace("[", "\\[")
-                            .as_str()
-                    );
-                }
-                label += "    </TABLE>";
-                label
-            };
-            dot_string += &format!(
-                //"    node_{i} [shape=record,style=filled,fillcolor=lightgrey,label=\"{label}\"];\n\n"
-                "    node_{i} [shape=plaintext,style=filled,fillcolor=lightgrey,label=<{label}>];\n\n"
-            );
-        }
-        //dot_string += "    node_end [shape=record,style=filled,fillcolor=lightgrey,label=\"{\\< EXIT \\>}\"];\n\n";
-        dot_string += "    node_end [shape=plaintext,style=filled,fillcolor=lightgrey,label=<<TABLE BORDER=\"0\" CELLBORDER=\"1\" CELLSPACING=\"0\"><TR><TD><B> EXIT </B></TD></TR></TABLE>>];\n\n";
-
-        for (i, node) in self.nodes.iter().enumerate() {
-            for j in &node.next_nodes {
-                if *j == i + 1 {
-                    dot_string += &format!(
-                        "    node_{i}:s -> node_{j}:n [style=\"solid,bold\",color=black,weight=100,constraint=true];\n"
-                    );
-                } else {
-                    dot_string += &format!(
-                        "    node_{i}:s -> node_{j}:n [style=\"solid,bold\",color=black,weight=10,constraint=true];\n"
-                    );
-                }
-            }
-            if node.next_nodes.is_empty() {
-                dot_string += &format!(
-                    "    node_{i}:s -> node_end:n [style=\"solid,bold\",color=black,weight=10,constraint=true];\n"
-                );
-            }
-        }
-        dot_string += "}\n";
-        dot_string
-    }
-}

androscalpel/src/code_analysis/mod.rs

@@ -1,8 +0,0 @@
-//! Module for more advanced code analysis.
-//!
-//! This is module is quite experimental but can be usefull.
-
-pub mod method_cfg;
-pub mod register_type;
-pub use method_cfg::*;
-pub use register_type::RegType;

androscalpel/src/code_analysis/register_type.rs

@@ -1,555 +0,0 @@
-//! Compute the register types at each label of a method.
-
-use super::{MethodCFG, MethodCFGNode};
-use crate::Instruction;
-use std::collections::HashMap;
-
-/// The different possible types of a register
-#[derive(Debug, PartialEq, Clone, Copy)]
-pub enum RegType {
-    /// The register content is not yet defined
-    Undefined,
-    /// The register contains a java object. It can be a classical object
-    /// an array, a type, etc
-    Object,
-    /// The register contains a 32 bit scalar
-    SimpleScalar,
-    /// The register contains the first 32 bits of a wide register. If not
-    /// followed by a `SecondWideScalar`, it should be considered as a `SimpleScalar`?
-    FirstWideScalar,
-    /// The register contains the last 32 bits of a wide register. If not
-    /// preceded by a `FirstWideScalar`, it should be considered as a `SimpleScalar`?
-    SecondWideScalar,
-    /// The register can be either a scalar or an object.
-    Any,
-}
-
-impl RegType {
-    pub fn to_str(&self) -> &'static str {
-        match self {
-            RegType::Undefined => "undef",
-            RegType::Object => "object",
-            RegType::SimpleScalar => "scalar",
-            RegType::FirstWideScalar => "wide1",
-            RegType::SecondWideScalar => "wide2",
-            RegType::Any => "any",
-        }
-    }
-}
-
-impl MethodCFG<'_> {
-    pub fn get_reg_types(&self) -> HashMap<String, Vec<RegType>> {
-        let code = if let Some(code) = self.method.code.as_ref() {
-            code
-        } else {
-            return HashMap::new();
-        };
-        let nb_reg = code.registers_size as usize;
-        let mut end_block_reg_tys: Vec<_> = self
-            .nodes
-            .iter()
-            .map(|_| vec![RegType::Undefined; nb_reg])
-            .collect();
-        if end_block_reg_tys.is_empty() {
-            return HashMap::new();
-        }
-        // Initialize the entry block from function signature:
-        let mut i = (code.registers_size - code.ins_size(self.method)) as usize;
-        if !self.method.is_static {
-            end_block_reg_tys[0][i] = RegType::Object; // 'this'
-            i += 1;
-        }
-        for arg in &self.method.descriptor.proto.get_parameters() {
-            if arg.is_class() || arg.is_array() {
-                end_block_reg_tys[0][i] = RegType::Object;
-                i += 1;
-            } else if arg.is_long() || arg.is_double() {
-                end_block_reg_tys[0][i] = RegType::FirstWideScalar;
-                i += 1;
-                end_block_reg_tys[0][i] = RegType::SecondWideScalar;
-                i += 1;
-            } else {
-                end_block_reg_tys[0][i] = RegType::SimpleScalar;
-                i += 1;
-            }
-        }
-        let mut changed = true;
-        while changed {
-            // The first node is empty and depend on the function signature, it must not change
-            let mut new_end_block_reg_tys = vec![end_block_reg_tys[0].clone()];
-            for node in self.nodes.iter().skip(1) {
-                new_end_block_reg_tys.push(transform_reg_ty(
-                    &merge_input(node, nb_reg, &end_block_reg_tys),
-                    node,
-                ))
-            }
-            changed = end_block_reg_tys != new_end_block_reg_tys;
-            end_block_reg_tys = new_end_block_reg_tys;
-        }
-        let start_block_reg_tys = &self
-            .nodes
-            .iter()
-            .map(|node| merge_input(node, nb_reg, &end_block_reg_tys))
-            .collect::<Vec<_>>();
-        self.nodes
-            .iter()
-            .enumerate()
-            .flat_map(|(i, node)| {
-                node.labels
-                    .iter()
-                    .map(move |label| (label.clone(), start_block_reg_tys[i].clone()))
-            })
-            .collect()
-    }
-
-    pub(crate) fn reg_types_dot(&self) -> String {
-        let types = self.get_reg_types();
-        let mut dot_string = format!(
-            "subgraph \"cluster_reg_types_{}\" {{\n",
-            self.dot_sanitized_method_dscr()
-        );
-        dot_string += "    style=\"dashed\";\n";
-        dot_string += "    color=\"black\";\n";
-        dot_string += "    rankdir=\"TB\";\n";
-        dot_string += &format!(
-            "    label=\"Register Types for {}\";\n",
-            self.method.descriptor.__str__()
-        );
-
-        let mut labels: Vec<_> = types.keys().collect();
-        labels.sort(); // Order more or less by addresses, good enought
-        let mut prev = None;
-        for label in labels.into_iter() {
-            let regs = types.get(label).unwrap();
-            let mut node_label = String::new();
-            node_label += "<TABLE BORDER=\"0\" CELLBORDER=\"1\" CELLSPACING=\"0\">\n";
-            node_label += &format!(
-                "        <TR><TD COLSPAN=\"{}\"><B>register types at {label}</B></TD></TR>\n        <TR>", regs.len(),
-            );
-            for i in 0..regs.len() {
-                node_label += &format!("        <TD>{i}</TD>\n");
-            }
-            node_label += "    </TR><TR>\n";
-            for reg in regs {
-                //node_label += "|";
-                //node_label += reg.to_str();
-                node_label += &format!("        <TD>{}</TD>\n", reg.to_str());
-            }
-            //node_label += "|";
-            node_label += "    </TR></TABLE>";
-            dot_string += &format!(
-                //"    node_{label} [shape=record,style=filled,fillcolor=lightgrey,label=\"{node_label}\"];\n"
-                "    node_{label} [shape=plaintext,style=filled,fillcolor=lightgrey,label=<{node_label}>];\n"
-            );
-            if let Some(prev) = prev {
-                // Add invisible edge to ensure the nodes are verticals
-                dot_string += &format!("    {prev}:s -> node_{label}:n [style=\"invis\"];\n");
-            }
-            prev = Some(format!("node_{label}"));
-        }
-
-        dot_string += "}\n";
-        dot_string += "\n";
-        for (i, node) in self.nodes.iter().enumerate() {
-            for (j, ins) in node.code_block.iter().enumerate() {
-                if let Instruction::Label { name } = ins {
-                    let mid = self.dot_sanitized_method_dscr();
-                    dot_string += &format!(
-                        "    node_{i}:i{j}:e -> node_{name} \
-                        [ltail=cluster_{mid},lhead=cluster_reg_types_{mid},style=\"solid,bold\",color=grey,weight=10,constraint=true];\n",
-                    );
-                }
-            }
-        }
-        dot_string
-    }
-}
-
-fn merge_input(node: &MethodCFGNode, nb_reg: usize, inputs: &[Vec<RegType>]) -> Vec<RegType> {
-    use RegType::*;
-    let mut reg_tys = vec![Undefined; nb_reg];
-    for i in &node.prev_nodes {
-        for (r, ty) in inputs[*i].iter().enumerate() {
-            reg_tys[r] = match (reg_tys[r], ty) {
-                (Undefined, _) => *ty,
-                (_, Undefined) => reg_tys[r],
-                (_, Any) => Any,
-                (Any, _) => Any,
-
-                (Object, Object) => Object,
-                (Object, SimpleScalar) => Any,
-                (Object, FirstWideScalar) => Any,
-                (Object, SecondWideScalar) => Any,
-
-                (SimpleScalar, Object) => Any,
-                (SimpleScalar, SimpleScalar) => SimpleScalar,
-                (SimpleScalar, FirstWideScalar) => SimpleScalar,
-                (SimpleScalar, SecondWideScalar) => SimpleScalar,
-
-                (FirstWideScalar, Object) => Any,
-                (FirstWideScalar, SimpleScalar) => SimpleScalar,
-                (FirstWideScalar, FirstWideScalar) => FirstWideScalar,
-                (FirstWideScalar, SecondWideScalar) => SimpleScalar,
-
-                (SecondWideScalar, Object) => Any,
-                (SecondWideScalar, SimpleScalar) => SimpleScalar,
-                (SecondWideScalar, FirstWideScalar) => SimpleScalar,
-                (SecondWideScalar, SecondWideScalar) => SecondWideScalar,
-            }
-        }
-    }
-    reg_tys
-}
-
-fn transform_reg_ty(input_types: &[RegType], cfg: &MethodCFGNode) -> Vec<RegType> {
-    use Instruction::*;
-    use RegType::*;
-    let mut types = input_types.to_vec();
-    for ins in cfg.code_block {
-        match ins {
-            Move { to, .. } => types[*to as usize] = SimpleScalar, // E: mism
-            MoveWide { to, .. } => {
-                types[*to as usize] = FirstWideScalar;
-                types[*to as usize + 1] = SecondWideScalar;
-            }
-            MoveObject { to, .. } => types[*to as usize] = Object,
-            MoveResult { to: reg } | Return { reg } | Const { reg, .. } | Switch { reg, .. } => {
-                types[*reg as usize] = SimpleScalar
-            }
-            MoveResultWide { to: reg } | ReturnWide { reg } | ConstWide { reg, .. } => {
-                types[*reg as usize] = FirstWideScalar;
-                types[*reg as usize + 1] = SecondWideScalar;
-            }
-            MoveResultObject { to: reg }
-            | MoveException { to: reg }
-            | ReturnObject { reg }
-            | ConstString { reg, .. }
-            | ConstClass { reg, .. }
-            | NewInstance { reg, .. }
-            | Throw { reg }
-            | ConstMethodHandle { to: reg, .. }
-            | ConstMethodType { to: reg, .. } => types[*reg as usize] = Object,
-            InstanceOf { dest, obj: _, .. } => {
-                // types[*obj as usize] = Object; not sure about this one
-                types[*dest as usize] = SimpleScalar;
-            }
-            ArrayLength { dest, arr } => {
-                types[*arr as usize] = Object;
-                types[*dest as usize] = SimpleScalar;
-            }
-            NewArray { reg, size_reg, .. } => {
-                types[*reg as usize] = Object;
-                types[*size_reg as usize] = SimpleScalar;
-            }
-            FilledNewArray { type_, reg_values } => {
-                let reg_ty = if type_.is_class() || type_.is_array() {
-                    Object
-                //} else if type_.is_long() || type_.is_double() {
-                //    SimpleScalar // Not supposed to happend so default to simple scalar just in
-                //    case
-                } else {
-                    SimpleScalar
-                };
-                for i in reg_values {
-                    types[*i as usize] = reg_ty;
-                }
-            }
-            CmpLFloat { dest, b, c } | CmpGFloat { dest, b, c } => {
-                types[*dest as usize] = SimpleScalar;
-                types[*b as usize] = SimpleScalar;
-                types[*c as usize] = SimpleScalar;
-            }
-            CmpLDouble { dest, b, c } | CmpGDouble { dest, b, c } | CmpLong { dest, b, c } => {
-                types[*dest as usize] = SimpleScalar;
-                types[*b as usize] = FirstWideScalar;
-                types[*b as usize + 1] = SecondWideScalar;
-                types[*c as usize] = FirstWideScalar;
-                types[*c as usize + 1] = SecondWideScalar;
-            }
-            IfEq { a, b, .. }
-            | IfNe { a, b, .. }
-            | IfLt { a, b, .. }
-            | IfGe { a, b, .. }
-            | IfGt { a, b, .. }
-            | IfLe { a, b, .. } => {
-                types[*a as usize] = SimpleScalar;
-                types[*b as usize] = SimpleScalar;
-            }
-            IfEqZ { a, .. }
-            | IfNeZ { a, .. }
-            | IfLtZ { a, .. }
-            | IfGeZ { a, .. }
-            | IfGtZ { a, .. }
-            | IfLeZ { a, .. } => {
-                types[*a as usize] = SimpleScalar;
-            }
-            AGet {
-                dest: r,
-                arr: obj,
-                idx,
-            }
-            | AGetBoolean {
-                dest: r,
-                arr: obj,
-                idx,
-            }
-            | AGetByte {
-                dest: r,
-                arr: obj,
-                idx,
-            }
-            | AGetChar {
-                dest: r,
-                arr: obj,
-                idx,
-            }
-            | AGetShort {
-                dest: r,
-                arr: obj,
-                idx,
-            }
-            | APut {
-                from: r,
-                arr: obj,
-                idx,
-            }
-            | APutBoolean {
-                from: r,
-                arr: obj,
-                idx,
-            }
-            | APutByte {
-                from: r,
-                arr: obj,
-                idx,
-            }
-            | APutChar {
-                from: r,
-                arr: obj,
-                idx,
-            }
-            | APutShort {
-                from: r,
-                arr: obj,
-                idx,
-            } => {
-                types[*r as usize] = SimpleScalar;
-                types[*obj as usize] = Object;
-                types[*idx as usize] = SimpleScalar;
-            }
-            AGetWide {
-                dest: r,
-                arr: obj,
-                idx,
-            }
-            | APutWide {
-                from: r,
-                arr: obj,
-                idx,
-            } => {
-                types[*r as usize] = FirstWideScalar;
-                types[*r as usize + 1] = SecondWideScalar;
-                types[*obj as usize] = Object;
-                types[*idx as usize] = SimpleScalar;
-            }
-            AGetObject {
-                dest: r,
-                arr: obj,
-                idx,
-            }
-            | APutObject {
-                from: r,
-                arr: obj,
-                idx,
-            } => {
-                types[*r as usize] = Object;
-                types[*obj as usize] = Object;
-                types[*idx as usize] = SimpleScalar;
-            }
-            IGet { to: r, obj, .. }
-            | IGetBoolean { to: r, obj, .. }
-            | IGetByte { to: r, obj, .. }
-            | IGetChar { to: r, obj, .. }
-            | IGetShort { to: r, obj, .. }
-            | IPut { from: r, obj, .. }
-            | IPutBoolean { from: r, obj, .. }
-            | IPutByte { from: r, obj, .. }
-            | IPutChar { from: r, obj, .. }
-            | IPutShort { from: r, obj, .. } => {
-                types[*r as usize] = SimpleScalar;
-                types[*obj as usize] = Object;
-            }
-            IGetWide { to: r, obj, .. } | IPutWide { from: r, obj, .. } => {
-                types[*r as usize] = FirstWideScalar;
-                types[*r as usize] = SecondWideScalar;
-                types[*obj as usize] = Object;
-            }
-            IGetObject { to: r, obj, .. } | IPutObject { from: r, obj, .. } => {
-                types[*r as usize] = Object;
-                types[*obj as usize] = Object;
-            }
-            SGet { to: r, .. }
-            | SGetBoolean { to: r, .. }
-            | SGetByte { to: r, .. }
-            | SGetChar { to: r, .. }
-            | SGetShort { to: r, .. }
-            | SPut { from: r, .. }
-            | SPutBoolean { from: r, .. }
-            | SPutByte { from: r, .. }
-            | SPutChar { from: r, .. }
-            | SPutShort { from: r, .. } => {
-                types[*r as usize] = SimpleScalar;
-            }
-            SGetWide { to: r, .. } | SPutWide { from: r, .. } => {
-                types[*r as usize] = FirstWideScalar;
-                types[*r as usize] = SecondWideScalar;
-            }
-            SGetObject { to: r, .. } | SPutObject { from: r, .. } => {
-                types[*r as usize] = Object;
-            }
-            /* They are information to get from the method type, but, meh, this is not
-             * necessary (we should have the type from when the reg was initialized)
-            InvokeVirtual { method: _, args: _ }
-            | InvokeSuper { method: _, args: _ }
-            | InvokeDirect { method: _, args: _ }
-            | InvokeStatic { method: _, args: _ }
-            | InvokeInterface { method: _, args: _ } => todo!(),
-            InvokePolymorphic { method:_, proto: _, args: _ } => todo!(),
-            InvokeCustom { call_site: _, args: _ } => todo!(),
-            */
-            NegInt { dest, val }
-            | NotInt { dest, val }
-            | NegFloat { dest, val }
-            | NegDouble { dest, val }
-            | IntToFloat { dest, val }
-            | FloatToInt { dest, val }
-            | IntToByte { dest, val }
-            | IntToChar { dest, val }
-            | IntToShort { dest, val } => {
-                types[*dest as usize] = SimpleScalar;
-                types[*val as usize] = SimpleScalar;
-            }
-            NegLong { dest, val }
-            | NotLong { dest, val }
-            | LongToDouble { dest, val }
-            | DoubleToLong { dest, val } => {
-                types[*dest as usize] = FirstWideScalar;
-                types[*dest as usize + 1] = SecondWideScalar;
-                types[*val as usize] = FirstWideScalar;
-                types[*val as usize + 1] = SecondWideScalar;
-            }
-            IntToLong { dest, val }
-            | IntToDouble { dest, val }
-            | FloatToLong { dest, val }
-            | FloatToDouble { dest, val } => {
-                types[*dest as usize] = FirstWideScalar;
-                types[*dest as usize + 1] = SecondWideScalar;
-                types[*val as usize] = SimpleScalar;
-            }
-            LongToInt { dest, val }
-            | LongToFloat { dest, val }
-            | DoubleToInt { dest, val }
-            | DoubleToFloat { dest, val } => {
-                types[*dest as usize] = SimpleScalar;
-                types[*val as usize] = FirstWideScalar;
-                types[*val as usize + 1] = SecondWideScalar;
-            }
-            AddInt { dest, b, c }
-            | SubInt { dest, b, c }
-            | MulInt { dest, b, c }
-            | DivInt { dest, b, c }
-            | RemInt { dest, b, c }
-            | AndInt { dest, b, c }
-            | OrInt { dest, b, c }
-            | XorInt { dest, b, c }
-            | ShlInt { dest, b, c }
-            | ShrInt { dest, b, c }
-            | UshrInt { dest, b, c }
-            | AddFloat { dest, b, c }
-            | SubFloat { dest, b, c }
-            | MulFloat { dest, b, c }
-            | DivFloat { dest, b, c }
-            | RemFloat { dest, b, c } => {
-                types[*dest as usize] = SimpleScalar;
-                types[*b as usize] = SimpleScalar;
-                types[*c as usize] = SimpleScalar;
-            }
-            AddLong { dest, b, c }
-            | SubLong { dest, b, c }
-            | MulLong { dest, b, c }
-            | DivLong { dest, b, c }
-            | RemLong { dest, b, c }
-            | AndLong { dest, b, c }
-            | OrLong { dest, b, c }
-            | XorLong { dest, b, c }
-            | ShlLong { dest, b, c }
-            | ShrLong { dest, b, c }
-            | UshrLong { dest, b, c }
-            | AddDouble { dest, b, c }
-            | SubDouble { dest, b, c }
-            | MulDouble { dest, b, c }
-            | DivDouble { dest, b, c }
-            | RemDouble { dest, b, c } => {
-                types[*dest as usize] = FirstWideScalar;
-                types[*dest as usize + 1] = SecondWideScalar;
-                types[*b as usize] = FirstWideScalar;
-                types[*b as usize + 1] = SecondWideScalar;
-                types[*c as usize] = FirstWideScalar;
-                types[*c as usize + 1] = SecondWideScalar;
-            }
-            AddInt2Addr { dest, b }
-            | SubInt2Addr { dest, b }
-            | MulInt2Addr { dest, b }
-            | DivInt2Addr { dest, b }
-            | RemInt2Addr { dest, b }
-            | AndInt2Addr { dest, b }
-            | OrInt2Addr { dest, b }
-            | XorInt2Addr { dest, b }
-            | ShlInt2Addr { dest, b }
-            | ShrInt2Addr { dest, b }
-            | UshrInt2Addr { dest, b }
-            | AddFloat2Addr { dest, b }
-            | SubFloat2Addr { dest, b }
-            | MulFloat2Addr { dest, b }
-            | DivFloat2Addr { dest, b }
-            | RemFloat2Addr { dest, b }
-            | AddIntLit { dest, b, .. }
-            | RsubIntLit { dest, b, .. }
-            | MulIntLit { dest, b, .. }
-            | DivIntLit { dest, b, .. }
-            | RemIntLit { dest, b, .. }
-            | AndIntLit { dest, b, .. }
-            | OrIntLit { dest, b, .. }
-            | XorIntLit { dest, b, .. }
-            | ShlIntLit { dest, b, .. }
-            | ShrIntLit { dest, b, .. }
-            | UshrIntLit { dest, b, .. } => {
-                types[*dest as usize] = SimpleScalar;
-                types[*b as usize] = SimpleScalar;
-            }
-            AddLong2Addr { dest, b }
-            | SubLong2Addr { dest, b }
-            | MulLong2Addr { dest, b }
-            | DivLong2Addr { dest, b }
-            | RemLong2Addr { dest, b }
-            | AndLong2Addr { dest, b }
-            | OrLong2Addr { dest, b }
-            | XorLong2Addr { dest, b }
-            | ShlLong2Addr { dest, b }
-            | ShrLong2Addr { dest, b }
-            | UshrLong2Addr { dest, b }
-            | AddDouble2Addr { dest, b }
-            | SubDouble2Addr { dest, b }
-            | MulDouble2Addr { dest, b }
-            | DivDouble2Addr { dest, b }
-            | RemDouble2Addr { dest, b } => {
-                types[*dest as usize] = FirstWideScalar;
-                types[*dest as usize + 1] = SecondWideScalar;
-                types[*b as usize] = FirstWideScalar;
-                types[*b as usize + 1] = SecondWideScalar;
-            }
-            _ => (),
-        }
-    }
-    types
-}

androscalpel/src/dex_id.rs

@@ -2,20 +2,21 @@
 
 use serde::{Deserialize, Serialize};
 use std::cmp::{Ord, Ordering, PartialOrd};
+use std::collections::hash_map::DefaultHasher;
 use std::collections::HashSet;
-use std::hash::Hash;
+use std::hash::{Hash, Hasher};
 
 use anyhow::{anyhow, bail, Context};
-#[cfg(feature = "python")]
+use pyo3::class::basic::CompareOp;
 use pyo3::prelude::*;
 
-use crate::{scalar::*, DexString, DexValue, Result, Visitable, VisitableMut, Visitor, VisitorMut};
+use crate::{scalar::*, DexString, DexValue, Result};
 use androscalpel_serializer::{StringDataItem, Uleb128};
 
 /// The type of a method. The shorty is formated as described in
 /// <https://source.android.com/docs/core/runtime/dex-format#shortydescriptor>
-#[cfg_attr(feature = "python", pyclass(eq, ord, hash, frozen))]
-#[derive(Debug, Clone, PartialEq, Eq, Hash)]
+#[pyclass]
+#[derive(Debug, Clone, PartialEq, Eq, Hash, Deserialize, Serialize)]
 pub struct IdMethodType {
     /// Type formated as described by  <https://source.android.com/docs/core/runtime/dex-format#shortydescriptor>
     pub(crate) shorty: DexString, // Redondant, but same as in the encoding, keep it in case we ever
@@ -24,30 +25,6 @@ pub struct IdMethodType {
     pub(crate) parameters: Vec<IdType>,
 }
 
-impl<V: Visitor> Visitable<V> for IdMethodType {
-    fn default_visit(&self, v: &mut V) -> Result<()> {
-        v.visit_string(&self.shorty)?;
-        v.visit_type(&self.return_type)?;
-        for ty in &self.parameters {
-            v.visit_type(ty)?;
-        }
-        Ok(())
-    }
-}
-
-impl<V: VisitorMut> VisitableMut<V> for IdMethodType {
-    fn default_visit_mut(mut self, v: &mut V) -> Result<Self> {
-        self.shorty = v.visit_string(self.shorty)?;
-        self.return_type = v.visit_type(self.return_type)?;
-        self.parameters = self
-            .parameters
-            .into_iter()
-            .map(|ty| v.visit_type(ty))
-            .collect::<Result<_>>()?;
-        Ok(self)
-    }
-}
-
 impl Ord for IdMethodType {
     fn cmp(&self, other: &Self) -> Ordering {
         self.return_type
@@ -63,18 +40,18 @@ impl PartialOrd for IdMethodType {
     }
 }
 
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl IdMethodType {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(return_type: IdType, parameters: Vec<IdType>) -> Self {
         Self {
             shorty: Self::compute_shorty(&return_type, &parameters),
@@ -86,12 +63,12 @@ impl IdMethodType {
     /// Try to parse a smali representation of a prototype into a IdMethodType.
     ///
     /// ```
-    /// use androscalpel::{IdMethodType, IdType};
+    /// use androscalpel::IdMethodType;
     ///
     /// let proto = IdMethodType::from_smali("(Landroidx/core/util/Predicate;Landroidx/core/util/Predicate;Ljava/lang/Object;)Z").unwrap();
     /// assert_eq!(
     ///   proto,
-    ///   IdMethodType::new(
+    ///   IdMethodType(
     ///     IdType::boolean(),
     ///     vec![
     ///       IdType::class("androidx/core/util/Predicate"),
@@ -101,7 +78,7 @@ impl IdMethodType {
     ///   )
     /// );
     /// ```
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_smali(smali_repr: &str) -> Result<Self> {
         if smali_repr.len() < 2 {
             bail!("'{smali_repr}' is to short to be a prototype");
@@ -166,6 +143,12 @@ impl IdMethodType {
         self.parameters.clone()
     }
 
+    pub fn __hash__(&self) -> u64 {
+        let mut hasher = DefaultHasher::new();
+        self.hash(&mut hasher);
+        hasher.finish()
+    }
+
     /// Return all strings referenced in the Id.
     pub fn get_all_strings(&self) -> HashSet<DexString> {
         let mut strings = HashSet::new();
@@ -191,24 +174,9 @@ impl IdMethodType {
         protos.insert(self.clone());
         protos
     }
-}
 
-impl SmaliName for IdMethodType {
-    /// Convert a descriptor to its smali representation.
-    fn try_to_smali(&self) -> Result<String> {
-        Ok(format!(
-            "({}){}",
-            self.parameters
-                .iter()
-                .map(|param| param.try_to_smali())
-                .collect::<Result<Vec<String>>>()?
-                .join(""),
-            self.return_type.try_to_smali()?
-        ))
-    }
-    /// Convert a smali representation to its descriptor.
-    fn try_from_smali(smali: &str) -> Result<Self> {
-        Self::from_smali(smali)
+    fn __richcmp__(&self, other: &Self, op: CompareOp) -> bool {
+        op.matches(self.cmp(other))
     }
 }
 
@@ -216,10 +184,9 @@ impl IdMethodType {
     /// Compute the format for the shorty as described in
     /// <https://source.android.com/docs/core/runtime/dex-format#shortydescriptor>
     pub fn compute_shorty(return_type: &IdType, parameters: &[IdType]) -> DexString {
-        // TODO: computing on dex string instead of string? that a lot of doubious conversion
-        let mut shorty: String = return_type.get_shorty().__str__();
+        let mut shorty: String = return_type.get_shorty().into();
         for ty in parameters {
-            let ty: String = ty.get_shorty().__str__();
+            let ty: String = ty.get_shorty().into();
             shorty.push_str(&ty);
         }
         shorty.into()
@@ -231,39 +198,23 @@ impl IdMethodType {
 /// as described here <https://source.android.com/docs/core/runtime/dex-format#typedescriptor>
 // Not a clean rust enum because we want to be compatible with python, and maybe support strange
 // malware edge case?
-#[cfg_attr(feature = "python", pyclass(eq, ord, hash, frozen))]
-#[derive(Debug, Clone, PartialEq, Eq, Hash, PartialOrd, Ord)]
+#[pyclass]
+#[derive(Debug, Clone, PartialEq, Eq, Hash, PartialOrd, Ord, Deserialize, Serialize)]
 pub struct IdType(pub(crate) DexString);
-
-impl<V: Visitor> Visitable<V> for IdType {
-    fn default_visit(&self, v: &mut V) -> Result<()> {
-        v.visit_string(&self.0)
-    }
-}
-
-impl<V: VisitorMut> VisitableMut<V> for IdType {
-    fn default_visit_mut(self, v: &mut V) -> Result<Self> {
-        Ok(Self(v.visit_string(self.0)?))
-    }
-}
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl IdType {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(
-        #[cfg_attr(
-            feature = "python",
-            pyo3(from_py_with = "crate::dex_string::as_dex_string")
-        )]
-        ty: DexString,
+        #[pyo3(from_py_with = "crate::dex_string::as_dex_string")] ty: DexString,
     ) -> Result<Self> {
         // TODO: check format
         let ty = Self(ty);
@@ -272,7 +223,7 @@ impl IdType {
     }
 
     /// Return a type from its string representation without checking its format
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn unchecked_new(ty: DexString) -> Self {
         Self(ty)
     }
@@ -283,7 +234,7 @@ impl IdType {
     /// id types.
     ///
     /// ```
-    /// use androscalpel::IdType;
+    /// use androscalpel::IdType
     ///
     /// let id_type = IdType::from_smali(
     ///   "Landroidx/core/util/Predicate;"
@@ -294,7 +245,7 @@ impl IdType {
     ///   IdType::class("androidx/core/util/Predicate")
     /// );
     /// ```
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_smali(smali_repr: &str) -> Result<Self> {
         Self::new(smali_repr.into())
     }
@@ -316,7 +267,7 @@ impl IdType {
     ///   IdType::boolean(),
     /// ])
     /// ```
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn get_list_from_str(type_list: &str) -> Result<Vec<Self>> {
         let mut lst = vec![];
         let mut chars = type_list.chars();
@@ -332,18 +283,15 @@ impl IdType {
                 'J' => Some(Self::long()),
                 'F' => Some(Self::float()),
                 'D' => Some(Self::double()),
-                '[' => {
-                    array_dimmention += 1;
-                    None
-                }
+                '[' => { array_dimmention += 1; None },
                 'L' => {
                     let mut class_name = String::new();
-                    for cc in chars.by_ref() {
-                        if cc == ';' {
-                            break;
-                        } else {
+                    for cc in chars.by_ref(){
+                        if cc == ';' { break;}
+                        else {
                             class_name.push(cc);
                         }
+
                     }
                     Some(Self::class(&class_name))
                 }
@@ -373,77 +321,66 @@ impl IdType {
     }
 
     /// Return the void type (for return type)
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn void() -> Self {
         Self("V".into())
     }
 
     /// Return the boolean type
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn boolean() -> Self {
         Self("Z".into())
     }
     /// Return the byte type
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn byte() -> Self {
         Self("B".into())
     }
 
     /// Return the short type
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn short() -> Self {
         Self("S".into())
     }
 
     /// Return the char type
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn char() -> Self {
         Self("C".into())
     }
 
     /// Return the int type
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn int() -> Self {
         Self("I".into())
     }
 
     /// Return the long type
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn long() -> Self {
         Self("J".into())
     }
 
     /// Return the float type
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn float() -> Self {
         Self("F".into())
     }
 
     /// Return the double type
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn double() -> Self {
         Self("D".into())
     }
 
     /// Return the type for the class of fully qualified name `name`
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn class(name: &str) -> Self {
         Self(format!("L{name};").into())
     }
 
-    /// Return the type for the class of fully qualified name `name`.
-    /// Same as [`IdType::class`] but with a [`DexString`] name.
-    #[cfg_attr(feature = "python", staticmethod)]
-    pub fn class_from_dex_string(name: &DexString) -> Self {
-        let mut repr = name.clone();
-        repr.0.utf16_size.0 += 2;
-        repr.0.data.insert(0, b'L');
-        repr.0.data.push(b';');
-        Self(repr)
-    }
-
     /// Return the type for an array of the specify `type_`
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn array(type_: &IdType) -> Self {
         let mut ty = type_.clone();
         ty.0 .0.utf16_size.0 += 1;
@@ -456,11 +393,11 @@ impl IdType {
     }
 
     pub fn __str__(&self) -> String {
-        self.0.__str__()
+        (&self.0).into()
     }
 
     pub fn __repr__(&self) -> String {
-        let name: String = self.0.__str__();
+        let name: String = (&self.0).into();
         format!("IdType(\"{name}\")")
     }
 
@@ -577,7 +514,7 @@ impl IdType {
         {
             Ok(())
         } else {
-            let format: String = self.0.__str__();
+            let format: String = (&self.0).into();
             Err(anyhow!("{format} is not a valid type"))
         }
     }
@@ -628,6 +565,12 @@ impl IdType {
         }
     }
 
+    pub fn __hash__(&self) -> u64 {
+        let mut hasher = DefaultHasher::new();
+        self.hash(&mut hasher);
+        hasher.finish()
+    }
+
     /// Return all strings referenced in the Id.
     pub fn get_all_strings(&self) -> HashSet<DexString> {
         let mut strings = HashSet::new();
@@ -642,81 +585,41 @@ impl IdType {
         types
     }
 
-    /// Check if the class is a platform class (ie in the android SDK or a hidden API).
-    #[cfg(feature = "platform-list")]
-    pub fn is_platform_class(&self) -> bool {
-        match self.try_to_smali() {
-            Ok(smali) => androscalpel_platform_api_list::is_platform_class(&smali),
-            Err(_) => false,
-        }
+    fn __richcmp__(&self, other: &Self, op: CompareOp) -> bool {
+        op.matches(self.cmp(other))
     }
-
     // TODO: TESTS
 }
 
-impl SmaliName for IdType {
-    /// Convert a descriptor to its smali representation.
-    fn try_to_smali(&self) -> Result<String> {
-        let r = (&self.0 .0).try_into()?; // Anyhow conversion stuff
-        Ok(r)
-    }
-    /// Convert a smali representation to its descriptor.
-    fn try_from_smali(smali: &str) -> Result<Self> {
-        Self::from_smali(smali)
-    }
-}
-
-#[cfg_attr(feature = "python", pyclass(eq, ord, hash, frozen))]
-#[derive(Debug, Clone, PartialEq, Eq, Hash)]
+#[pyclass]
+#[derive(Debug, Clone, PartialEq, Eq, Hash, Deserialize, Serialize)]
 pub struct IdField {
     /// The name of the field, format described at
     /// <https://source.android.com/docs/core/runtime/dex-format#membername>
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub name: DexString,
     /// The type of the field.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub type_: IdType,
     /// The class that own the field.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub class_: IdType,
 }
 
-impl<V: Visitor> Visitable<V> for IdField {
-    fn default_visit(&self, v: &mut V) -> Result<()> {
-        v.visit_string(&self.name)?;
-        v.visit_type(&self.type_)?;
-        v.visit_type(&self.class_)?;
-        Ok(())
-    }
-}
-
-impl<V: VisitorMut> VisitableMut<V> for IdField {
-    fn default_visit_mut(mut self, v: &mut V) -> Result<Self> {
-        self.name = v.visit_string(self.name)?;
-        self.type_ = v.visit_type(self.type_)?;
-        self.class_ = v.visit_type(self.class_)?;
-        Ok(self)
-    }
-}
-
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl IdField {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(
-        #[cfg_attr(
-            feature = "python",
-            pyo3(from_py_with = "crate::dex_string::as_dex_string")
-        )]
-        name: DexString,
+        #[pyo3(from_py_with = "crate::dex_string::as_dex_string")] name: DexString,
         type_: IdType,
         class_: IdType,
     ) -> Self {
@@ -731,19 +634,19 @@ impl IdField {
     /// Try to parse a smali representation of a field into a IdField.
     ///
     /// ```
-    /// use androscalpel::{IdField, IdType};
+    /// use androscalpel::IdField;
     ///
-    /// let proto = IdField::from_smali("Ljava/lang/annotation/ElementType;->FIELD:Ljava/lang/annotation/ElementType;").unwrap();
+    /// let proto = IdField::from_smali("Ljava/lang/annotation/ElementType;->METHOD:Ljava/lang/annotation/ElementType;").unwrap();
     /// assert_eq!(
     ///   proto,
     ///   IdField::new(
-    ///     "FIELD".into(),
+    ///     "METHOD".into(),
     ///     IdType::class("java/lang/annotation/ElementType"),
     ///     IdType::class("java/lang/annotation/ElementType"),
     ///   )
     /// );
     /// ```
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_smali(smali_repr: &str) -> Result<Self> {
         let arrow_i = smali_repr.find("->");
         if arrow_i.is_none() {
@@ -769,19 +672,25 @@ impl IdField {
     }
 
     pub fn __str__(&self) -> String {
-        let class: String = self.class_.get_name().__str__();
-        let name: String = self.name.__str__();
-        let ty: String = self.type_.get_name().__str__();
+        let class: String = self.class_.get_name().into();
+        let name: String = (&self.name).into();
+        let ty: String = self.type_.get_name().into();
         format!("{class}->{name}:{ty}")
     }
 
     pub fn __repr__(&self) -> String {
         let class: String = self.class_.__repr__();
-        let name: String = self.name.__str__();
+        let name: String = (&self.name).into();
         let ty: String = self.type_.__repr__();
         format!("IdField(\"{name}\", {ty}, {class})")
     }
 
+    pub fn __hash__(&self) -> u64 {
+        let mut hasher = DefaultHasher::new();
+        self.hash(&mut hasher);
+        hasher.finish()
+    }
+
     /// Return all strings referenced in the Id.
     pub fn get_all_strings(&self) -> HashSet<DexString> {
         let mut strings = HashSet::new();
@@ -806,13 +715,8 @@ impl IdField {
         fields
     }
 
-    /// Check if the field is a platform field (ie in the android SDK or a hidden API).
-    #[cfg(feature = "platform-list")]
-    pub fn is_platform_field(&self) -> bool {
-        match self.try_to_smali() {
-            Ok(smali) => androscalpel_platform_api_list::is_platform_field(&smali),
-            Err(_) => false,
-        }
+    fn __richcmp__(&self, other: &Self, op: CompareOp) -> bool {
+        op.matches(self.cmp(other))
     }
 }
 
@@ -831,71 +735,35 @@ impl PartialOrd for IdField {
     }
 }
 
-impl SmaliName for IdField {
-    /// Convert a descriptor to its smali representation.
-    fn try_to_smali(&self) -> Result<String> {
-        let class: String = self.class_.try_to_smali()?;
-        let name: String = (&self.name.0).try_into()?;
-        let ty: String = self.type_.try_to_smali()?;
-        Ok(format!("{class}->{name}:{ty}"))
-    }
-    /// Convert a smali representation to its descriptor.
-    fn try_from_smali(smali: &str) -> Result<Self> {
-        Self::from_smali(smali)
-    }
-}
-
 /// The Id of a method.
-#[cfg_attr(feature = "python", pyclass(eq, ord, hash, frozen))]
-#[derive(Debug, Clone, PartialEq, Eq, Hash)]
+#[pyclass]
+#[derive(Debug, Clone, PartialEq, Eq, Hash, Deserialize, Serialize)]
 pub struct IdMethod {
     /// The class containing the method.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub class_: IdType,
     /// The prototype (aka type or signature) of the method.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub proto: IdMethodType,
     /// The name of the method.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub name: DexString,
 }
 
-impl<V: Visitor> Visitable<V> for IdMethod {
-    fn default_visit(&self, v: &mut V) -> Result<()> {
-        v.visit_string(&self.name)?;
-        v.visit_method_type(&self.proto)?;
-        v.visit_type(&self.class_)?;
-        Ok(())
-    }
-}
-
-impl<V: VisitorMut> VisitableMut<V> for IdMethod {
-    fn default_visit_mut(mut self, v: &mut V) -> Result<Self> {
-        self.name = v.visit_string(self.name)?;
-        self.proto = v.visit_method_type(self.proto)?;
-        self.class_ = v.visit_type(self.class_)?;
-        Ok(self)
-    }
-}
-
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl IdMethod {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(
-        #[cfg_attr(
-            feature = "python",
-            pyo3(from_py_with = "crate::dex_string::as_dex_string")
-        )]
-        name: DexString,
+        #[pyo3(from_py_with = "crate::dex_string::as_dex_string")] name: DexString,
         proto: IdMethodType,
         class_: IdType,
     ) -> Self {
@@ -909,7 +777,7 @@ impl IdMethod {
     /// Try to parse a smali representation of method into a IdMethod.
     ///
     /// ```
-    /// use androscalpel::{IdType, IdMethod, IdMethodType};
+    /// use androscalpel::IdMethod;
     ///
     /// let id_method = IdMethod::from_smali(
     ///   "Landroidx/core/util/Predicate;->lambda$and$0(Landroidx/core/util/Predicate;Landroidx/core/util/Predicate;Ljava/lang/Object;)Z"
@@ -931,7 +799,7 @@ impl IdMethod {
     ///   )
     /// );
     /// ```
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_smali(smali_repr: &str) -> Result<Self> {
         let arrow_i = smali_repr.find("->");
         if arrow_i.is_none() {
@@ -981,6 +849,12 @@ impl IdMethod {
         )
     }
 
+    pub fn __hash__(&self) -> u64 {
+        let mut hasher = DefaultHasher::new();
+        self.hash(&mut hasher);
+        hasher.finish()
+    }
+
     /// Return all strings referenced in the Id.
     pub fn get_all_strings(&self) -> HashSet<DexString> {
         let mut strings = HashSet::new();
@@ -1012,13 +886,8 @@ impl IdMethod {
         method_ids
     }
 
-    /// Check if the method is a platform method (ie in the android SDK or a hidden API).
-    #[cfg(feature = "platform-list")]
-    pub fn is_platform_method(&self) -> bool {
-        match self.try_to_smali() {
-            Ok(smali) => androscalpel_platform_api_list::is_platform_method(&smali),
-            Err(_) => false,
-        }
+    fn __richcmp__(&self, other: &Self, op: CompareOp) -> bool {
+        op.matches(self.cmp(other))
     }
 }
 
@@ -1037,51 +906,22 @@ impl PartialOrd for IdMethod {
     }
 }
 
-impl SmaliName for IdMethod {
-    /// Convert a descriptor to its smali representation.
-    fn try_to_smali(&self) -> Result<String> {
-        let name: String = (&self.name.0).try_into()?;
-        Ok(format!(
-            "{}->{}{}",
-            self.class_.try_to_smali()?,
-            name,
-            self.proto.try_to_smali()?,
-        ))
-    }
-    /// Convert a smali representation to its descriptor.
-    fn try_from_smali(smali: &str) -> Result<Self> {
-        Self::from_smali(smali)
-    }
-}
-
-#[cfg_attr(feature = "python", pyclass(eq, ord, hash, frozen))]
-#[derive(Debug, Clone, PartialEq, Eq, Ord, PartialOrd, Hash)]
+#[pyclass]
+#[derive(Debug, Clone, PartialEq, Eq, Deserialize, Serialize, Ord, PartialOrd)]
 pub struct IdEnum(pub IdField);
 
-impl<V: Visitor> Visitable<V> for IdEnum {
-    fn default_visit(&self, v: &mut V) -> Result<()> {
-        v.visit_field_id(&self.0)
-    }
-}
-
-impl<V: VisitorMut> VisitableMut<V> for IdEnum {
-    fn default_visit_mut(self, v: &mut V) -> Result<Self> {
-        Ok(Self(v.visit_field_id(self.0)?))
-    }
-}
-
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl IdEnum {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(val: IdField) -> Self {
         Self(val)
     }
@@ -1112,62 +952,8 @@ impl IdEnum {
     pub fn get_all_field_ids(&self) -> HashSet<IdField> {
         self.0.get_all_field_ids()
     }
-}
 
-// Not to sure about this one
-impl SmaliName for IdEnum {
-    /// Convert a descriptor to its smali representation.
-    fn try_to_smali(&self) -> Result<String> {
-        self.0.try_to_smali()
-    }
-    /// Convert a smali representation to its descriptor.
-    fn try_from_smali(smali: &str) -> Result<Self> {
-        Ok(Self(IdField::from_smali(smali)?))
+    fn __richcmp__(&self, other: &Self, op: CompareOp) -> bool {
+        op.matches(self.cmp(other))
     }
 }
-
-pub trait SmaliName: Sized {
-    /// Convert a descriptor to its smali representation.
-    fn try_to_smali(&self) -> Result<String>;
-    /// Convert a smali representation to its descriptor.
-    fn try_from_smali(smali: &str) -> Result<Self>;
-}
-
-macro_rules! serde_serialize_to_smali {
-    // This macro takes an argument of designator `ident` and
-    // implement Serialize and Deserialize for the type, assuming
-    // it implement SmaliName.
-    ($type_name:ident) => {
-        impl serde::Serialize for $type_name {
-            fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
-            where
-                S: serde::Serializer,
-            {
-                Serialize::serialize(
-                    &self
-                        .try_to_smali()
-                        .expect(&format!("Failed to convert {} to smali", self.__str__())),
-                    serializer,
-                )
-            }
-        }
-
-        impl<'de> serde::Deserialize<'de> for $type_name {
-            fn deserialize<D>(deserializer: D) -> Result<$type_name, D::Error>
-            where
-                D: serde::Deserializer<'de>,
-            {
-                <String as Deserialize>::deserialize(deserializer).map(|string| {
-                    $type_name::try_from_smali(&string)
-                        .expect(&format!("Failed to convert {string} as smali"))
-                })
-            }
-        }
-    };
-}
-
-serde_serialize_to_smali!(IdMethodType);
-serde_serialize_to_smali!(IdType);
-serde_serialize_to_smali!(IdMethod);
-serde_serialize_to_smali!(IdField);
-serde_serialize_to_smali!(IdEnum);

androscalpel/src/dex_string.rs

@@ -1,56 +1,41 @@
-use crate::{Result, Visitable, VisitableMut, Visitor, VisitorMut};
-use anyhow::{Context, Error};
+use crate::Result;
 use serde::{Deserialize, Deserializer, Serialize, Serializer};
 use std::cmp::{Ord, PartialOrd};
+use std::collections::hash_map::DefaultHasher;
 use std::collections::HashSet;
-use std::hash::Hash;
+use std::hash::{Hash, Hasher};
 
-#[cfg(feature = "python")]
-use pyo3::{exceptions::PyTypeError, prelude::*};
+use pyo3::class::basic::CompareOp;
+use pyo3::exceptions::PyTypeError;
+use pyo3::prelude::*;
 
-#[cfg_attr(feature = "python", pyclass(eq, ord, frozen, hash))]
+#[pyclass]
 #[derive(Clone, PartialEq, Eq, Ord, PartialOrd)]
 pub struct DexString(pub androscalpel_serializer::StringDataItem);
 
-// Kinda useless
-impl<V: Visitor> Visitable<V> for DexString {
-    fn default_visit(&self, _: &mut V) -> Result<()> {
-        Ok(())
-    }
-}
-
-impl<V: VisitorMut> VisitableMut<V> for DexString {
-    fn default_visit_mut(self, _: &mut V) -> Result<Self> {
-        Ok(self)
-    }
-}
-
 impl std::fmt::Debug for DexString {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> Result<(), std::fmt::Error> {
-        match TryInto::<String>::try_into(self) {
-            Ok(string) => {
-                f.write_str(&format!(
-                    "DexString({}, {:#x})",
-                    string, self.0.utf16_size.0
-                ))
-                /*
-                f.debug_tuple("DexString")
-                    .field(&string)
-                    .field(&self.0.utf16_size.0)
-                    .finish()
-                */
-            }
-            _ => {
-                f.write_str(&format!(
-                    "DexString({:?}, {:#x})",
-                    self.0.data, self.0.utf16_size.0
-                ))
-                /*f.debug_tuple("DexString")
-                    .field(&self.0.data)
-                    .field(&self.0.utf16_size.0)
-                    .finish()
-                */
-            }
+        if let Ok(string) = TryInto::<String>::try_into(self) {
+            f.write_str(&format!(
+                "DexString({}, {:#x})",
+                string, self.0.utf16_size.0
+            ))
+            /*
+            f.debug_tuple("DexString")
+                .field(&string)
+                .field(&self.0.utf16_size.0)
+                .finish()
+            */
+        } else {
+            f.write_str(&format!(
+                "DexString({:?}, {:#x})",
+                self.0.data, self.0.utf16_size.0
+            ))
+            /*f.debug_tuple("DexString")
+                .field(&self.0.data)
+                .field(&self.0.utf16_size.0)
+                .finish()
+            */
         }
     }
 }
@@ -120,19 +105,17 @@ impl From<androscalpel_serializer::StringDataItem> for DexString {
     }
 }
 
-impl TryFrom<&DexString> for String {
-    type Error = Error;
-    fn try_from(DexString(string): &DexString) -> Result<Self> {
+impl From<&DexString> for String {
+    fn from(DexString(string): &DexString) -> Self {
         string
             .try_into()
-            .with_context(|| format!("InvalidEncoding:{:x?}", string.data))
+            .unwrap_or(format!("InvalidEncoding:{:x?}", string.data))
     }
 }
 
-impl TryFrom<DexString> for String {
-    type Error = Error;
-    fn try_from(string: DexString) -> Result<Self> {
-        (&string).try_into()
+impl From<DexString> for String {
+    fn from(string: DexString) -> Self {
+        (&string).into()
     }
 }
 
@@ -155,11 +138,10 @@ impl Hash for DexString {
     }
 }
 
-#[cfg(feature = "python")]
-pub fn as_dex_string(obj: &Bound<'_, PyAny>) -> PyResult<DexString> {
-    if let Ok(string) = DexString::extract_bound(obj) {
+pub fn as_dex_string(obj: &PyAny) -> PyResult<DexString> {
+    if let Ok(string) = DexString::extract(obj) {
         Ok(string)
-    } else if let Ok(string) = String::extract_bound(obj) {
+    } else if let Ok(string) = String::extract(obj) {
         Ok(string.into())
     } else {
         Err(PyErr::new::<PyTypeError, _>(format!(
@@ -169,18 +151,18 @@ pub fn as_dex_string(obj: &Bound<'_, PyAny>) -> PyResult<DexString> {
     }
 }
 
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl DexString {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(s: &str) -> Self {
         s.into()
     }
@@ -196,16 +178,17 @@ impl DexString {
     }
 
     pub fn __str__(&self) -> String {
-        match TryInto::<String>::try_into(self) {
-            Ok(string) => string,
-            _ => {
-                format!("string{:02x?}", self.0.data)
-            }
-        }
+        self.into()
     }
 
     pub fn __repr__(&self) -> String {
-        format!("{:?}", self)
+        self.into()
+    }
+
+    fn __hash__(&self) -> u64 {
+        let mut hasher = DefaultHasher::new();
+        self.hash(&mut hasher);
+        hasher.finish()
     }
 
     /// Return all strings references in the value.
@@ -215,19 +198,10 @@ impl DexString {
         strings
     }
 
-    /// Build the concatenation of two string.
-    pub fn concatenate(&self, other: &Self) -> Self {
-        let Self(androscalpel_serializer::StringDataItem {
-            utf16_size: androscalpel_serializer::Uleb128(size1),
-            data: data1,
-        }) = self.clone();
-        let Self(androscalpel_serializer::StringDataItem {
-            utf16_size: androscalpel_serializer::Uleb128(size2),
-            data: data2,
-        }) = other.clone();
-        Self(androscalpel_serializer::StringDataItem {
-            utf16_size: androscalpel_serializer::Uleb128(size1 + size2),
-            data: data1.into_iter().chain(data2).collect(),
-        })
+    fn __richcmp__(&self, other: &PyAny, op: CompareOp) -> PyResult<bool> {
+        let other: Self = other
+            .extract()
+            .or(<String as FromPyObject>::extract(other).map(|string| string.into()))?;
+        Ok(op.matches(self.0.cmp(&other.0)))
     }
 }

androscalpel/src/field.rs

@@ -3,55 +3,51 @@
 use serde::{Deserialize, Serialize};
 use std::collections::HashSet;
 
-#[cfg(feature = "python")]
 use pyo3::prelude::*;
 
 use crate::{
-    DexAnnotationItem, DexString, DexValue, HiddenApiData, IdField, IdMethod, IdMethodType, IdType,
-    MethodHandle, Result, Visitable, VisitableMut, Visitor, VisitorMut,
+    DexAnnotationItem, DexString, DexValue, IdField, IdMethod, IdMethodType, IdType, MethodHandle,
+    Result,
 };
 use androscalpel_serializer::consts::*;
 
 /// Represent a field.
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, PartialEq, Deserialize, Serialize)]
 pub struct Field {
     /// The structure used to reference this field.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub descriptor: IdField,
     /// The field visibility
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub visibility: FieldVisibility,
     /// If the field is defined for the class globally
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_static: bool,
     /// If the field is immutable after construction
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_final: bool,
     /// For thread safety
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_volatile: bool,
     /// If the field should **not** be saved by default serialization
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_transient: bool,
     /// If the field is not defined in the source code
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_synthetic: bool,
     /// If the field is an enumerated value
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_enum: bool,
     /// The default value of this field
     pub value: Option<DexValue>,
     /// The annotations for this field
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub annotations: Vec<DexAnnotationItem>,
-    /// Hidden Api data.
-    #[cfg_attr(feature = "python", pyo3(get))]
-    pub hiddenapi: Option<HiddenApiData>,
 }
 
 /// Represent the visibility of a field
-#[cfg_attr(feature = "python", pyclass(eq, eq_int))]
+#[pyclass]
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize, Serialize)]
 pub enum FieldVisibility {
     Public,
@@ -60,18 +56,18 @@ pub enum FieldVisibility {
     None_, // Actually quite common
 }
 
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl Field {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(descriptor: IdField) -> Self {
         Self {
             descriptor,
@@ -84,7 +80,6 @@ impl Field {
             is_enum: false,
             value: None,
             annotations: vec![],
-            hiddenapi: None,
         }
     }
 
@@ -125,16 +120,14 @@ impl Field {
     }
 
     /// Return the value as a python object
-    #[cfg(feature = "python")]
     #[getter]
     pub fn get_value(&self) -> Option<DexValue> {
         self.value.clone()
     }
 
     /// Set the value from a python object
-    #[cfg(feature = "python")]
     #[setter]
-    pub fn set_value(&mut self, ob: &Bound<'_, PyAny>) -> PyResult<()> {
+    pub fn set_value(&mut self, ob: &PyAny) -> PyResult<()> {
         self.value = Some(ob.extract()?);
         // TODO: check type match
         Ok(())
@@ -256,46 +249,7 @@ impl Field {
         !self.annotations.is_empty()
     }
 
-    /// Check if the field is a platform field (ie in the android SDK or a hidden API).
-    #[cfg(feature = "platform-list")]
-    pub fn is_platform_field(&self) -> bool {
-        self.descriptor.is_platform_field()
-    }
-}
-
-impl<V: Visitor> Visitable<V> for Field {
-    fn default_visit(&self, v: &mut V) -> Result<()> {
-        v.visit_field_id(&self.descriptor)?;
-        v.visit_field_visibility(&self.visibility)?;
-        if let Some(val) = &self.value {
-            v.visit_value(val)?;
-        }
-        for annot in &self.annotations {
-            v.visit_annotation_item(annot)?;
-        }
-        if let Some(hiddenapi) = &self.hiddenapi {
-            v.visit_hidden_api_data(hiddenapi)?;
-        }
-
-        Ok(())
-    }
-}
-impl<V: VisitorMut> VisitableMut<V> for Field {
-    fn default_visit_mut(self, v: &mut V) -> Result<Self> {
-        Ok(Self {
-            descriptor: v.visit_field_id(self.descriptor)?,
-            visibility: v.visit_field_visibility(self.visibility)?,
-            value: self.value.map(|val| v.visit_value(val)).transpose()?,
-            annotations: self
-                .annotations
-                .into_iter()
-                .map(|annot| v.visit_annotation_item(annot))
-                .collect::<Result<_>>()?,
-            hiddenapi: self
-                .hiddenapi
-                .map(|hiddenapi| v.visit_hidden_api_data(hiddenapi))
-                .transpose()?,
-            ..self
-        })
+    pub fn __eq__(&self, other: &Self) -> bool {
+        self == other
     }
 }

androscalpel/src/hiddenapi.rs

@@ -1,211 +0,0 @@
-//! Representation of the hidden API data
-
-use crate::{Result, Visitable, VisitableMut, Visitor, VisitorMut};
-use log::warn;
-#[cfg(feature = "python")]
-use pyo3::prelude::*;
-use serde::{Deserialize, Serialize};
-
-#[cfg_attr(feature = "python", pyclass)]
-#[derive(Debug, PartialEq, Eq, Hash, Clone, Deserialize, Serialize)]
-pub struct HiddenApiData {
-    pub permission: HiddenApiPermission,
-    pub domain: HiddenApiDomain,
-}
-
-impl From<&androscalpel_serializer::HiddenApiFlags> for HiddenApiData {
-    fn from(flags: &androscalpel_serializer::HiddenApiFlags) -> Self {
-        Self {
-            permission: flags.value.into(),
-            domain: flags.domain_api.into(),
-        }
-    }
-}
-impl From<androscalpel_serializer::HiddenApiFlags> for HiddenApiData {
-    fn from(flags: androscalpel_serializer::HiddenApiFlags) -> Self {
-        (&flags).into()
-    }
-}
-
-impl From<&HiddenApiData> for androscalpel_serializer::HiddenApiFlags {
-    fn from(flags: &HiddenApiData) -> Self {
-        Self {
-            value: (&flags.permission).into(),
-            domain_api: (&flags.domain).into(),
-        }
-    }
-}
-impl From<HiddenApiData> for androscalpel_serializer::HiddenApiFlags {
-    fn from(flags: HiddenApiData) -> Self {
-        (&flags).into()
-    }
-}
-
-impl<V: Visitor> Visitable<V> for HiddenApiData {
-    fn default_visit(&self, v: &mut V) -> Result<()> {
-        v.visit_hidden_api_permission(&self.permission)?;
-        v.visit_hidden_api_domain(&self.domain)?;
-        Ok(())
-    }
-}
-impl<V: VisitorMut> VisitableMut<V> for HiddenApiData {
-    fn default_visit_mut(self, v: &mut V) -> Result<Self> {
-        Ok(Self {
-            permission: v.visit_hidden_api_permission(self.permission)?,
-            domain: v.visit_hidden_api_domain(self.domain)?,
-        })
-    }
-}
-
-#[cfg_attr(feature = "python", pyclass)]
-#[derive(Debug, PartialEq, Eq, Hash, Clone, Deserialize, Serialize)]
-pub enum HiddenApiPermission {
-    /// Interfaces that can be freely used and are supported as
-    /// part of the officially documented Android framework Package Index
-    Whitelist {},
-    /// Non-SDK interfaces that can be used regardless of the
-    /// application's target API level
-    Greylist {},
-    /// Non-SDK interfaces that cannot be used regardless of the
-    /// application's target API level. Accessing one of these
-    /// interfaces causes a runtime error.
-    Blacklist {},
-    /// Non-SDK interfaces that can be used for Android 8.x and
-    /// below unless they are restricted (targetSdkVersion <= 27 (O_MR1)).
-    GreylistMaxO {},
-    /// Non-SDK interfaces that can be used for Android 9.x unless
-    /// they are restricted.
-    GreylistMaxP {},
-    /// Non-SDK interfaces that can be used for Android 10.x unless
-    /// they are restricted.
-    GreylistMaxQ {},
-    /// Non-SDK interfaces that can be used for Android 11.x unless
-    /// they are restricted.
-    GreylistMaxR {},
-    GreylistMaxS {},
-    /// Unknown flag, either an error or this crate is out of date.
-    Unknwon {
-        value: u8,
-    },
-}
-
-impl HiddenApiPermission {
-    pub fn to_smali_name(&self) -> &'static str {
-        match self {
-            Self::Whitelist {} => "whitelist",
-            Self::Greylist {} => "greylist",
-            Self::Blacklist {} => "blacklist",
-            Self::GreylistMaxO {} => "greylist-max-o",
-            Self::GreylistMaxP {} => "greylist-max-p",
-            Self::GreylistMaxQ {} => "greylist-max-q",
-            Self::GreylistMaxR {} => "greylist-max-r",
-            Self::GreylistMaxS {} => "greylist-max-s",
-            Self::Unknwon { .. } => "???list",
-        }
-    }
-}
-
-impl From<&androscalpel_serializer::HiddenApiValue> for HiddenApiPermission {
-    fn from(permission: &androscalpel_serializer::HiddenApiValue) -> Self {
-        match permission {
-            androscalpel_serializer::HiddenApiValue::Whitelist => Self::Whitelist {},
-            androscalpel_serializer::HiddenApiValue::Greylist => Self::Greylist {},
-            androscalpel_serializer::HiddenApiValue::Blacklist => Self::Blacklist {},
-            androscalpel_serializer::HiddenApiValue::GreylistMaxO => Self::GreylistMaxO {},
-            androscalpel_serializer::HiddenApiValue::GreylistMaxP => Self::GreylistMaxP {},
-            androscalpel_serializer::HiddenApiValue::GreylistMaxQ => Self::GreylistMaxQ {},
-            androscalpel_serializer::HiddenApiValue::GreylistMaxR => Self::GreylistMaxR {},
-            androscalpel_serializer::HiddenApiValue::GreylistMaxS => Self::GreylistMaxS {},
-            androscalpel_serializer::HiddenApiValue::Unknwon(val) => Self::Unknwon { value: *val },
-        }
-    }
-}
-impl From<androscalpel_serializer::HiddenApiValue> for HiddenApiPermission {
-    fn from(permission: androscalpel_serializer::HiddenApiValue) -> Self {
-        (&permission).into()
-    }
-}
-
-impl From<&HiddenApiPermission> for androscalpel_serializer::HiddenApiValue {
-    fn from(permission: &HiddenApiPermission) -> Self {
-        match permission {
-            HiddenApiPermission::Whitelist {} => Self::Whitelist,
-            HiddenApiPermission::Greylist {} => Self::Greylist,
-            HiddenApiPermission::Blacklist {} => Self::Blacklist,
-            HiddenApiPermission::GreylistMaxO {} => Self::GreylistMaxO,
-            HiddenApiPermission::GreylistMaxP {} => Self::GreylistMaxP,
-            HiddenApiPermission::GreylistMaxQ {} => Self::GreylistMaxQ,
-            HiddenApiPermission::GreylistMaxR {} => Self::GreylistMaxR,
-            HiddenApiPermission::GreylistMaxS {} => Self::GreylistMaxS,
-            HiddenApiPermission::Unknwon { value } => Self::Unknwon(*value),
-        }
-    }
-}
-
-impl From<HiddenApiPermission> for androscalpel_serializer::HiddenApiValue {
-    fn from(permission: HiddenApiPermission) -> Self {
-        (&permission).into()
-    }
-}
-
-#[cfg_attr(feature = "python", pyclass)]
-#[derive(Debug, PartialEq, Eq, Hash, Clone, Deserialize, Serialize)]
-pub struct HiddenApiDomain {
-    pub is_core_platform_api: bool,
-    pub is_test_api: bool,
-    /// Unknown domains.
-    pub unknown_flags: u32,
-}
-
-impl HiddenApiDomain {
-    pub fn to_smali_name(&self) -> String {
-        let mut value = String::new();
-        if self.is_core_platform_api {
-            value += "core-platform-api";
-        }
-        if self.is_test_api {
-            if !value.is_empty() {
-                value += "|";
-            }
-            value += "test-api";
-        }
-        if self.unknown_flags != 0 {
-            warn!(
-                "This attribut has an unknown hiddenapi domain set, this will not be display for compatibility with apktool"
-            );
-        }
-        value
-    }
-}
-
-impl From<&androscalpel_serializer::HiddenApiDomain> for HiddenApiDomain {
-    fn from(domain: &androscalpel_serializer::HiddenApiDomain) -> Self {
-        Self {
-            is_core_platform_api: domain.is_core_platform_api,
-            is_test_api: domain.is_test_api,
-            unknown_flags: domain.unknown_flags,
-        }
-    }
-}
-
-impl From<androscalpel_serializer::HiddenApiDomain> for HiddenApiDomain {
-    fn from(domain: androscalpel_serializer::HiddenApiDomain) -> Self {
-        (&domain).into()
-    }
-}
-
-impl From<&HiddenApiDomain> for androscalpel_serializer::HiddenApiDomain {
-    fn from(domain: &HiddenApiDomain) -> androscalpel_serializer::HiddenApiDomain {
-        Self {
-            is_core_platform_api: domain.is_core_platform_api,
-            is_test_api: domain.is_test_api,
-            unknown_flags: domain.unknown_flags,
-        }
-    }
-}
-
-impl From<HiddenApiDomain> for androscalpel_serializer::HiddenApiDomain {
-    fn from(domain: HiddenApiDomain) -> androscalpel_serializer::HiddenApiDomain {
-        (&domain).into()
-    }
-}

androscalpel/src/lib.rs

@@ -1,6 +1,5 @@
-pub use anyhow::Result;
+use anyhow::Result;
 
-#[cfg(feature = "python")]
 use pyo3::prelude::*;
 
 pub mod annotation;
@@ -12,19 +11,12 @@ pub mod dex_string;
 pub mod dex_writer;
 pub mod field;
 mod hashmap_vectorize;
-pub mod hiddenapi;
 pub mod instructions;
 pub mod method;
 pub mod method_handle;
-#[cfg(feature = "python")]
 pub mod py_utils;
 pub mod scalar;
-pub mod utils;
 pub mod value;
-pub mod visitor;
-
-#[cfg(feature = "code-analysis")]
-pub mod code_analysis;
 
 pub use annotation::*;
 pub use apk::*;
@@ -34,26 +26,18 @@ pub use dex_id::*;
 pub use dex_string::*;
 pub use dex_writer::*;
 pub use field::*;
-pub use hiddenapi::*;
 pub use instructions as ins;
-pub use instructions::*;
 pub use method::*;
 pub use method_handle::*;
 pub use scalar::*;
-pub use utils::*;
 pub use value::*;
-pub use visitor::*;
-
-#[cfg(feature = "code-analysis")]
-pub use code_analysis::*;
 
 #[cfg(test)]
 mod tests;
 
 /// Androscalpel.
-#[cfg(feature = "python")]
 #[pymodule]
-fn androscalpel(py: Python, m: &Bound<'_, PyModule>) -> PyResult<()> {
+fn androscalpel(py: Python, m: &PyModule) -> PyResult<()> {
     pyo3_log::init();
     m.add_class::<DexNull>()?;
     m.add_class::<DexBoolean>()?;
@@ -72,9 +56,15 @@ fn androscalpel(py: Python, m: &Bound<'_, PyModule>) -> PyResult<()> {
     m.add_class::<IdMethod>()?;
     m.add_class::<IdEnum>()?;
 
-    m.add_class::<HiddenApiData>()?;
-    m.add_class::<HiddenApiPermission>()?;
-    m.add_class::<HiddenApiDomain>()?;
+    m.add_class::<StaticPut>()?;
+    m.add_class::<StaticGet>()?;
+    m.add_class::<InstancePut>()?;
+    m.add_class::<InstanceGet>()?;
+    m.add_class::<InvokeStatic>()?;
+    m.add_class::<InvokeInstance>()?;
+    m.add_class::<InvokeConstructor>()?;
+    m.add_class::<InvokeDirect>()?;
+    m.add_class::<InvokeInterface>()?;
 
     m.add_class::<DexAnnotationItem>()?;
     m.add_class::<DexAnnotation>()?;
@@ -88,19 +78,211 @@ fn androscalpel(py: Python, m: &Bound<'_, PyModule>) -> PyResult<()> {
     m.add_class::<Apk>()?;
 
     let ins_module = PyModule::new(py, "ins")?;
-    androscalpel_ins(py, &ins_module)?;
-    m.add_submodule(&ins_module)?;
+    androscalpel_ins(py, ins_module)?;
+    m.add_submodule(ins_module)?;
     let utils_module = PyModule::new(py, "utils")?;
-    py_utils::export_module(py, &utils_module)?;
-    m.add_submodule(&utils_module)?;
+    py_utils::export_module(py, utils_module)?;
+    m.add_submodule(utils_module)?;
 
     Ok(())
 }
 
 /// Dalvik opcode for Androscalpel.
-#[cfg(feature = "python")]
-fn androscalpel_ins(_py: Python, m: &Bound<'_, PyModule>) -> PyResult<()> {
+fn androscalpel_ins(_py: Python, m: &PyModule) -> PyResult<()> {
     m.add_class::<ins::CallSite>()?;
-    m.add_class::<ins::Instruction>()?;
+    m.add_class::<ins::Nop>()?;
+    m.add_class::<ins::Move>()?;
+    m.add_class::<ins::MoveWide>()?;
+    m.add_class::<ins::MoveObject>()?;
+    m.add_class::<ins::MoveResult>()?;
+    m.add_class::<ins::MoveResultWide>()?;
+    m.add_class::<ins::MoveResultObject>()?;
+    m.add_class::<ins::MoveException>()?;
+    m.add_class::<ins::ReturnVoid>()?;
+    m.add_class::<ins::Return>()?;
+    m.add_class::<ins::ReturnWide>()?;
+    m.add_class::<ins::ReturnObject>()?;
+    m.add_class::<ins::Const>()?;
+    m.add_class::<ins::ConstWide>()?;
+    m.add_class::<ins::ConstString>()?;
+    m.add_class::<ins::ConstClass>()?;
+    m.add_class::<ins::MonitorEnter>()?;
+    m.add_class::<ins::MonitorExit>()?;
+    m.add_class::<ins::CheckCast>()?;
+    m.add_class::<ins::InstanceOf>()?;
+    m.add_class::<ins::ArrayLength>()?;
+    m.add_class::<ins::NewInstance>()?;
+    m.add_class::<ins::NewArray>()?;
+    m.add_class::<ins::FilledNewArray>()?;
+    m.add_class::<ins::FillArrayData>()?;
+    m.add_class::<ins::Throw>()?;
+    m.add_class::<ins::Goto>()?;
+    m.add_class::<ins::Switch>()?;
+    m.add_class::<ins::CmpLFloat>()?;
+    m.add_class::<ins::CmpGFloat>()?;
+    m.add_class::<ins::CmpLDouble>()?;
+    m.add_class::<ins::CmpGDouble>()?;
+    m.add_class::<ins::CmpLong>()?;
+    m.add_class::<ins::IfEq>()?;
+    m.add_class::<ins::IfNe>()?;
+    m.add_class::<ins::IfLt>()?;
+    m.add_class::<ins::IfGe>()?;
+    m.add_class::<ins::IfGt>()?;
+    m.add_class::<ins::IfLe>()?;
+    m.add_class::<ins::IfEqZ>()?;
+    m.add_class::<ins::IfNeZ>()?;
+    m.add_class::<ins::IfLtZ>()?;
+    m.add_class::<ins::IfGeZ>()?;
+    m.add_class::<ins::IfGtZ>()?;
+    m.add_class::<ins::IfLeZ>()?;
+    m.add_class::<ins::AGet>()?;
+    m.add_class::<ins::AGetWide>()?;
+    m.add_class::<ins::AGetObject>()?;
+    m.add_class::<ins::AGetBoolean>()?;
+    m.add_class::<ins::AGetByte>()?;
+    m.add_class::<ins::AGetChar>()?;
+    m.add_class::<ins::AGetShort>()?;
+    m.add_class::<ins::APut>()?;
+    m.add_class::<ins::APutWide>()?;
+    m.add_class::<ins::APutObject>()?;
+    m.add_class::<ins::APutBoolean>()?;
+    m.add_class::<ins::APutByte>()?;
+    m.add_class::<ins::APutChar>()?;
+    m.add_class::<ins::APutShort>()?;
+    m.add_class::<ins::IGet>()?;
+    m.add_class::<ins::IGetWide>()?;
+    m.add_class::<ins::IGetObject>()?;
+    m.add_class::<ins::IGetBoolean>()?;
+    m.add_class::<ins::IGetByte>()?;
+    m.add_class::<ins::IGetChar>()?;
+    m.add_class::<ins::IGetShort>()?;
+    m.add_class::<ins::IPut>()?;
+    m.add_class::<ins::IPutWide>()?;
+    m.add_class::<ins::IPutObject>()?;
+    m.add_class::<ins::IPutBoolean>()?;
+    m.add_class::<ins::IPutByte>()?;
+    m.add_class::<ins::IPutChar>()?;
+    m.add_class::<ins::IPutShort>()?;
+    m.add_class::<ins::SGet>()?;
+    m.add_class::<ins::SGetWide>()?;
+    m.add_class::<ins::SGetObject>()?;
+    m.add_class::<ins::SGetBoolean>()?;
+    m.add_class::<ins::SGetByte>()?;
+    m.add_class::<ins::SGetChar>()?;
+    m.add_class::<ins::SGetShort>()?;
+    m.add_class::<ins::SPut>()?;
+    m.add_class::<ins::SPutWide>()?;
+    m.add_class::<ins::SPutObject>()?;
+    m.add_class::<ins::SPutBoolean>()?;
+    m.add_class::<ins::SPutByte>()?;
+    m.add_class::<ins::SPutChar>()?;
+    m.add_class::<ins::SPutShort>()?;
+    m.add_class::<ins::InvokeVirtual>()?;
+    m.add_class::<ins::InvokeSuper>()?;
+    m.add_class::<ins::InvokeDirect>()?;
+    m.add_class::<ins::InvokeStatic>()?;
+    m.add_class::<ins::InvokeInterface>()?;
+    m.add_class::<ins::NegInt>()?;
+    m.add_class::<ins::NotInt>()?;
+    m.add_class::<ins::NegLong>()?;
+    m.add_class::<ins::NotLong>()?;
+    m.add_class::<ins::NegFloat>()?;
+    m.add_class::<ins::NegDouble>()?;
+    m.add_class::<ins::IntToLong>()?;
+    m.add_class::<ins::IntToFloat>()?;
+    m.add_class::<ins::IntToDouble>()?;
+    m.add_class::<ins::LongToInt>()?;
+    m.add_class::<ins::LongToFloat>()?;
+    m.add_class::<ins::LongToDouble>()?;
+    m.add_class::<ins::FloatToInt>()?;
+    m.add_class::<ins::FloatToLong>()?;
+    m.add_class::<ins::FloatToDouble>()?;
+    m.add_class::<ins::DoubleToInt>()?;
+    m.add_class::<ins::DoubleToLong>()?;
+    m.add_class::<ins::DoubleToFloat>()?;
+    m.add_class::<ins::IntToByte>()?;
+    m.add_class::<ins::IntToChar>()?;
+    m.add_class::<ins::IntToShort>()?;
+    m.add_class::<ins::AddInt>()?;
+    m.add_class::<ins::SubInt>()?;
+    m.add_class::<ins::MulInt>()?;
+    m.add_class::<ins::DivInt>()?;
+    m.add_class::<ins::RemInt>()?;
+    m.add_class::<ins::AndInt>()?;
+    m.add_class::<ins::OrInt>()?;
+    m.add_class::<ins::XorInt>()?;
+    m.add_class::<ins::ShlInt>()?;
+    m.add_class::<ins::ShrInt>()?;
+    m.add_class::<ins::UshrInt>()?;
+    m.add_class::<ins::AddLong>()?;
+    m.add_class::<ins::SubLong>()?;
+    m.add_class::<ins::MulLong>()?;
+    m.add_class::<ins::DivLong>()?;
+    m.add_class::<ins::RemLong>()?;
+    m.add_class::<ins::AndLong>()?;
+    m.add_class::<ins::OrLong>()?;
+    m.add_class::<ins::XorLong>()?;
+    m.add_class::<ins::ShlLong>()?;
+    m.add_class::<ins::ShrLong>()?;
+    m.add_class::<ins::UshrLong>()?;
+    m.add_class::<ins::AddFloat>()?;
+    m.add_class::<ins::SubFloat>()?;
+    m.add_class::<ins::MulFloat>()?;
+    m.add_class::<ins::DivFloat>()?;
+    m.add_class::<ins::RemFloat>()?;
+    m.add_class::<ins::AddDouble>()?;
+    m.add_class::<ins::SubDouble>()?;
+    m.add_class::<ins::MulDouble>()?;
+    m.add_class::<ins::DivDouble>()?;
+    m.add_class::<ins::RemDouble>()?;
+    m.add_class::<ins::AddInt2Addr>()?;
+    m.add_class::<ins::SubInt2Addr>()?;
+    m.add_class::<ins::MulInt2Addr>()?;
+    m.add_class::<ins::DivInt2Addr>()?;
+    m.add_class::<ins::RemInt2Addr>()?;
+    m.add_class::<ins::AndInt2Addr>()?;
+    m.add_class::<ins::OrInt2Addr>()?;
+    m.add_class::<ins::XorInt2Addr>()?;
+    m.add_class::<ins::ShlInt2Addr>()?;
+    m.add_class::<ins::ShrInt2Addr>()?;
+    m.add_class::<ins::UshrInt2Addr>()?;
+    m.add_class::<ins::AddLong2Addr>()?;
+    m.add_class::<ins::SubLong2Addr>()?;
+    m.add_class::<ins::MulLong2Addr>()?;
+    m.add_class::<ins::DivLong2Addr>()?;
+    m.add_class::<ins::RemLong2Addr>()?;
+    m.add_class::<ins::AndLong2Addr>()?;
+    m.add_class::<ins::OrLong2Addr>()?;
+    m.add_class::<ins::XorLong2Addr>()?;
+    m.add_class::<ins::ShlLong2Addr>()?;
+    m.add_class::<ins::ShrLong2Addr>()?;
+    m.add_class::<ins::UshrLong2Addr>()?;
+    m.add_class::<ins::AddFloat2Addr>()?;
+    m.add_class::<ins::SubFloat2Addr>()?;
+    m.add_class::<ins::MulFloat2Addr>()?;
+    m.add_class::<ins::DivFloat2Addr>()?;
+    m.add_class::<ins::RemFloat2Addr>()?;
+    m.add_class::<ins::AddDouble2Addr>()?;
+    m.add_class::<ins::SubDouble2Addr>()?;
+    m.add_class::<ins::MulDouble2Addr>()?;
+    m.add_class::<ins::DivDouble2Addr>()?;
+    m.add_class::<ins::RemDouble2Addr>()?;
+    m.add_class::<ins::AddIntLit>()?;
+    m.add_class::<ins::RsubIntLit>()?;
+    m.add_class::<ins::MulIntLit>()?;
+    m.add_class::<ins::DivIntLit>()?;
+    m.add_class::<ins::RemIntLit>()?;
+    m.add_class::<ins::AndIntLit>()?;
+    m.add_class::<ins::OrIntLit>()?;
+    m.add_class::<ins::XorIntLit>()?;
+    m.add_class::<ins::ShlIntLit>()?;
+    m.add_class::<ins::ShrIntLit>()?;
+    m.add_class::<ins::UshrIntLit>()?;
+    m.add_class::<ins::InvokePolymorphic>()?;
+    m.add_class::<ins::InvokeCustom>()?;
+    m.add_class::<ins::ConstMethodHandle>()?;
+    m.add_class::<ins::ConstMethodType>()?;
+    m.add_class::<ins::Try>()?;
+    m.add_class::<ins::Label>()?;
     Ok(())
 }

androscalpel/src/method.rs

@@ -3,77 +3,73 @@
 use serde::{Deserialize, Serialize};
 use std::collections::HashSet;
 
-#[cfg(feature = "python")]
 use pyo3::prelude::*;
 
 use crate::{
-    Code, DexAnnotationItem, DexString, HiddenApiData, IdField, IdMethod, IdMethodType, IdType,
-    MethodHandle, Result, Visitable, VisitableMut, Visitor, VisitorMut,
+    Code, DexAnnotationItem, DexString, IdField, IdMethod, IdMethodType, IdType, MethodHandle,
+    Result,
 };
 use androscalpel_serializer::consts::*;
 
 /// Represent a method.
-#[cfg_attr(feature = "python", pyclass(eq))]
-#[derive(Debug, Clone, Deserialize, Serialize, PartialEq)]
+#[pyclass]
+#[derive(Debug, Clone, PartialEq, Deserialize, Serialize)]
 pub struct Method {
     /// The structure used to reference this method.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub descriptor: IdMethod,
     /// The field visibility.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub visibility: MethodVisibility,
     /// Static methods do not take this in argument.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_static: bool,
     /// Final methods are not averridable.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_final: bool,
     /// Synchronized method automatically acquire their associated lock around call.
     /// Can only be set in native method, (`[Self::is_native] = true`).
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_synchronized: bool,
     /// Bridge are automatically added by the compiler as a type-safe bridge.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_bridge: bool,
     /// If the last argument should be treated as a "rest" argument by compiler
     /// (for method of variable number of argument).
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_varargs: bool,
     /// If the method is a native method.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_native: bool,
     /// Abstract methods are not implemented by the class.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_abstract: bool,
     /// If the method must use strict rules for floating point arithmetic.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_strictfp: bool,
     /// Synthetic method are not directly defined in the source code.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_synthetic: bool,
     /// If the method is a constructor.
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_constructor: bool,
     /// If the method is declared as synchronize (just indicatif)
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub is_declared_syncrhonized: bool,
     /// The annotations for this method
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub annotations: Vec<DexAnnotationItem>,
     /// The annotations for the parameters of this method method
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub parameters_annotations: Vec<Vec<DexAnnotationItem>>,
-    /// Hidden Api data.
-    #[cfg_attr(feature = "python", pyo3(get))]
-    pub hiddenapi: Option<HiddenApiData>,
 
     /// The code of the method
-    #[cfg_attr(feature = "python", pyo3(get))]
+    #[pyo3(get)]
     pub code: Option<Code>,
 }
 
 /// Represent the visibility of a field
-#[cfg_attr(feature = "python", pyclass(eq, eq_int))]
+#[pyclass]
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize, Serialize)]
 pub enum MethodVisibility {
     Public,
@@ -82,18 +78,18 @@ pub enum MethodVisibility {
     None_, // Actually quite common
 }
 
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl Method {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(descriptor: IdMethod) -> Self {
         // TODO: take code option as arg and set the default flags accordingly
         Self {
@@ -113,7 +109,6 @@ impl Method {
             annotations: vec![],
             parameters_annotations: vec![],
             code: None,
-            hiddenapi: None,
         }
     }
 
@@ -295,79 +290,7 @@ impl Method {
                 .any(|list| !list.is_empty())
     }
 
-    /// Compute the `ins_size` field. This is the number of register parameters, including the
-    /// `this` parameter for non-static methods.
-    /// This information is stored in the code item in dex files, but is not computable from the code
-    /// (as opposed to `outs_size`). The [`Method`] struct is needed to compute it.
-    pub fn ins_size(&self) -> u16 {
-        let mut ins = 0;
-        if !self.is_static {
-            ins += 1; // this
-        }
-        for param in &self.descriptor.proto.parameters {
-            if param.is_long() || param.is_double() {
-                ins += 2;
-            } else {
-                ins += 1;
-            }
-        }
-        ins
-    }
-
-    /// Check if the method is a platform method (ie in the android SDK or a hidden API).
-    #[cfg(feature = "platform-list")]
-    pub fn is_platform_method(&self) -> bool {
-        self.descriptor.is_platform_method()
-    }
-}
-
-impl<V: Visitor> Visitable<V> for Method {
-    fn default_visit(&self, v: &mut V) -> Result<()> {
-        v.visit_method_id(&self.descriptor)?;
-        v.visit_method_visibility(&self.visibility)?;
-        for annot in &self.annotations {
-            v.visit_annotation_item(annot)?;
-        }
-        for parameter_annotations in &self.parameters_annotations {
-            for annot in parameter_annotations {
-                v.visit_annotation_item(annot)?;
-            }
-        }
-        if let Some(hiddenapi) = &self.hiddenapi {
-            v.visit_hidden_api_data(hiddenapi)?;
-        }
-        if let Some(code) = &self.code {
-            v.visit_code(code)?;
-        }
-        Ok(())
-    }
-}
-impl<V: VisitorMut> VisitableMut<V> for Method {
-    fn default_visit_mut(self, v: &mut V) -> Result<Self> {
-        Ok(Self {
-            descriptor: v.visit_method_id(self.descriptor)?,
-            visibility: v.visit_method_visibility(self.visibility)?,
-            annotations: self
-                .annotations
-                .into_iter()
-                .map(|annot| v.visit_annotation_item(annot))
-                .collect::<Result<_>>()?,
-            parameters_annotations: self
-                .parameters_annotations
-                .into_iter()
-                .map(|parameter_annotations| {
-                    parameter_annotations
-                        .into_iter()
-                        .map(|annot| v.visit_annotation_item(annot))
-                        .collect::<Result<_>>()
-                })
-                .collect::<Result<_>>()?,
-            hiddenapi: self
-                .hiddenapi
-                .map(|hiddenapi| v.visit_hidden_api_data(hiddenapi))
-                .transpose()?,
-            code: self.code.map(|code| v.visit_code(code)).transpose()?,
-            ..self
-        })
+    pub fn __eq__(&self, other: &Self) -> bool {
+        self == other
     }
 }

androscalpel/src/method_handle.rs

@@ -4,151 +4,790 @@ use serde::{Deserialize, Serialize};
 use std::collections::HashSet;
 use std::hash::Hash;
 
-#[cfg(feature = "python")]
+use pyo3::exceptions::PyTypeError;
 use pyo3::prelude::*;
 
 use crate::dex_id::*;
-use crate::{
-    DexString, FieldIdCollector, MethodHandleCollector, MethodIdCollector, MethodTypeCollector,
-    Result, StringCollector, TypeCollector, Visitable, VisitableMut, Visitor, VisitorMut,
-};
+use crate::DexString;
+use crate::Result;
 
 /// The structure use to reference a method invocation.
-#[cfg_attr(feature = "python", pyclass(eq))]
 #[derive(Debug, Clone, PartialEq, Eq, Hash, Deserialize, Serialize)]
 pub enum MethodHandle {
-    StaticPut { field: IdField },
-    StaticGet { field: IdField },
-    InstancePut { field: IdField },
-    InstanceGet { field: IdField },
-    InvokeStatic { method: IdMethod },
-    InvokeInstance { method: IdMethod },
-    InvokeConstructor { method: IdMethod },
-    InvokeDirect { method: IdMethod },
-    InvokeInterface { method: IdMethod },
+    StaticPut(StaticPut),
+    StaticGet(StaticGet),
+    InstancePut(InstancePut),
+    InstanceGet(InstanceGet),
+    InvokeStatic(InvokeStatic),
+    InvokeInstance(InvokeInstance),
+    InvokeConstructor(InvokeConstructor),
+    InvokeDirect(InvokeDirect),
+    InvokeInterface(InvokeInterface),
 }
 
-impl<V: Visitor> Visitable<V> for MethodHandle {
-    fn default_visit(&self, v: &mut V) -> Result<()> {
-        match self {
-            Self::StaticPut { field } => v.visit_field_id(field)?,
-            Self::StaticGet { field } => v.visit_field_id(field)?,
-            Self::InstancePut { field } => v.visit_field_id(field)?,
-            Self::InstanceGet { field } => v.visit_field_id(field)?,
-            Self::InvokeStatic { method } => v.visit_method_id(method)?,
-            Self::InvokeInstance { method } => v.visit_method_id(method)?,
-            Self::InvokeConstructor { method } => v.visit_method_id(method)?,
-            Self::InvokeDirect { method } => v.visit_method_id(method)?,
-            Self::InvokeInterface { method } => v.visit_method_id(method)?,
-        }
-        Ok(())
-    }
-}
+#[pyclass]
+#[derive(Debug, Clone, PartialEq, Eq, Hash, Deserialize, Serialize)]
+pub struct StaticPut(pub IdField);
 
-impl<V: VisitorMut> VisitableMut<V> for MethodHandle {
-    fn default_visit_mut(self, v: &mut V) -> Result<Self> {
-        match self {
-            Self::StaticPut { field } => Ok(Self::StaticPut {
-                field: v.visit_field_id(field)?,
-            }),
-            Self::StaticGet { field } => Ok(Self::StaticGet {
-                field: v.visit_field_id(field)?,
-            }),
-            Self::InstancePut { field } => Ok(Self::InstancePut {
-                field: v.visit_field_id(field)?,
-            }),
-            Self::InstanceGet { field } => Ok(Self::InstanceGet {
-                field: v.visit_field_id(field)?,
-            }),
-            Self::InvokeStatic { method } => Ok(Self::InvokeStatic {
-                method: v.visit_method_id(method)?,
-            }),
-            Self::InvokeInstance { method } => Ok(Self::InvokeInstance {
-                method: v.visit_method_id(method)?,
-            }),
-            Self::InvokeConstructor { method } => Ok(Self::InvokeConstructor {
-                method: v.visit_method_id(method)?,
-            }),
-            Self::InvokeDirect { method } => Ok(Self::InvokeDirect {
-                method: v.visit_method_id(method)?,
-            }),
-            Self::InvokeInterface { method } => Ok(Self::InvokeInterface {
-                method: v.visit_method_id(method)?,
-            }),
-        }
-    }
-}
-
-#[cfg_attr(feature = "python", pymethods)]
-impl MethodHandle {
+#[pymethods]
+impl StaticPut {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
+    #[new]
+    pub fn new(val: IdField) -> Self {
+        Self(val)
+    }
+
+    pub fn get_field(&self) -> IdField {
+        self.0.clone()
+    }
+
     pub fn __str__(&self) -> String {
         self.__repr__()
     }
 
     pub fn __repr__(&self) -> String {
+        format!("StaticPut({})", self.0.__str__())
+    }
+
+    /// Return all strings referenced in the handle.
+    pub fn get_all_strings(&self) -> HashSet<DexString> {
+        self.0.get_all_strings()
+    }
+
+    /// Return all types referenced in the handle.
+    pub fn get_all_types(&self) -> HashSet<IdType> {
+        self.0.get_all_types()
+    }
+
+    /// Return all prototypes referenced in the handle.
+    pub fn get_all_protos(&self) -> HashSet<IdMethodType> {
+        HashSet::new()
+    }
+
+    /// Return all field ids referenced in the handle.
+    pub fn get_all_field_ids(&self) -> HashSet<IdField> {
+        let mut fields = HashSet::new();
+        fields.insert(self.0.clone());
+        fields
+    }
+
+    /// Return all method ids referenced in the handle.
+    pub fn get_all_method_ids(&self) -> HashSet<IdMethod> {
+        HashSet::new()
+    }
+
+    /// Return all method handles referenced in the handle.
+    pub fn get_all_method_handles(&self) -> HashSet<MethodHandle> {
+        let mut methods = HashSet::new();
+        methods.insert(MethodHandle::StaticPut(self.clone()));
+        methods
+    }
+}
+#[pyclass]
+#[derive(Debug, Clone, PartialEq, Eq, Hash, Deserialize, Serialize)]
+pub struct StaticGet(pub IdField);
+
+#[pymethods]
+impl StaticGet {
+    pub fn to_json(&self) -> Result<String> {
+        Ok(serde_json::to_string(self)?)
+    }
+
+    #[staticmethod]
+    pub fn from_json(json: &str) -> Result<Self> {
+        Ok(serde_json::from_str(json)?)
+    }
+
+    #[new]
+    pub fn new(val: IdField) -> Self {
+        Self(val)
+    }
+
+    pub fn get_field(&self) -> IdField {
+        self.0.clone()
+    }
+
+    pub fn __str__(&self) -> String {
+        self.__repr__()
+    }
+
+    pub fn __repr__(&self) -> String {
+        format!("StaticGet({})", self.0.__str__())
+    }
+
+    /// Return all strings referenced in the handle.
+    pub fn get_all_strings(&self) -> HashSet<DexString> {
+        self.0.get_all_strings()
+    }
+
+    /// Return all types referenced in the handle.
+    pub fn get_all_types(&self) -> HashSet<IdType> {
+        self.0.get_all_types()
+    }
+
+    /// Return all prototypes referenced in the handle.
+    pub fn get_all_protos(&self) -> HashSet<IdMethodType> {
+        HashSet::new()
+    }
+
+    /// Return all field ids referenced in the handle.
+    pub fn get_all_field_ids(&self) -> HashSet<IdField> {
+        let mut fields = HashSet::new();
+        fields.insert(self.0.clone());
+        fields
+    }
+
+    /// Return all method ids referenced in the handle.
+    pub fn get_all_method_ids(&self) -> HashSet<IdMethod> {
+        HashSet::new()
+    }
+
+    /// Return all method handles referenced in the handle.
+    pub fn get_all_method_handles(&self) -> HashSet<MethodHandle> {
+        let mut methods = HashSet::new();
+        methods.insert(MethodHandle::StaticGet(self.clone()));
+        methods
+    }
+}
+#[pyclass]
+#[derive(Debug, Clone, PartialEq, Eq, Hash, Deserialize, Serialize)]
+pub struct InstancePut(pub IdField);
+
+#[pymethods]
+impl InstancePut {
+    pub fn to_json(&self) -> Result<String> {
+        Ok(serde_json::to_string(self)?)
+    }
+
+    #[staticmethod]
+    pub fn from_json(json: &str) -> Result<Self> {
+        Ok(serde_json::from_str(json)?)
+    }
+
+    #[new]
+    pub fn new(val: IdField) -> Self {
+        Self(val)
+    }
+
+    pub fn get_field(&self) -> IdField {
+        self.0.clone()
+    }
+
+    pub fn __str__(&self) -> String {
+        self.__repr__()
+    }
+
+    pub fn __repr__(&self) -> String {
+        format!("InstancePut({})", self.0.__str__())
+    }
+
+    /// Return all strings referenced in the handle.
+    pub fn get_all_strings(&self) -> HashSet<DexString> {
+        self.0.get_all_strings()
+    }
+
+    /// Return all types referenced in the handle.
+    pub fn get_all_types(&self) -> HashSet<IdType> {
+        self.0.get_all_types()
+    }
+
+    /// Return all prototypes referenced in the handle.
+    pub fn get_all_protos(&self) -> HashSet<IdMethodType> {
+        HashSet::new()
+    }
+
+    /// Return all field ids referenced in the handle.
+    pub fn get_all_field_ids(&self) -> HashSet<IdField> {
+        let mut fields = HashSet::new();
+        fields.insert(self.0.clone());
+        fields
+    }
+
+    /// Return all method ids referenced in the handle.
+    pub fn get_all_method_ids(&self) -> HashSet<IdMethod> {
+        HashSet::new()
+    }
+
+    /// Return all method handles referenced in the handle.
+    pub fn get_all_method_handles(&self) -> HashSet<MethodHandle> {
+        let mut methods = HashSet::new();
+        methods.insert(MethodHandle::InstancePut(self.clone()));
+        methods
+    }
+}
+#[pyclass]
+#[derive(Debug, Clone, PartialEq, Eq, Hash, Deserialize, Serialize)]
+pub struct InstanceGet(pub IdField);
+
+#[pymethods]
+impl InstanceGet {
+    pub fn to_json(&self) -> Result<String> {
+        Ok(serde_json::to_string(self)?)
+    }
+
+    #[staticmethod]
+    pub fn from_json(json: &str) -> Result<Self> {
+        Ok(serde_json::from_str(json)?)
+    }
+
+    #[new]
+    pub fn new(val: IdField) -> Self {
+        Self(val)
+    }
+
+    pub fn get_field(&self) -> IdField {
+        self.0.clone()
+    }
+
+    pub fn __str__(&self) -> String {
+        self.__repr__()
+    }
+
+    pub fn __repr__(&self) -> String {
+        format!("InstanceGet({})", self.0.__str__())
+    }
+
+    /// Return all strings referenced in the handle.
+    pub fn get_all_strings(&self) -> HashSet<DexString> {
+        self.0.get_all_strings()
+    }
+
+    /// Return all types referenced in the handle.
+    pub fn get_all_types(&self) -> HashSet<IdType> {
+        self.0.get_all_types()
+    }
+
+    /// Return all prototypes referenced in the handle.
+    pub fn get_all_protos(&self) -> HashSet<IdMethodType> {
+        HashSet::new()
+    }
+
+    /// Return all field ids referenced in the handle.
+    pub fn get_all_field_ids(&self) -> HashSet<IdField> {
+        let mut fields = HashSet::new();
+        fields.insert(self.0.clone());
+        fields
+    }
+
+    /// Return all method ids referenced in the handle.
+    pub fn get_all_method_ids(&self) -> HashSet<IdMethod> {
+        HashSet::new()
+    }
+
+    /// Return all method handles referenced in the handle.
+    pub fn get_all_method_handles(&self) -> HashSet<MethodHandle> {
+        let mut methods = HashSet::new();
+        methods.insert(MethodHandle::InstanceGet(self.clone()));
+        methods
+    }
+}
+
+#[pyclass]
+#[derive(Debug, Clone, PartialEq, Eq, Hash, Deserialize, Serialize)]
+pub struct InvokeStatic(pub IdMethod);
+
+#[pymethods]
+impl InvokeStatic {
+    pub fn to_json(&self) -> Result<String> {
+        Ok(serde_json::to_string(self)?)
+    }
+
+    #[staticmethod]
+    pub fn from_json(json: &str) -> Result<Self> {
+        Ok(serde_json::from_str(json)?)
+    }
+
+    #[new]
+    pub fn new(val: IdMethod) -> Self {
+        Self(val)
+    }
+
+    pub fn get_method(&self) -> IdMethod {
+        self.0.clone()
+    }
+
+    pub fn __str__(&self) -> String {
+        self.__repr__()
+    }
+
+    pub fn __repr__(&self) -> String {
+        format!("InvokeStatic({})", self.0.__str__())
+    }
+
+    /// Return all strings referenced in the handle.
+    pub fn get_all_strings(&self) -> HashSet<DexString> {
+        self.0.get_all_strings()
+    }
+
+    /// Return all types referenced in the handle.
+    pub fn get_all_types(&self) -> HashSet<IdType> {
+        self.0.get_all_types()
+    }
+
+    /// Return all prototypes referenced in the handle.
+    pub fn get_all_protos(&self) -> HashSet<IdMethodType> {
+        self.0.get_all_protos()
+    }
+
+    /// Return all field ids referenced in the handle.
+    pub fn get_all_field_ids(&self) -> HashSet<IdField> {
+        HashSet::new()
+    }
+
+    /// Return all method ids referenced in the handle.
+    pub fn get_all_method_ids(&self) -> HashSet<IdMethod> {
+        let mut methods = HashSet::new();
+        methods.insert(self.0.clone());
+        methods
+    }
+
+    /// Return all method handles referenced in the handle.
+    pub fn get_all_method_handles(&self) -> HashSet<MethodHandle> {
+        let mut methods = HashSet::new();
+        methods.insert(MethodHandle::InvokeStatic(self.clone()));
+        methods
+    }
+}
+
+#[pyclass]
+#[derive(Debug, Clone, PartialEq, Eq, Hash, Deserialize, Serialize)]
+pub struct InvokeInstance(pub IdMethod);
+
+#[pymethods]
+impl InvokeInstance {
+    pub fn to_json(&self) -> Result<String> {
+        Ok(serde_json::to_string(self)?)
+    }
+
+    #[staticmethod]
+    pub fn from_json(json: &str) -> Result<Self> {
+        Ok(serde_json::from_str(json)?)
+    }
+
+    #[new]
+    pub fn new(val: IdMethod) -> Self {
+        Self(val)
+    }
+
+    pub fn get_method(&self) -> IdMethod {
+        self.0.clone()
+    }
+
+    pub fn __str__(&self) -> String {
+        self.__repr__()
+    }
+
+    pub fn __repr__(&self) -> String {
+        format!("InvokeInstance({})", self.0.__str__())
+    }
+
+    /// Return all strings referenced in the handle.
+    pub fn get_all_strings(&self) -> HashSet<DexString> {
+        self.0.get_all_strings()
+    }
+
+    /// Return all types referenced in the handle.
+    pub fn get_all_types(&self) -> HashSet<IdType> {
+        self.0.get_all_types()
+    }
+
+    /// Return all prototypes referenced in the handle.
+    pub fn get_all_protos(&self) -> HashSet<IdMethodType> {
+        self.0.get_all_protos()
+    }
+
+    /// Return all field ids referenced in the handle.
+    pub fn get_all_field_ids(&self) -> HashSet<IdField> {
+        HashSet::new()
+    }
+
+    /// Return all method ids referenced in the handle.
+    pub fn get_all_method_ids(&self) -> HashSet<IdMethod> {
+        let mut methods = HashSet::new();
+        methods.insert(self.0.clone());
+        methods
+    }
+
+    /// Return all method handles referenced in the handle.
+    pub fn get_all_method_handles(&self) -> HashSet<MethodHandle> {
+        let mut methods = HashSet::new();
+        methods.insert(MethodHandle::InvokeInstance(self.clone()));
+        methods
+    }
+}
+
+#[pyclass]
+#[derive(Debug, Clone, PartialEq, Eq, Hash, Deserialize, Serialize)]
+pub struct InvokeConstructor(pub IdMethod);
+
+#[pymethods]
+impl InvokeConstructor {
+    pub fn to_json(&self) -> Result<String> {
+        Ok(serde_json::to_string(self)?)
+    }
+
+    #[staticmethod]
+    pub fn from_json(json: &str) -> Result<Self> {
+        Ok(serde_json::from_str(json)?)
+    }
+
+    #[new]
+    pub fn new(val: IdMethod) -> Self {
+        Self(val)
+    }
+
+    pub fn get_method(&self) -> IdMethod {
+        self.0.clone()
+    }
+
+    pub fn __str__(&self) -> String {
+        self.__repr__()
+    }
+
+    pub fn __repr__(&self) -> String {
+        format!("InvokeConstructor({})", self.0.__str__())
+    }
+
+    /// Return all strings referenced in the handle.
+    pub fn get_all_strings(&self) -> HashSet<DexString> {
+        self.0.get_all_strings()
+    }
+
+    /// Return all types referenced in the handle.
+    pub fn get_all_types(&self) -> HashSet<IdType> {
+        self.0.get_all_types()
+    }
+
+    /// Return all prototypes referenced in the handle.
+    pub fn get_all_protos(&self) -> HashSet<IdMethodType> {
+        self.0.get_all_protos()
+    }
+
+    /// Return all field ids referenced in the handle.
+    pub fn get_all_field_ids(&self) -> HashSet<IdField> {
+        HashSet::new()
+    }
+
+    /// Return all method ids referenced in the handle.
+    pub fn get_all_method_ids(&self) -> HashSet<IdMethod> {
+        let mut methods = HashSet::new();
+        methods.insert(self.0.clone());
+        methods
+    }
+
+    /// Return all method handles referenced in the handle.
+    pub fn get_all_method_handles(&self) -> HashSet<MethodHandle> {
+        let mut methods = HashSet::new();
+        methods.insert(MethodHandle::InvokeConstructor(self.clone()));
+        methods
+    }
+}
+
+#[pyclass]
+#[derive(Debug, Clone, PartialEq, Eq, Hash, Deserialize, Serialize)]
+pub struct InvokeDirect(pub IdMethod);
+
+#[pymethods]
+impl InvokeDirect {
+    pub fn to_json(&self) -> Result<String> {
+        Ok(serde_json::to_string(self)?)
+    }
+
+    #[staticmethod]
+    pub fn from_json(json: &str) -> Result<Self> {
+        Ok(serde_json::from_str(json)?)
+    }
+
+    #[new]
+    pub fn new(val: IdMethod) -> Self {
+        Self(val)
+    }
+
+    pub fn get_method(&self) -> IdMethod {
+        self.0.clone()
+    }
+
+    pub fn __str__(&self) -> String {
+        self.__repr__()
+    }
+
+    pub fn __repr__(&self) -> String {
+        format!("InvokeDirect({})", self.0.__str__())
+    }
+
+    /// Return all strings referenced in the handle.
+    pub fn get_all_strings(&self) -> HashSet<DexString> {
+        self.0.get_all_strings()
+    }
+
+    /// Return all types referenced in the handle.
+    pub fn get_all_types(&self) -> HashSet<IdType> {
+        self.0.get_all_types()
+    }
+
+    /// Return all prototypes referenced in the handle.
+    pub fn get_all_protos(&self) -> HashSet<IdMethodType> {
+        self.0.get_all_protos()
+    }
+
+    /// Return all field ids referenced in the handle.
+    pub fn get_all_field_ids(&self) -> HashSet<IdField> {
+        HashSet::new()
+    }
+
+    /// Return all method ids referenced in the handle.
+    pub fn get_all_method_ids(&self) -> HashSet<IdMethod> {
+        let mut methods = HashSet::new();
+        methods.insert(self.0.clone());
+        methods
+    }
+
+    /// Return all method handles referenced in the handle.
+    pub fn get_all_method_handles(&self) -> HashSet<MethodHandle> {
+        let mut methods = HashSet::new();
+        methods.insert(MethodHandle::InvokeDirect(self.clone()));
+        methods
+    }
+}
+
+#[pyclass]
+#[derive(Debug, Clone, PartialEq, Eq, Hash, Deserialize, Serialize)]
+pub struct InvokeInterface(pub IdMethod);
+
+#[pymethods]
+impl InvokeInterface {
+    pub fn to_json(&self) -> Result<String> {
+        Ok(serde_json::to_string(self)?)
+    }
+
+    #[staticmethod]
+    pub fn from_json(json: &str) -> Result<Self> {
+        Ok(serde_json::from_str(json)?)
+    }
+
+    #[new]
+    pub fn new(val: IdMethod) -> Self {
+        Self(val)
+    }
+
+    pub fn get_method(&self) -> IdMethod {
+        self.0.clone()
+    }
+
+    pub fn __str__(&self) -> String {
+        self.__repr__()
+    }
+
+    pub fn __repr__(&self) -> String {
+        format!("InvokeInterface({})", self.0.__str__())
+    }
+
+    /// Return all strings referenced in the handle.
+    pub fn get_all_strings(&self) -> HashSet<DexString> {
+        self.0.get_all_strings()
+    }
+
+    /// Return all types referenced in the handle.
+    pub fn get_all_types(&self) -> HashSet<IdType> {
+        self.0.get_all_types()
+    }
+
+    /// Return all prototypes referenced in the handle.
+    pub fn get_all_protos(&self) -> HashSet<IdMethodType> {
+        self.0.get_all_protos()
+    }
+
+    /// Return all field ids referenced in the handle.
+    pub fn get_all_field_ids(&self) -> HashSet<IdField> {
+        HashSet::new()
+    }
+
+    /// Return all method ids referenced in the handle.
+    pub fn get_all_method_ids(&self) -> HashSet<IdMethod> {
+        let mut methods = HashSet::new();
+        methods.insert(self.0.clone());
+        methods
+    }
+
+    /// Return all method handles referenced in the handle.
+    pub fn get_all_method_handles(&self) -> HashSet<MethodHandle> {
+        let mut methods = HashSet::new();
+        methods.insert(MethodHandle::InvokeInterface(self.clone()));
+        methods
+    }
+}
+
+impl<'source> FromPyObject<'source> for MethodHandle {
+    fn extract(ob: &'source PyAny) -> PyResult<Self> {
+        if let Ok(val) = StaticPut::extract(ob) {
+            Ok(Self::StaticPut(val))
+        } else if let Ok(val) = StaticGet::extract(ob) {
+            Ok(Self::StaticGet(val))
+        } else if let Ok(val) = InstancePut::extract(ob) {
+            Ok(Self::InstancePut(val))
+        } else if let Ok(val) = InstanceGet::extract(ob) {
+            Ok(Self::InstanceGet(val))
+        } else if let Ok(val) = InvokeStatic::extract(ob) {
+            Ok(Self::InvokeStatic(val))
+        } else if let Ok(val) = InvokeInstance::extract(ob) {
+            Ok(Self::InvokeInstance(val))
+        } else if let Ok(val) = InvokeConstructor::extract(ob) {
+            Ok(Self::InvokeConstructor(val))
+        } else if let Ok(val) = InvokeDirect::extract(ob) {
+            Ok(Self::InvokeDirect(val))
+        } else if let Ok(val) = InvokeInterface::extract(ob) {
+            Ok(Self::InvokeInterface(val))
+        } else {
+            Err(PyErr::new::<PyTypeError, _>(format!(
+                "{} is not a castable as a MethodHandle",
+                ob.repr()?
+            )))
+        }
+    }
+}
+
+impl IntoPy<PyObject> for MethodHandle {
+    fn into_py(self, py: Python<'_>) -> PyObject {
         match self {
-            Self::StaticPut { field } => format!("StaticPut({})", field.__str__()),
-            Self::StaticGet { field } => format!("StaticGet({})", field.__str__()),
-            Self::InstancePut { field } => format!("InstancePut({})", field.__str__()),
-            Self::InstanceGet { field } => format!("InstanceGet({})", field.__str__()),
-            Self::InvokeStatic { method } => format!("InvokeStatic({})", method.__str__()),
-            Self::InvokeInstance { method } => format!("InvokeInstance({})", method.__str__()),
-            Self::InvokeConstructor { method } => {
-                format!("InvokeConstructor({})", method.__str__())
-            }
-            Self::InvokeDirect { method } => format!("InvokeDirect({})", method.__str__()),
-            Self::InvokeInterface { method } => format!("InvokeInterface({})", method.__str__()),
+            Self::StaticPut(val) => val.into_py(py),
+            Self::StaticGet(val) => val.into_py(py),
+            Self::InstancePut(val) => val.into_py(py),
+            Self::InstanceGet(val) => val.into_py(py),
+            Self::InvokeStatic(val) => val.into_py(py),
+            Self::InvokeInstance(val) => val.into_py(py),
+            Self::InvokeConstructor(val) => val.into_py(py),
+            Self::InvokeDirect(val) => val.into_py(py),
+            Self::InvokeInterface(val) => val.into_py(py),
+        }
+    }
+}
+
+impl MethodHandle {
+    // Not exposed to python, but meh, let's keep it coherent
+    pub fn __str__(&self) -> String {
+        match self {
+            Self::StaticPut(val) => val.__str__(),
+            Self::StaticGet(val) => val.__str__(),
+            Self::InstancePut(val) => val.__str__(),
+            Self::InstanceGet(val) => val.__str__(),
+            Self::InvokeStatic(val) => val.__str__(),
+            Self::InvokeInstance(val) => val.__str__(),
+            Self::InvokeConstructor(val) => val.__str__(),
+            Self::InvokeDirect(val) => val.__str__(),
+            Self::InvokeInterface(val) => val.__str__(),
+        }
+    }
+
+    // Not exposed to python, but meh, let's keep it coherent
+    pub fn __repr__(&self) -> String {
+        match self {
+            Self::StaticPut(val) => val.__repr__(),
+            Self::StaticGet(val) => val.__repr__(),
+            Self::InstancePut(val) => val.__repr__(),
+            Self::InstanceGet(val) => val.__repr__(),
+            Self::InvokeStatic(val) => val.__repr__(),
+            Self::InvokeInstance(val) => val.__repr__(),
+            Self::InvokeConstructor(val) => val.__str__(),
+            Self::InvokeDirect(val) => val.__repr__(),
+            Self::InvokeInterface(val) => val.__repr__(),
         }
     }
 
     /// Return all strings referenced in the Handle.
     pub fn get_all_strings(&self) -> HashSet<DexString> {
-        let mut visitor = StringCollector::default();
-        visitor.visit_method_handle(self).unwrap();
-        visitor.result()
+        match self {
+            Self::StaticPut(val) => val.get_all_strings(),
+            Self::StaticGet(val) => val.get_all_strings(),
+            Self::InstancePut(val) => val.get_all_strings(),
+            Self::InstanceGet(val) => val.get_all_strings(),
+            Self::InvokeStatic(val) => val.get_all_strings(),
+            Self::InvokeInstance(val) => val.get_all_strings(),
+            Self::InvokeConstructor(val) => val.get_all_strings(),
+            Self::InvokeDirect(val) => val.get_all_strings(),
+            Self::InvokeInterface(val) => val.get_all_strings(),
+        }
     }
 
     /// Return all types referenced in the Handle.
     pub fn get_all_types(&self) -> HashSet<IdType> {
-        let mut visitor = TypeCollector::default();
-        visitor.visit_method_handle(self).unwrap();
-        visitor.result()
+        match self {
+            Self::StaticPut(val) => val.get_all_types(),
+            Self::StaticGet(val) => val.get_all_types(),
+            Self::InstancePut(val) => val.get_all_types(),
+            Self::InstanceGet(val) => val.get_all_types(),
+            Self::InvokeStatic(val) => val.get_all_types(),
+            Self::InvokeInstance(val) => val.get_all_types(),
+            Self::InvokeConstructor(val) => val.get_all_types(),
+            Self::InvokeDirect(val) => val.get_all_types(),
+            Self::InvokeInterface(val) => val.get_all_types(),
+        }
     }
 
     /// Return all prototypes referenced in the handle.
     pub fn get_all_protos(&self) -> HashSet<IdMethodType> {
-        let mut visitor = MethodTypeCollector::default();
-        visitor.visit_method_handle(self).unwrap();
-        visitor.result()
+        match self {
+            Self::StaticPut(val) => val.get_all_protos(),
+            Self::StaticGet(val) => val.get_all_protos(),
+            Self::InstancePut(val) => val.get_all_protos(),
+            Self::InstanceGet(val) => val.get_all_protos(),
+            Self::InvokeStatic(val) => val.get_all_protos(),
+            Self::InvokeInstance(val) => val.get_all_protos(),
+            Self::InvokeConstructor(val) => val.get_all_protos(),
+            Self::InvokeDirect(val) => val.get_all_protos(),
+            Self::InvokeInterface(val) => val.get_all_protos(),
+        }
     }
 
     /// Return all field ids referenced in the handle.
     pub fn get_all_field_ids(&self) -> HashSet<IdField> {
-        let mut visitor = FieldIdCollector::default();
-        visitor.visit_method_handle(self).unwrap();
-        visitor.result()
+        match self {
+            Self::StaticPut(val) => val.get_all_field_ids(),
+            Self::StaticGet(val) => val.get_all_field_ids(),
+            Self::InstancePut(val) => val.get_all_field_ids(),
+            Self::InstanceGet(val) => val.get_all_field_ids(),
+            Self::InvokeStatic(val) => val.get_all_field_ids(),
+            Self::InvokeInstance(val) => val.get_all_field_ids(),
+            Self::InvokeConstructor(val) => val.get_all_field_ids(),
+            Self::InvokeDirect(val) => val.get_all_field_ids(),
+            Self::InvokeInterface(val) => val.get_all_field_ids(),
+        }
     }
 
     /// Return all method ids referenced in the handle.
     pub fn get_all_method_ids(&self) -> HashSet<IdMethod> {
-        let mut visitor = MethodIdCollector::default();
-        visitor.visit_method_handle(self).unwrap();
-        visitor.result()
+        match self {
+            Self::StaticPut(val) => val.get_all_method_ids(),
+            Self::StaticGet(val) => val.get_all_method_ids(),
+            Self::InstancePut(val) => val.get_all_method_ids(),
+            Self::InstanceGet(val) => val.get_all_method_ids(),
+            Self::InvokeStatic(val) => val.get_all_method_ids(),
+            Self::InvokeInstance(val) => val.get_all_method_ids(),
+            Self::InvokeConstructor(val) => val.get_all_method_ids(),
+            Self::InvokeDirect(val) => val.get_all_method_ids(),
+            Self::InvokeInterface(val) => val.get_all_method_ids(),
+        }
     }
 
     /// Return all method handles referenced in the handle.
     pub fn get_all_method_handles(&self) -> HashSet<MethodHandle> {
-        let mut visitor = MethodHandleCollector::default();
-        visitor.visit_method_handle(self).unwrap();
-        visitor.result()
+        match self {
+            Self::StaticPut(val) => val.get_all_method_handles(),
+            Self::StaticGet(val) => val.get_all_method_handles(),
+            Self::InstancePut(val) => val.get_all_method_handles(),
+            Self::InstanceGet(val) => val.get_all_method_handles(),
+            Self::InvokeStatic(val) => val.get_all_method_handles(),
+            Self::InvokeInstance(val) => val.get_all_method_handles(),
+            Self::InvokeConstructor(val) => val.get_all_method_handles(),
+            Self::InvokeDirect(val) => val.get_all_method_handles(),
+            Self::InvokeInterface(val) => val.get_all_method_handles(),
+        }
     }
 }

androscalpel/src/py_utils.rs

@@ -1,18 +1,13 @@
 //! Function that are can be usefull on the python side.
 
 use pyo3::prelude::*;
-use pyo3::types::{PyBytes, PyBytesMethods};
+use pyo3::types::PyBytes;
 
-use std::collections::{HashMap, HashSet};
-use std::fs::File;
-use std::io::{Cursor, Seek, SeekFrom};
+use std::io::Cursor;
 use std::path::PathBuf;
 
-use crate::{Apk, IdType, Result};
-use androscalpel_serializer::{
-    DexFileReader, HeaderItem, Serializable, Sleb128, Uleb128, Uleb128p1,
-};
-use apk_frauder::{ZipFileReader, end_of_central_directory::EndCentralDirectory};
+use crate::Result;
+use androscalpel_serializer::{Serializable, Sleb128, Uleb128, Uleb128p1};
 
 /// Convert an integer to the uleb128 byte encoding
 #[pyfunction]
@@ -49,32 +44,6 @@ pub fn sleb128_to_int(b: &[u8]) -> Result<i32> {
     Ok(Sleb128::deserialize_from_slice(b)?.0)
 }
 
-// TODO: list_defined_classes, is_dex, is_zip take only &[u8] or file, but should allow to also read from both
-
-/// List all classes defined in a dex file.
-#[pyfunction]
-pub fn list_defined_classes(dex: &[u8]) -> Result<HashSet<IdType>> {
-    let dex = DexFileReader::new(dex)?;
-    dex.get_class_defs()
-        .iter()
-        .map(|cdef| Apk::get_id_type_from_idx(cdef.class_idx as usize, &dex))
-        .collect()
-}
-
-/// Test if a file is as .dex file an return the dex version if it is, else return None.
-#[pyfunction]
-pub fn is_dex(file: PathBuf) -> Result<Option<usize>> {
-    let mut file = File::open(file)?;
-    crate::utils::is_dex(file)
-}
-
-/// Test if a file is a zip file.
-#[pyfunction]
-pub fn is_zip(file: PathBuf) -> Result<bool> {
-    let mut file = File::open(file)?;
-    crate::utils::is_zip(file)
-}
-
 /// Replace the dex files a an apk an resigned the apk.
 ///
 /// # Warning
@@ -82,69 +51,28 @@ pub fn is_zip(file: PathBuf) -> Result<bool> {
 /// For now, only jks keystore are allowed.
 ///
 /// The `zipalign` and `apksigner` args allow to use a specific version of the
-/// toimpl Read + Seekols instead of the one in the PATH (if it even exist)
-///
-/// `additionnal_files` is a dict of file to add, modify or remove in the apk.
-/// The keys are the file names and the values are `None` to remove the file, or
-/// `bytes` for the content of the file.
+/// tools instead of the one in the PATH (if it even exist)
 #[pyfunction]
-#[pyo3(signature = (
-    apk,
-    dst,
-    dexfiles,
-    keystore,
-    zipalign=None,
-    apksigner=None,
-    additionnal_files=None
-))]
 pub fn replace_dex(
     apk: PathBuf,
     dst: PathBuf,
-    dexfiles: Vec<Bound<'_, PyBytes>>,
+    dexfiles: Vec<&[u8]>,
     keystore: PathBuf,
     zipalign: Option<PathBuf>,
     apksigner: Option<PathBuf>,
-    additionnal_files: Option<HashMap<String, Option<Bound<'_, PyBytes>>>>,
 ) {
-    let mut dexfiles: Vec<_> = dexfiles
-        .iter()
-        .map(PyBytesMethods::as_bytes)
-        .map(Cursor::new)
-        .collect();
-    let additionnal_files: Option<HashMap<_, _>> =
-        additionnal_files.as_ref().map(|additionnal_files| {
-            additionnal_files
-                .iter()
-                .map(|(k, v)| {
-                    (
-                        k.clone(),
-                        v.as_ref().map(|bytes| bytes.as_bytes()).map(Cursor::new),
-                    )
-                })
-                .collect()
-        });
-    apk_frauder::replace_dex(
-        apk,
-        dst,
-        &mut dexfiles,
-        keystore,
-        zipalign,
-        apksigner,
-        additionnal_files,
-    )
+    let mut dexfiles: Vec<_> = dexfiles.iter().map(Cursor::new).collect();
+    apk_frauder::replace_dex(apk, dst, &mut dexfiles, keystore, zipalign, apksigner)
 }
 
 /// export the function in a python module
-pub(crate) fn export_module(_py: Python, m: &Bound<'_, PyModule>) -> PyResult<()> {
+pub(crate) fn export_module(_py: Python, m: &PyModule) -> PyResult<()> {
     m.add_function(wrap_pyfunction!(int_to_uleb128, m)?)?;
     m.add_function(wrap_pyfunction!(int_to_uleb128p1, m)?)?;
     m.add_function(wrap_pyfunction!(int_to_sleb128, m)?)?;
     m.add_function(wrap_pyfunction!(uleb128_to_int, m)?)?;
     m.add_function(wrap_pyfunction!(uleb128p1_to_int, m)?)?;
     m.add_function(wrap_pyfunction!(sleb128_to_int, m)?)?;
-    m.add_function(wrap_pyfunction!(list_defined_classes, m)?)?;
-    m.add_function(wrap_pyfunction!(is_dex, m)?)?;
-    m.add_function(wrap_pyfunction!(is_zip, m)?)?;
     m.add_function(wrap_pyfunction!(replace_dex, m)?)?;
     Ok(())
 }

androscalpel/src/scalar.rs

@@ -4,29 +4,24 @@ use crate::Result;
 use serde::{Deserialize, Serialize};
 use std::collections::HashSet;
 
-use crate::{
-    DexString, DexValue, IdField, IdMethod, IdMethodType, IdType, MethodHandle, Visitable,
-    VisitableMut, Visitor, VisitorMut,
-};
-#[cfg(feature = "python")]
+use crate::{DexString, DexValue, IdField, IdMethod, IdMethodType, IdType, MethodHandle};
 use pyo3::prelude::*;
 
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize, Serialize)]
 pub struct DexByte(pub i8);
-
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl DexByte {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(val: i8) -> Self {
         Self(val)
     }
@@ -44,21 +39,21 @@ impl DexByte {
     }
 }
 
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize, Serialize)]
 pub struct DexShort(pub i16);
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl DexShort {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(val: i16) -> Self {
         Self(val)
     }
@@ -76,21 +71,21 @@ impl DexShort {
     }
 }
 
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize, Serialize)]
 pub struct DexChar(pub u16);
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl DexChar {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(val: u16) -> Self {
         Self(val)
     }
@@ -108,21 +103,21 @@ impl DexChar {
     }
 }
 
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize, Serialize)]
 pub struct DexInt(pub i32);
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl DexInt {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(val: i32) -> Self {
         Self(val)
     }
@@ -140,21 +135,21 @@ impl DexInt {
     }
 }
 
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize, Serialize)]
 pub struct DexLong(pub i64);
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl DexLong {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(val: i64) -> Self {
         Self(val)
     }
@@ -172,21 +167,21 @@ impl DexLong {
     }
 }
 
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, Copy, PartialEq, Deserialize, Serialize)]
 pub struct DexFloat(pub f32);
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl DexFloat {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(val: f32) -> Self {
         Self(val)
     }
@@ -204,21 +199,21 @@ impl DexFloat {
     }
 }
 
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, Copy, PartialEq, Deserialize, Serialize)]
 pub struct DexDouble(pub f64);
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl DexDouble {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(val: f64) -> Self {
         Self(val)
     }
@@ -237,21 +232,21 @@ impl DexDouble {
 }
 
 /* DexString is already define in lib.rs, TODO: move the version in lib.rs here
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 pub struct DexString(pub u32);
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl DexString {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-   #[cfg_attr(feature = "python", staticmethod)]
+   #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(val: u32) -> Self {
         Self(val)
     }
@@ -270,21 +265,21 @@ impl DexString {
 }
 */
 
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize, Serialize)]
 pub struct DexNull;
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl DexNull {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn _new() -> Self {
         Self
     }
@@ -298,21 +293,21 @@ impl DexNull {
     }
 }
 
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize, Serialize)]
 pub struct DexBoolean(pub bool);
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl DexBoolean {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn new(val: bool) -> Self {
         Self(val)
     }
@@ -330,21 +325,21 @@ impl DexBoolean {
     }
 }
 
-#[cfg_attr(feature = "python", pyclass(eq))]
+#[pyclass]
 #[derive(Debug, Clone, PartialEq, Deserialize, Serialize)]
 pub struct DexArray(pub Vec<DexValue>);
-#[cfg_attr(feature = "python", pymethods)]
+#[pymethods]
 impl DexArray {
     pub fn to_json(&self) -> Result<String> {
         Ok(serde_json::to_string(self)?)
     }
 
-    #[cfg_attr(feature = "python", staticmethod)]
+    #[staticmethod]
     pub fn from_json(json: &str) -> Result<Self> {
         Ok(serde_json::from_str(json)?)
     }
 
-    #[cfg_attr(feature = "python", new)]
+    #[new]
     pub fn _new(arr: Vec<DexValue>) -> Self {
         Self(arr)
     }
@@ -422,23 +417,3 @@ impl DexArray {
         methods
     }
 }
-
-impl<V: Visitor> Visitable<V> for DexArray {
-    fn default_visit(&self, v: &mut V) -> Result<()> {
-        for val in &self.0 {
-            v.visit_value(val)?;
-        }
-        Ok(())
-    }
-}
-
-impl<V: VisitorMut> VisitableMut<V> for DexArray {
-    fn default_visit_mut(self, v: &mut V) -> Result<Self> {
-        Ok(Self(
-            self.0
-                .into_iter()
-                .map(|val| v.visit_value(val))
-                .collect::<Result<_>>()?,
-        ))
-    }
-}

androscalpel/src/tests/apktool_parse_hiddenapi.py

@@ -1,103 +0,0 @@
-#!/usr/bin/python3
-
-# require zip and apktool installed
-
-import sys
-import os
-import tempfile
-from pathlib import Path
-import json
-
-
-def parse_smali(path: Path) -> tuple[str, dict, dict]:
-    class_name: str | None = None
-    fields = {}
-    methods = {}
-    with path.open() as file:
-        for line in file.readlines():
-            if line.startswith(".class"):
-                if class_name is None:
-                    class_name = line.strip().split(" ")[-1]
-                else:
-                    raise RuntimeError(f"Two classes found in {path}")
-            if line.startswith(".field"):
-                k, v = parse_field(line)
-                fields[k] = v
-            elif line.startswith(".method"):
-                k, v = parse_method(line)
-                methods[k] = v
-    if class_name is None:
-        raise RuntimeError(f"No classe found in {path}")
-    return (class_name, fields, methods)
-
-
-def parse_field(line: str) -> tuple[str, dict]:
-    data: dict[str, None | str | list[str]] = {
-        "value": None,
-        "other": [],
-        "hiddenapi": None,
-        "hiddenapi_domain": None,
-    }
-    if "=" in line:
-        line, val = line.split("=")[0], "=".join(line.split("=")[1:])
-        data["value"] = val.strip()
-    line, ty = line.split(":")
-    vals = list(map(str.strip, line.split(" ")))
-    name = f"{vals[-1]}:{ty.strip()}"
-    vals = vals[:-1]
-    if vals:
-        if "api" in vals[-1]:
-            data["hiddenapi_domain"] = vals[-1]
-            vals = vals[:-1]
-    if vals:
-        if "list" in vals[-1]:
-            data["hiddenapi"] = vals[-1]
-            vals = vals[:-1]
-    data["other"] = vals
-    return name, data
-
-
-def parse_method(line: str) -> tuple[str, dict]:
-    data: dict[str, None | str | list[str]] = {
-        "other": [],
-        "hiddenapi": None,
-        "hiddenapi_domain": None,
-    }
-    vals = list(map(str.strip, line.split(" ")))
-    name = vals[-1]
-    vals = vals[:-1]
-    if vals:
-        if "api" in vals[-1]:
-            data["hiddenapi_domain"] = vals[-1]
-            vals = vals[:-1]
-    if vals:
-        if "list" in vals[-1]:
-            data["hiddenapi"] = vals[-1]
-            vals = vals[:-1]
-    data["other"] = vals
-    return name, data
-
-
-def main():
-    if len(sys.argv) != 3:
-        print("rought smali parser")
-        print("usage:")
-        print(f"    {sys.argv[0]} some_classes.dex output.json")
-        exit()
-    file = sys.argv[1]
-    dst = sys.argv[2]
-    data = {}
-    with tempfile.TemporaryDirectory() as tmp:
-        os.system(f"zip {tmp}/apk.apk {file}")
-        os.system(f"apktool d -o {tmp}/apktool.out {tmp}/apk.apk")
-        smalidir = Path(f"{tmp}/apktool.out/smali_{file.removesuffix('.dex')}")
-        for root, dirs, files in smalidir.walk():
-            for file in files:
-                class_name, fields, methods = parse_smali(root / file)
-                data[class_name] = {"fields": fields, "methods": methods}
-    with open(dst, "w") as file:
-        json.dump(data, file)
-
-
-if __name__ == "__main__":
-    main()

androscalpel/src/tests/app1.json

@@ -1,676 +0,0 @@
-{
-  "dex_files": {
-    "classes.dex": {
-      "classes": {
-        "Lcom/example/testclassloader/TestA;": {
-          "descriptor": "Lcom/example/testclassloader/TestA;",
-          "is_public": true,
-          "is_final": false,
-          "is_interface": false,
-          "is_abstract": false,
-          "is_synthetic": false,
-          "is_annotation": false,
-          "is_enum": false,
-          "superclass": "Ljava/lang/Object;",
-          "interfaces": [],
-          "source_file": {
-            "String": "TestA.java"
-          },
-          "static_fields": {
-            "Lcom/example/testclassloader/TestA;->value:Ljava/lang/String;": {
-              "descriptor": "Lcom/example/testclassloader/TestA;->value:Ljava/lang/String;",
-              "visibility": "Public",
-              "is_static": true,
-              "is_final": false,
-              "is_volatile": false,
-              "is_transient": false,
-              "is_synthetic": false,
-              "is_enum": false,
-              "value": null,
-              "annotations": []
-            }
-          },
-          "instance_fields": {},
-          "direct_methods": {
-            "Lcom/example/testclassloader/TestA;-><init>()V": {
-              "descriptor": "Lcom/example/testclassloader/TestA;-><init>()V",
-              "visibility": "Public",
-              "is_static": false,
-              "is_final": false,
-              "is_synchronized": false,
-              "is_bridge": false,
-              "is_varargs": false,
-              "is_native": false,
-              "is_abstract": false,
-              "is_strictfp": false,
-              "is_synthetic": false,
-              "is_constructor": true,
-              "is_declared_syncrhonized": false,
-              "annotations": [],
-              "parameters_annotations": [],
-              "code": {
-                "registers_size": 1,
-                "ins_size": 1,
-                "outs_size": 1,
-                "debug_info": [
-                  3,
-                  [
-                    14,
-                    0
-                  ]
-                ],
-                "parameter_names": [],
-                "insns": [
-                  {
-                    "Label": {
-                      "name": "label_00000000"
-                    }
-                  },
-                  {
-                    "InvokeDirect": {
-                      "method": "Ljava/lang/Object;-><init>()V",
-                      "args": [
-                        0
-                      ]
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000003"
-                    }
-                  },
-                  {
-                    "ReturnVoid": {}
-                  }
-                ]
-              }
-            },
-            "Lcom/example/testclassloader/TestA;-><clinit>()V": {
-              "descriptor": "Lcom/example/testclassloader/TestA;-><clinit>()V",
-              "visibility": "None_",
-              "is_static": true,
-              "is_final": false,
-              "is_synchronized": false,
-              "is_bridge": false,
-              "is_varargs": false,
-              "is_native": false,
-              "is_abstract": false,
-              "is_strictfp": false,
-              "is_synthetic": false,
-              "is_constructor": true,
-              "is_declared_syncrhonized": false,
-              "annotations": [],
-              "parameters_annotations": [],
-              "code": {
-                "registers_size": 1,
-                "ins_size": 0,
-                "outs_size": 0,
-                "debug_info": [
-                  5,
-                  [
-                    14,
-                    0
-                  ]
-                ],
-                "parameter_names": [],
-                "insns": [
-                  {
-                    "Label": {
-                      "name": "label_00000000"
-                    }
-                  },
-                  {
-                    "ConstString": {
-                      "reg": 0,
-                      "lit": {
-                        "String": "Pirat\\')"
-                      }
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000002"
-                    }
-                  },
-                  {
-                    "SPutObject": {
-                      "from": 0,
-                      "field": "Lcom/example/testclassloader/TestA;->value:Ljava/lang/String;"
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000004"
-                    }
-                  },
-                  {
-                    "ReturnVoid": {}
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000005"
-                    }
-                  },
-                  {
-                    "Nop": {}
-                  }
-                ]
-              }
-            }
-          },
-          "virtual_methods": {
-            "Lcom/example/testclassloader/TestA;->get_value()Ljava/lang/String;": {
-              "descriptor": "Lcom/example/testclassloader/TestA;->get_value()Ljava/lang/String;",
-              "visibility": "Public",
-              "is_static": false,
-              "is_final": false,
-              "is_synchronized": false,
-              "is_bridge": false,
-              "is_varargs": false,
-              "is_native": false,
-              "is_abstract": false,
-              "is_strictfp": false,
-              "is_synthetic": false,
-              "is_constructor": false,
-              "is_declared_syncrhonized": false,
-              "annotations": [],
-              "parameters_annotations": [],
-              "code": {
-                "registers_size": 2,
-                "ins_size": 1,
-                "outs_size": 0,
-                "debug_info": [
-                  11,
-                  [
-                    14,
-                    0
-                  ]
-                ],
-                "parameter_names": [],
-                "insns": [
-                  {
-                    "Label": {
-                      "name": "label_00000000"
-                    }
-                  },
-                  {
-                    "SGetObject": {
-                      "to": 0,
-                      "field": "Lcom/example/testclassloader/TestA;->value:Ljava/lang/String;"
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000002"
-                    }
-                  },
-                  {
-                    "ReturnObject": {
-                      "reg": 0
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000003"
-                    }
-                  },
-                  {
-                    "Nop": {}
-                  }
-                ]
-              }
-            },
-            "Lcom/example/testclassloader/TestA;->get_cl()Ljava/lang/String;": {
-              "descriptor": "Lcom/example/testclassloader/TestA;->get_cl()Ljava/lang/String;",
-              "visibility": "Public",
-              "is_static": false,
-              "is_final": false,
-              "is_synchronized": false,
-              "is_bridge": false,
-              "is_varargs": false,
-              "is_native": false,
-              "is_abstract": false,
-              "is_strictfp": false,
-              "is_synthetic": false,
-              "is_constructor": false,
-              "is_declared_syncrhonized": false,
-              "annotations": [],
-              "parameters_annotations": [],
-              "code": {
-                "registers_size": 2,
-                "ins_size": 1,
-                "outs_size": 1,
-                "debug_info": [
-                  7,
-                  [
-                    14,
-                    0
-                  ]
-                ],
-                "parameter_names": [],
-                "insns": [
-                  {
-                    "Label": {
-                      "name": "label_00000000"
-                    }
-                  },
-                  {
-                    "InvokeVirtual": {
-                      "method": "Ljava/lang/Object;->getClass()Ljava/lang/Class;",
-                      "args": [
-                        1
-                      ]
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000003"
-                    }
-                  },
-                  {
-                    "MoveResultObject": {
-                      "to": 0
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000004"
-                    }
-                  },
-                  {
-                    "InvokeVirtual": {
-                      "method": "Ljava/lang/Class;->getClassLoader()Ljava/lang/ClassLoader;",
-                      "args": [
-                        0
-                      ]
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000007"
-                    }
-                  },
-                  {
-                    "MoveResultObject": {
-                      "to": 0
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000008"
-                    }
-                  },
-                  {
-                    "InvokeVirtual": {
-                      "method": "Ljava/lang/Object;->toString()Ljava/lang/String;",
-                      "args": [
-                        0
-                      ]
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_0000000B"
-                    }
-                  },
-                  {
-                    "MoveResultObject": {
-                      "to": 0
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_0000000C"
-                    }
-                  },
-                  {
-                    "ReturnObject": {
-                      "reg": 0
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_0000000D"
-                    }
-                  },
-                  {
-                    "Nop": {}
-                  }
-                ]
-              }
-            }
-          },
-          "annotations": []
-        },
-        "Lcom/example/testclassloader/TestB;": {
-          "descriptor": "Lcom/example/testclassloader/TestB;",
-          "is_public": true,
-          "is_final": false,
-          "is_interface": false,
-          "is_abstract": false,
-          "is_synthetic": false,
-          "is_annotation": false,
-          "is_enum": false,
-          "superclass": "Ljava/lang/Object;",
-          "interfaces": [],
-          "source_file": {
-            "String": "TestB.java"
-          },
-          "static_fields": {},
-          "instance_fields": {},
-          "direct_methods": {
-            "Lcom/example/testclassloader/TestB;-><init>()V": {
-              "descriptor": "Lcom/example/testclassloader/TestB;-><init>()V",
-              "visibility": "Public",
-              "is_static": false,
-              "is_final": false,
-              "is_synchronized": false,
-              "is_bridge": false,
-              "is_varargs": false,
-              "is_native": false,
-              "is_abstract": false,
-              "is_strictfp": false,
-              "is_synthetic": false,
-              "is_constructor": true,
-              "is_declared_syncrhonized": false,
-              "annotations": [],
-              "parameters_annotations": [],
-              "code": {
-                "registers_size": 1,
-                "ins_size": 1,
-                "outs_size": 1,
-                "debug_info": [
-                  3,
-                  [
-                    14,
-                    0
-                  ]
-                ],
-                "parameter_names": [],
-                "insns": [
-                  {
-                    "Label": {
-                      "name": "label_00000000"
-                    }
-                  },
-                  {
-                    "InvokeDirect": {
-                      "method": "Ljava/lang/Object;-><init>()V",
-                      "args": [
-                        0
-                      ]
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000003"
-                    }
-                  },
-                  {
-                    "ReturnVoid": {}
-                  }
-                ]
-              }
-            },
-            "Lcom/example/testclassloader/TestB;-><clinit>()V": {
-              "descriptor": "Lcom/example/testclassloader/TestB;-><clinit>()V",
-              "visibility": "None_",
-              "is_static": true,
-              "is_final": false,
-              "is_synchronized": false,
-              "is_bridge": false,
-              "is_varargs": false,
-              "is_native": false,
-              "is_abstract": false,
-              "is_strictfp": false,
-              "is_synthetic": false,
-              "is_constructor": true,
-              "is_declared_syncrhonized": false,
-              "annotations": [],
-              "parameters_annotations": [],
-              "code": {
-                "registers_size": 1,
-                "ins_size": 0,
-                "outs_size": 0,
-                "debug_info": [
-                  5,
-                  [
-                    14,
-                    0
-                  ]
-                ],
-                "parameter_names": [],
-                "insns": [
-                  {
-                    "Label": {
-                      "name": "label_00000000"
-                    }
-                  },
-                  {
-                    "ConstString": {
-                      "reg": 0,
-                      "lit": {
-                        "String": "Plopliplop"
-                      }
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000002"
-                    }
-                  },
-                  {
-                    "SPutObject": {
-                      "from": 0,
-                      "field": "Lcom/example/testclassloader/TestB;->value:Ljava/lang/String;"
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000004"
-                    }
-                  },
-                  {
-                    "ReturnVoid": {}
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000005"
-                    }
-                  },
-                  {
-                    "Nop": {}
-                  }
-                ]
-              }
-            }
-          },
-          "virtual_methods": {
-            "Lcom/example/testclassloader/TestB;->get_cl()Ljava/lang/String;": {
-              "descriptor": "Lcom/example/testclassloader/TestB;->get_cl()Ljava/lang/String;",
-              "visibility": "Public",
-              "is_static": false,
-              "is_final": false,
-              "is_synchronized": false,
-              "is_bridge": false,
-              "is_varargs": false,
-              "is_native": false,
-              "is_abstract": false,
-              "is_strictfp": false,
-              "is_synthetic": false,
-              "is_constructor": false,
-              "is_declared_syncrhonized": false,
-              "annotations": [],
-              "parameters_annotations": [],
-              "code": {
-                "registers_size": 2,
-                "ins_size": 1,
-                "outs_size": 1,
-                "debug_info": [
-                  7,
-                  [
-                    14,
-                    0
-                  ]
-                ],
-                "parameter_names": [],
-                "insns": [
-                  {
-                    "Label": {
-                      "name": "label_00000000"
-                    }
-                  },
-                  {
-                    "InvokeVirtual": {
-                      "method": "Ljava/lang/Object;->getClass()Ljava/lang/Class;",
-                      "args": [
-                        1
-                      ]
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000003"
-                    }
-                  },
-                  {
-                    "MoveResultObject": {
-                      "to": 0
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000004"
-                    }
-                  },
-                  {
-                    "InvokeVirtual": {
-                      "method": "Ljava/lang/Class;->getClassLoader()Ljava/lang/ClassLoader;",
-                      "args": [
-                        0
-                      ]
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000007"
-                    }
-                  },
-                  {
-                    "MoveResultObject": {
-                      "to": 0
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000008"
-                    }
-                  },
-                  {
-                    "InvokeVirtual": {
-                      "method": "Ljava/lang/Object;->toString()Ljava/lang/String;",
-                      "args": [
-                        0
-                      ]
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_0000000B"
-                    }
-                  },
-                  {
-                    "MoveResultObject": {
-                      "to": 0
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_0000000C"
-                    }
-                  },
-                  {
-                    "ReturnObject": {
-                      "reg": 0
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_0000000D"
-                    }
-                  },
-                  {
-                    "Nop": {}
-                  }
-                ]
-              }
-            },
-            "Lcom/example/testclassloader/TestB;->get_value()Ljava/lang/String;": {
-              "descriptor": "Lcom/example/testclassloader/TestB;->get_value()Ljava/lang/String;",
-              "visibility": "Public",
-              "is_static": false,
-              "is_final": false,
-              "is_synchronized": false,
-              "is_bridge": false,
-              "is_varargs": false,
-              "is_native": false,
-              "is_abstract": false,
-              "is_strictfp": false,
-              "is_synthetic": false,
-              "is_constructor": false,
-              "is_declared_syncrhonized": false,
-              "annotations": [],
-              "parameters_annotations": [],
-              "code": {
-                "registers_size": 2,
-                "ins_size": 1,
-                "outs_size": 0,
-                "debug_info": [
-                  11,
-                  [
-                    14,
-                    0
-                  ]
-                ],
-                "parameter_names": [],
-                "insns": [
-                  {
-                    "Label": {
-                      "name": "label_00000000"
-                    }
-                  },
-                  {
-                    "SGetObject": {
-                      "to": 0,
-                      "field": "Lcom/example/testclassloader/TestA;->value:Ljava/lang/String;"
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000002"
-                    }
-                  },
-                  {
-                    "ReturnObject": {
-                      "reg": 0
-                    }
-                  },
-                  {
-                    "Label": {
-                      "name": "label_00000003"
-                    }
-                  },
-                  {
-                    "Nop": {}
-                  }
-                ]
-              }
-            }
-          },
-          "annotations": []
-        }
-      },
-      "not_referenced_strings": []
-    }
-  }
-}

androscalpel/src/tests/mod.rs

@@ -1,11 +1,9 @@
 use super::*;
 use androscalpel_serializer::Instruction as InsFormat;
 use androscalpel_serializer::*;
-use serde_json as sj;
-use std::collections::HashSet;
 use std::fs::File;
 use std::io;
-use std::io::{Read, Write};
+use std::io::Write;
 use std::ops::Deref;
 use std::sync::{Mutex, OnceLock};
 use std::time::Instant;
@@ -13,14 +11,13 @@ use std::time::Instant;
 fn write_to_report(data: &str) {
     static REPORT_FILE: Mutex<Option<File>> = Mutex::new(None);
     let mut report_file = REPORT_FILE.lock().unwrap();
-    let mut report_file = match report_file.deref() {
-        Some(report_file) => report_file,
-        _ => {
-            *report_file = Some(
-                File::create(&format!("{}/test_repport.txt", env!("CARGO_MANIFEST_DIR"),)).unwrap(),
-            );
-            report_file.deref().as_ref().unwrap()
-        }
+    let mut report_file = if let Some(report_file) = report_file.deref() {
+        report_file
+    } else {
+        *report_file = Some(
+            File::create(&format!("{}/test_repport.txt", env!("CARGO_MANIFEST_DIR"),)).unwrap(),
+        );
+        report_file.deref().as_ref().unwrap()
     };
     writeln!(report_file, "{data}").unwrap();
 }
@@ -43,8 +40,7 @@ fn get_hello_world_apk() -> &'static Apk {
     HELLO_WORLD_APK.get_or_init(|| {
         let mut apk = Apk::new();
         let start = Instant::now();
-        apk.add_dex_file("classes.dex", get_hello_world_dex(), |_, _, _| None, false)
-            .unwrap();
+        apk.add_dex_file(get_hello_world_dex()).unwrap();
         let duration = start.elapsed();
         write_to_report(&format!("Parsing classes_hello_world.dex: {duration:?}"));
         apk
@@ -55,11 +51,7 @@ fn get_hello_world_recompilled() -> &'static [u8] {
     static HELLO_WORLD_RECOMP: OnceLock<Vec<u8>> = OnceLock::new();
     HELLO_WORLD_RECOMP.get_or_init(|| {
         let start = Instant::now();
-        let dex = get_hello_world_apk()
-            .gen_raw_dex()
-            .unwrap()
-            .remove("classes.dex")
-            .unwrap();
+        let dex = get_hello_world_apk().gen_raw_dex().unwrap().pop().unwrap();
         let duration = start.elapsed();
         write_to_report(&format!("Recompile classes_hello_world.dex: {duration:?}"));
         dex
@@ -85,7 +77,7 @@ fn get_class_dex<'a, 'b>(name: &'a str, dex: &'b DexFileReader) -> Option<&'b Cl
 
 fn get_method_code_dex(name: &str, dex: &DexFileReader) -> Option<CodeItem> {
     let method = IdMethod::from_smali(name).unwrap();
-    let class_name: String = (&method.class_.0).try_into().unwrap();
+    let class_name: String = (&method.class_.0).into();
     let class = get_class_dex(&class_name, dex);
     let class = if let Some(class) = class {
         class
@@ -142,30 +134,7 @@ fn test_generated_data_size() {
 fn test_generated_apk_equivalence() {
     let new_dex = get_hello_world_recompilled();
     let mut new_apk = Apk::new();
-    new_apk
-        .add_dex_file("classes.dex", &new_dex, |_, _, _| None, false)
-        .unwrap();
-
-    /*
-    use pretty_assertions::assert_eq;
-    let method = IdMethod::from_smali(
-        "Lcom/google/android/material/datepicker/DateFormatTextWatcher;->lambda$new$0$com-google-android-material-datepicker-DateFormatTextWatcher(Ljava/lang/String;)V"
-    )
-    .unwrap();
-    assert_eq!(
-        get_hello_world_apk()
-            .get_class(&method.class_)
-            .unwrap()
-            .virtual_methods
-            .get(&method)
-            .unwrap(),
-        new_apk
-            .get_class(&method.class_)
-            .unwrap()
-            .virtual_methods
-            .get(&method)
-            .unwrap()
-    );*/
+    new_apk.add_dex_file(&new_dex).unwrap();
     assert_eq!(get_hello_world_apk(), &new_apk);
 }
 
@@ -383,344 +352,3 @@ fn test_sort_strings() {
     assert_eq!(list1, list2);
     assert_eq!(list1, vec![s1.clone(), s2.clone()]);
 }
-
-/// Emulate test https://cs.android.com/android/platform/superproject/main/+/main:art/libdexfile/dex/dex_file_verifier.cc;drc=e8da7cd1d0e7d3535c82f8d05adcef3edd43cd40;l=581
-fn check_valid_offset_and_size(
-    dex: &DexFileReader,
-    offset: u32,
-    size: u32,
-    alignment: u32,
-    label: &str,
-) {
-    if size == 0 {
-        if offset != 0 {
-            panic!("Offset 0x{offset:x} should be zero when size is zero for {label}");
-        }
-        return;
-    }
-    // offset < hdr_offset is not relevent (we index from hdr_offset=0)
-    let file_size = dex.get_header().file_size;
-    if file_size <= offset {
-        panic!("Offset 0x{offset:x} sould be within file size 0x{file_size:x} for {label}");
-    }
-    if file_size - offset < size {
-        // size + offset could overflow
-        panic!(
-            "Section end 0x{:x} should be within file size 0x{file_size:x} for {label}",
-            size + offset
-        );
-    }
-    if alignment != 0 && !(offset & (alignment - 1) == 0) {
-        panic!("Offset 0x{offset:x} sould be aligned by {alignment} for {label}");
-    }
-}
-
-#[test]
-fn test_1_from_json() {
-    let filename = "test_class1.json";
-    let hello_world_dex = format!("{}/src/tests/{}", env!("CARGO_MANIFEST_DIR"), filename);
-    let mut file = File::open(&hello_world_dex).expect(&format!("{} not found", filename));
-    let mut json = String::new();
-    file.read_to_string(&mut json).unwrap();
-    let test_a: Class = serde_json::from_str(&json).unwrap();
-    let mut apk = Apk::new();
-    apk.add_class("classes.dex", test_a).unwrap();
-    let dex = apk.gen_raw_dex().unwrap().remove("classes.dex").unwrap();
-    let dex = DexFileReader::new(&dex).unwrap();
-
-    check_valid_offset_and_size(
-        &dex,
-        dex.get_header().link_off,
-        dex.get_header().link_size,
-        0,
-        "link",
-    );
-    check_valid_offset_and_size(&dex, dex.get_header().map_off, 4, 4, "map");
-    check_valid_offset_and_size(
-        &dex,
-        dex.get_header().string_ids_off,
-        dex.get_header().string_ids_size,
-        4,
-        "string-ids",
-    );
-    check_valid_offset_and_size(
-        &dex,
-        dex.get_header().type_ids_off,
-        dex.get_header().type_ids_size,
-        4,
-        "type-ids",
-    );
-    check_valid_offset_and_size(
-        &dex,
-        dex.get_header().proto_ids_off,
-        dex.get_header().proto_ids_size,
-        4,
-        "proto-ids",
-    );
-    check_valid_offset_and_size(
-        &dex,
-        dex.get_header().field_ids_off,
-        dex.get_header().field_ids_size,
-        4,
-        "field-ids",
-    );
-    check_valid_offset_and_size(
-        &dex,
-        dex.get_header().method_ids_off,
-        dex.get_header().method_ids_size,
-        4,
-        "method-ids",
-    );
-    check_valid_offset_and_size(
-        &dex,
-        dex.get_header().class_defs_off,
-        dex.get_header().class_defs_size,
-        4,
-        "class-defs",
-    );
-    check_valid_offset_and_size(
-        &dex,
-        dex.get_header().data_off,
-        dex.get_header().data_size,
-        0,
-        "data",
-    );
-    if dex.get_header().type_ids_size > (u16::MAX as u32) {
-        panic!(
-            "Size 0x{:x} should not exceed limit 0x{:x} for type-ids",
-            dex.get_header().type_ids_size,
-            u16::MAX
-        )
-    }
-    if dex.get_header().proto_ids_size > (u16::MAX as u32) {
-        panic!(
-            "Size 0x{:x} should not exceed limit 0x{:x} for proto-ids",
-            dex.get_header().proto_ids_size,
-            u16::MAX
-        )
-    }
-
-    let map = dex.get_map_list();
-    let mut last_offset = 0;
-    let mut last_type = MapItemType::UnkownType(0);
-    let mut data_item_count = 0;
-    let mut data_items_left = dex.get_header().data_size;
-    let file_size = dex.get_header().file_size;
-    let mut used_types = HashSet::new();
-    for (i, item) in map.list.iter().enumerate() {
-        if last_offset >= item.offset && i != 0 {
-            panic!(
-                "Out of order map item: 0x{:x} then 0x{:x} for type {:?} (last type was {:?})",
-                last_offset, item.offset, item.type_, last_type
-            )
-        }
-        if item.offset >= file_size {
-            panic!(
-                "Map item for type {:?} ends after file: 0x{:x} >= 0x{:x}",
-                item.type_, item.offset, file_size
-            );
-        }
-        if item.type_.is_data_section_type() {
-            if item.size > data_items_left {
-                panic!(
-                    "Too many items in data section: {} > {} (item type: {:?}",
-                    item.size + data_item_count,
-                    dex.get_header().data_size,
-                    item.type_
-                );
-            }
-            data_items_left -= item.size;
-            data_item_count += item.size;
-        }
-        if used_types.contains(&item.type_) {
-            panic!("Duplicate map section of type {:?}", item.type_);
-        }
-        used_types.insert(item.type_);
-        last_offset = item.offset;
-        last_type = item.type_;
-    }
-    if !used_types.contains(&MapItemType::HeaderItem) {
-        panic!("Map is missing Header entry");
-    }
-    if !used_types.contains(&MapItemType::MapList) {
-        panic!("Map is missing Map List entry");
-    }
-    if !used_types.contains(&MapItemType::StringIdItem) && (dex.get_header().string_ids_off != 0) {
-        panic!("Map is missing String Id entry");
-    }
-    if !used_types.contains(&MapItemType::TypeIdItem) && (dex.get_header().type_ids_off != 0) {
-        panic!("Map is missing Type Id entry");
-    }
-    if !used_types.contains(&MapItemType::ProtoIdItem) && (dex.get_header().proto_ids_off != 0) {
-        panic!("Map is missing Proto Id entry");
-    }
-    if !used_types.contains(&MapItemType::FieldIdItem) && (dex.get_header().field_ids_off != 0) {
-        panic!("Map is missing Field Id entry");
-    }
-    if !used_types.contains(&MapItemType::MethodIdItem) && (dex.get_header().method_ids_off != 0) {
-        panic!("Map is missing Method Id entry");
-    }
-    if !used_types.contains(&MapItemType::ClassDefItem) && (dex.get_header().class_defs_off != 0) {
-        panic!("Map is missing Class Def entry");
-    }
-}
-
-#[test]
-fn test_2_from_json() {
-    let filename = "app1.json";
-    let json_path = format!("{}/src/tests/{}", env!("CARGO_MANIFEST_DIR"), filename);
-    let mut file = File::open(&json_path).expect(&format!("{} not found", filename));
-    let mut json = String::new();
-    file.read_to_string(&mut json).unwrap();
-    let apk: Apk = serde_json::from_str(&json).unwrap();
-    let dex = apk.gen_raw_dex().unwrap().remove("classes.dex").unwrap();
-    let mut new_apk = Apk::new();
-    new_apk
-        .add_dex_file("classes.dex", &dex, |_, _, _| None, false)
-        .unwrap();
-    assert_eq!(apk, new_apk);
-}
-
-#[test]
-fn test_hidden_api() {
-    let dex_raw = get_dex("core-oj-33_classes.dex");
-    // The data are generated using src/tests/apktool_parse_hiddenapi.py src/tests/core-oj-33_classes.dex src/tests/core-oj-33_hiddenapi.json
-    let apktool_result = format!(
-        "{}/src/tests/{}",
-        env!("CARGO_MANIFEST_DIR"),
-        "core-oj-33_hiddenapi.json"
-    );
-    fn f_desc_to_pointer(f_dsc: &IdField) -> String {
-        let name: String = String::try_from(&f_dsc.name.0).unwrap().replace("/", "~1");
-        let ty: String = f_dsc.type_.try_to_smali().unwrap().replace("/", "~1");
-        format!(
-            "/{}/fields/{name}:{ty}",
-            f_dsc.class_.try_to_smali().unwrap().replace("/", "~1")
-        )
-    }
-    fn m_desc_to_pointer(m_dsc: &IdMethod) -> String {
-        let name: String = String::try_from(&m_dsc.name.0).unwrap().replace("/", "~1");
-        let ty: String = m_dsc.proto.try_to_smali().unwrap().replace("/", "~1");
-        format!(
-            "/{}/methods/{name}{ty}",
-            m_dsc.class_.try_to_smali().unwrap().replace("/", "~1")
-        )
-    }
-    fn compare_hidden_api(
-        apktool_hiddenapi: &sj::Value,
-        apktool_hiddenapi_domain: &sj::Value,
-        parsed_api: &Option<HiddenApiData>,
-        name: &str,
-    ) {
-        match (apktool_hiddenapi, apktool_hiddenapi_domain, parsed_api) {
-            (sj::Value::Null, sj::Value::Null, None) => (),
-            (
-                sj::Value::String(apktool_hiddenapi),
-                sj::Value::String(apktool_domain),
-                Some(HiddenApiData { permission, domain }),
-            ) if (permission.to_smali_name() == apktool_hiddenapi)
-                && (&domain.to_smali_name() == apktool_domain) =>
-            {
-                ()
-            }
-            (
-                sj::Value::String(apktool_hiddenapi),
-                sj::Value::Null,
-                Some(HiddenApiData { permission, domain }),
-            ) if (permission.to_smali_name() == apktool_hiddenapi)
-                && &domain.to_smali_name() == "" =>
-            {
-                ()
-            }
-
-            _ => panic!(
-                "Expected {apktool_hiddenapi:?} and {apktool_hiddenapi_domain:?}, found {parsed_api:?} in {name}"
-            ),
-        }
-    }
-    let apktool_result = File::open(&apktool_result).expect("core-oj-33_hiddenapi.json not found");
-    let apktool_result = std::io::BufReader::new(apktool_result);
-    let apktool_result: sj::Value = sj::from_reader(apktool_result).unwrap();
-    let mut apk = Apk::new();
-    apk.add_dex_file("classes.dex", &dex_raw, |_, _, _| None, false)
-        .unwrap();
-    for cls in apktool_result.as_object().unwrap().keys() {
-        assert!(
-            apk.dex_files
-                .get("classes.dex")
-                .unwrap()
-                .classes
-                .get(&dex_id::IdType(cls.as_str().into()))
-                .is_some(),
-            "{cls} not found in core-oj-33_classes.dex"
-        );
-    }
-    for cls in apk.dex_files.get("classes.dex").unwrap().classes.keys() {
-        assert!(
-            apktool_result
-                .get::<String>((&cls.0).try_into().unwrap())
-                .is_some(),
-            "{} not found in core-oj-33_hiddenapi.json",
-            cls.__str__()
-        );
-    }
-    for (_, cls) in &apk.dex_files.get("classes.dex").unwrap().classes {
-        for (f_dsc, field) in &cls.static_fields {
-            let pointer = f_desc_to_pointer(f_dsc);
-            let apktool_field = apktool_result.pointer(&pointer).unwrap();
-            let apktool_hiddenapi = apktool_field.get("hiddenapi").unwrap();
-            let apktool_hiddenapi_domain = apktool_field.get("hiddenapi_domain").unwrap();
-            compare_hidden_api(
-                apktool_hiddenapi,
-                &apktool_hiddenapi_domain,
-                &field.hiddenapi,
-                &f_dsc.try_to_smali().unwrap(),
-            );
-        }
-
-        for (f_dsc, field) in &cls.instance_fields {
-            let pointer = f_desc_to_pointer(f_dsc);
-            let apktool_field = apktool_result.pointer(&pointer).unwrap();
-            let apktool_hiddenapi = apktool_field.get("hiddenapi").unwrap();
-            let apktool_hiddenapi_domain = apktool_field.get("hiddenapi_domain").unwrap();
-            compare_hidden_api(
-                apktool_hiddenapi,
-                &apktool_hiddenapi_domain,
-                &field.hiddenapi,
-                &f_dsc.try_to_smali().unwrap(),
-            );
-        }
-        for (m_dsc, method) in &cls.direct_methods {
-            let pointer = m_desc_to_pointer(m_dsc);
-            let apktool_method = apktool_result.pointer(&pointer).unwrap();
-            let apktool_hiddenapi = apktool_method.get("hiddenapi").unwrap();
-            let apktool_hiddenapi_domain = apktool_method.get("hiddenapi_domain").unwrap();
-            compare_hidden_api(
-                apktool_hiddenapi,
-                &apktool_hiddenapi_domain,
-                &method.hiddenapi,
-                &m_dsc.try_to_smali().unwrap(),
-            );
-        }
-        for (m_dsc, method) in &cls.virtual_methods {
-            let pointer = m_desc_to_pointer(m_dsc);
-            let apktool_method = apktool_result.pointer(&pointer).unwrap();
-            let apktool_hiddenapi = apktool_method.get("hiddenapi").unwrap();
-            let apktool_hiddenapi_domain = apktool_method.get("hiddenapi_domain").unwrap();
-            compare_hidden_api(
-                apktool_hiddenapi,
-                &apktool_hiddenapi_domain,
-                &method.hiddenapi,
-                &m_dsc.try_to_smali().unwrap(),
-            );
-        }
-    }
-}
-
-#[test]
-fn test_is_platform_class() {
-    let activity = Class::new("Landroid/app/Activity;".into()).unwrap();
-    assert!(activity.is_platform_class());
-    let not_platform = Class::new("Landroid/app/NotAPlatformClass;".into()).unwrap();
-    assert!(!not_platform.is_platform_class());
-}

androscalpel/src/tests/test_class1.json

@@ -1,194 +0,0 @@
-{
-  "descriptor": "Lcom/example/testclassloader/TestB;",
-  "is_public": true,
-  "is_final": false,
-  "is_interface": false,
-  "is_abstract": false,
-  "is_synthetic": false,
-  "is_annotation": false,
-  "is_enum": false,
-  "superclass": "Ljava/lang/Object;",
-  "interfaces": [],
-  "source_file": {
-    "String": "TestB.java"
-  },
-  "static_fields": {},
-  "instance_fields": {},
-  "direct_methods": {
-    "Lcom/example/testclassloader/TestB;-><init>()V": {
-      "descriptor": "Lcom/example/testclassloader/TestB;-><init>()V",
-      "visibility": "Public",
-      "is_static": false,
-      "is_final": false,
-      "is_synchronized": false,
-      "is_bridge": false,
-      "is_varargs": false,
-      "is_native": false,
-      "is_abstract": false,
-      "is_strictfp": false,
-      "is_synthetic": false,
-      "is_constructor": true,
-      "is_declared_syncrhonized": false,
-      "annotations": [],
-      "parameters_annotations": [],
-      "code": {
-        "registers_size": 1,
-        "ins_size": 1,
-        "outs_size": 1,
-        "debug_info": [
-          3,
-          [
-            14,
-            0
-          ]
-        ],
-        "parameter_names": [],
-        "insns": [
-          {
-            "Label": {
-              "name": "label_00000000"
-            }
-          },
-          {
-            "InvokeDirect": {
-              "method": "Ljava/lang/Object;-><init>()V",
-              "args": [
-                0
-              ]
-            }
-          },
-          {
-            "Label": {
-              "name": "label_00000003"
-            }
-          },
-          {
-            "ReturnVoid": {}
-          }
-        ]
-      }
-    }
-  },
-  "virtual_methods": {
-    "Lcom/example/testclassloader/TestB;->val()Ljava/lang/String;": {
-      "descriptor": "Lcom/example/testclassloader/TestB;->val()Ljava/lang/String;",
-      "visibility": "Public",
-      "is_static": false,
-      "is_final": false,
-      "is_synchronized": false,
-      "is_bridge": false,
-      "is_varargs": false,
-      "is_native": false,
-      "is_abstract": false,
-      "is_strictfp": false,
-      "is_synthetic": false,
-      "is_constructor": false,
-      "is_declared_syncrhonized": false,
-      "annotations": [],
-      "parameters_annotations": [],
-      "code": {
-        "registers_size": 2,
-        "ins_size": 1,
-        "outs_size": 1,
-        "debug_info": [
-          6,
-          [
-            14,
-            0
-          ]
-        ],
-        "parameter_names": [],
-        "insns": [
-          {
-            "Label": {
-              "name": "label_00000000"
-            }
-          },
-          {
-            "InvokeVirtual": {
-              "method": "Ljava/lang/Object;->getClass()Ljava/lang/Class;",
-              "args": [
-                1
-              ]
-            }
-          },
-          {
-            "Label": {
-              "name": "label_00000003"
-            }
-          },
-          {
-            "MoveResultObject": {
-              "to": 0
-            }
-          },
-          {
-            "Label": {
-              "name": "label_00000004"
-            }
-          },
-          {
-            "InvokeVirtual": {
-              "method": "Ljava/lang/Class;->getClassLoader()Ljava/lang/ClassLoader;",
-              "args": [
-                0
-              ]
-            }
-          },
-          {
-            "Label": {
-              "name": "label_00000007"
-            }
-          },
-          {
-            "MoveResultObject": {
-              "to": 0
-            }
-          },
-          {
-            "Label": {
-              "name": "label_00000008"
-            }
-          },
-          {
-            "InvokeVirtual": {
-              "method": "Ljava/lang/Object;->toString()Ljava/lang/String;",
-              "args": [
-                0
-              ]
-            }
-          },
-          {
-            "Label": {
-              "name": "label_0000000B"
-            }
-          },
-          {
-            "MoveResultObject": {
-              "to": 0
-            }
-          },
-          {
-            "Label": {
-              "name": "label_0000000C"
-            }
-          },
-          {
-            "ReturnObject": {
-              "reg": 0
-            }
-          },
-          {
-            "Label": {
-              "name": "label_0000000D"
-            }
-          },
-          {
-            "Nop": {}
-          }
-        ]
-      }
-    }
-  },
-  "annotations": []
-}

androscalpel/src/utils.rs

@@ -1,34 +0,0 @@
-use std::io::{Read, Seek, SeekFrom};
-
-use crate::Result;
-use androscalpel_serializer::{HeaderItem, Serializable};
-use apk_frauder::{ZipFileReader, end_of_central_directory::EndCentralDirectory};
-
-/// Test if a file is as .dex file an return the dex version if it is, else return None.
-pub fn is_dex(file: &mut (impl Read + Seek)) -> Result<Option<usize>> {
-    let pos = file.stream_position()?;
-    let r = HeaderItem::deserialize(file)
-        .ok()
-        .and_then(|header| String::from_utf8(header.magic.version.to_vec()).ok())
-        .and_then(|version| version.parse::<usize>().ok());
-    file.seek(SeekFrom::Start(pos))?;
-    Ok(r)
-}
-
-/// Test if a file is a zip file.
-pub fn is_zip(mut file: impl Read + Seek) -> Result<bool> {
-    let pos = file.stream_position()?;
-    let ecd_off = match ZipFileReader::get_end_of_central_directory_offset(&mut file) {
-        Some(off) => off,
-        _ => {
-            return Ok(false);
-        }
-    };
-    file.seek(SeekFrom::Start(ecd_off))?;
-    let r = match apk_frauder::Signature::deserialize(&mut file) {
-        Ok(sig) => EndCentralDirectory::SIGNATURE == sig,
-        _ => false,
-    };
-    file.seek(SeekFrom::Start(pos))?;
-    Ok(r)
-}

androscalpel/src/value.rs

@@ -3,13 +3,10 @@
 use serde::{Deserialize, Serialize};
 use std::collections::HashSet;
 
-#[cfg(feature = "python")]
-use pyo3::{exceptions::PyTypeError, prelude::*};
+use pyo3::exceptions::PyTypeError;
+use pyo3::prelude::*;
 
-use crate::{
-    DexAnnotation, DexString, MethodHandle, Result, Visitable, VisitableMut, Visitor, VisitorMut,
-    dex_id::*, scalar::*,
-};
+use crate::{dex_id::*, scalar::*, DexAnnotation, DexString, MethodHandle};
 
 #[derive(Debug, Clone, PartialEq, Deserialize, Serialize)]
 pub enum DexValue {
@@ -33,94 +30,43 @@ pub enum DexValue {
     Boolean(DexBoolean),
 }
 
-impl<V: Visitor> Visitable<V> for DexValue {
-    fn default_visit(&self, v: &mut V) -> Result<()> {
-        match self {
-            Self::Byte(val) => v.visit_byte(val),
-            Self::Short(val) => v.visit_short(val),
-            Self::Char(val) => v.visit_char(val),
-            Self::Int(val) => v.visit_int(val),
-            Self::Long(val) => v.visit_long(val),
-            Self::Float(val) => v.visit_float(val),
-            Self::Double(val) => v.visit_double(val),
-            Self::MethodType(val) => v.visit_method_type(val),
-            Self::MethodHandle(val) => v.visit_method_handle(val),
-            Self::String(val) => v.visit_string(val),
-            Self::Type(val) => v.visit_type(val),
-            Self::Field(val) => v.visit_field_id(val),
-            Self::Method(val) => v.visit_method_id(val),
-            Self::Enum(val) => v.visit_enum_id(val),
-            Self::Array(val) => v.visit_array(val),
-            Self::Annotation(val) => v.visit_annotation(val),
-            Self::Null(val) => v.visit_null(val),
-            Self::Boolean(val) => v.visit_bool(val),
-        }
-    }
-}
-
-impl<V: VisitorMut> VisitableMut<V> for DexValue {
-    fn default_visit_mut(self, v: &mut V) -> Result<Self> {
-        match self {
-            Self::Byte(val) => Ok(Self::Byte(v.visit_byte(val)?)),
-            Self::Short(val) => Ok(Self::Short(v.visit_short(val)?)),
-            Self::Char(val) => Ok(Self::Char(v.visit_char(val)?)),
-            Self::Int(val) => Ok(Self::Int(v.visit_int(val)?)),
-            Self::Long(val) => Ok(Self::Long(v.visit_long(val)?)),
-            Self::Float(val) => Ok(Self::Float(v.visit_float(val)?)),
-            Self::Double(val) => Ok(Self::Double(v.visit_double(val)?)),
-            Self::MethodType(val) => Ok(Self::MethodType(v.visit_method_type(val)?)),
-            Self::MethodHandle(val) => Ok(Self::MethodHandle(v.visit_method_handle(val)?)),
-            Self::String(val) => Ok(Self::String(v.visit_string(val)?)),
-            Self::Type(val) => Ok(Self::Type(v.visit_type(val)?)),
-            Self::Field(val) => Ok(Self::Field(v.visit_field_id(val)?)),
-            Self::Method(val) => Ok(Self::Method(v.visit_method_id(val)?)),
-            Self::Enum(val) => Ok(Self::Enum(v.visit_enum_id(val)?)),
-            Self::Array(val) => Ok(Self::Array(v.visit_array(val)?)),
-            Self::Annotation(val) => Ok(Self::Annotation(v.visit_annotation(val)?)),
-            Self::Null(val) => Ok(Self::Null(v.visit_null(val)?)),
-            Self::Boolean(val) => Ok(Self::Boolean(v.visit_bool(val)?)),
-        }
-    }
-}
-
-#[cfg(feature = "python")]
 impl<'source> FromPyObject<'source> for DexValue {
-    fn extract_bound(ob: &Bound<'source, PyAny>) -> PyResult<Self> {
-        if let Ok(val) = DexByte::extract_bound(ob) {
+    fn extract(ob: &'source PyAny) -> PyResult<Self> {
+        if let Ok(val) = DexByte::extract(ob) {
             Ok(Self::Byte(val))
-        } else if let Ok(val) = DexShort::extract_bound(ob) {
+        } else if let Ok(val) = DexShort::extract(ob) {
             Ok(Self::Short(val))
-        } else if let Ok(val) = DexChar::extract_bound(ob) {
+        } else if let Ok(val) = DexChar::extract(ob) {
             Ok(Self::Char(val))
-        } else if let Ok(val) = DexInt::extract_bound(ob) {
+        } else if let Ok(val) = DexInt::extract(ob) {
             Ok(Self::Int(val))
-        } else if let Ok(val) = DexLong::extract_bound(ob) {
+        } else if let Ok(val) = DexLong::extract(ob) {
             Ok(Self::Long(val))
-        } else if let Ok(val) = DexFloat::extract_bound(ob) {
+        } else if let Ok(val) = DexFloat::extract(ob) {
             Ok(Self::Float(val))
-        } else if let Ok(val) = DexDouble::extract_bound(ob) {
+        } else if let Ok(val) = DexDouble::extract(ob) {
             Ok(Self::Double(val))
-        } else if let Ok(val) = DexString::extract_bound(ob) {
+        } else if let Ok(val) = DexString::extract(ob) {
             Ok(Self::String(val))
-        } else if let Ok(val) = IdType::extract_bound(ob) {
+        } else if let Ok(val) = IdType::extract(ob) {
             Ok(Self::Type(val))
-        } else if let Ok(val) = IdField::extract_bound(ob) {
+        } else if let Ok(val) = IdField::extract(ob) {
             Ok(Self::Field(val))
-        } else if let Ok(val) = IdMethod::extract_bound(ob) {
+        } else if let Ok(val) = IdMethod::extract(ob) {
             Ok(Self::Method(val))
-        } else if let Ok(val) = IdEnum::extract_bound(ob) {
+        } else if let Ok(val) = IdEnum::extract(ob) {
             Ok(Self::Enum(val))
-        } else if let Ok(val) = DexArray::extract_bound(ob) {
+        } else if let Ok(val) = DexArray::extract(ob) {
             Ok(Self::Array(val))
-        } else if let Ok(val) = DexAnnotation::extract_bound(ob) {
+        } else if let Ok(val) = DexAnnotation::extract(ob) {
             Ok(Self::Annotation(val))
-        } else if let Ok(val) = DexNull::extract_bound(ob) {
+        } else if let Ok(val) = DexNull::extract(ob) {
             Ok(Self::Null(val))
-        } else if let Ok(val) = DexBoolean::extract_bound(ob) {
+        } else if let Ok(val) = DexBoolean::extract(ob) {
             Ok(Self::Boolean(val))
-        } else if let Ok(val) = IdMethodType::extract_bound(ob) {
+        } else if let Ok(val) = IdMethodType::extract(ob) {
             Ok(Self::MethodType(val))
-        } else if let Ok(val) = MethodHandle::extract_bound(ob) {
+        } else if let Ok(val) = MethodHandle::extract(ob) {
             Ok(Self::MethodHandle(val))
         } else {
             Err(PyErr::new::<PyTypeError, _>(format!(
@@ -306,32 +252,28 @@ impl DexValue {
         }
     }
 }
-#[cfg(feature = "python")]
-impl<'py> IntoPyObject<'py> for DexValue {
-    type Target = PyAny;
-    type Output = Bound<'py, Self::Target>;
-    type Error = PyErr;
 
-    fn into_pyobject(self, py: Python<'py>) -> Result<Self::Output, Self::Error> {
+impl IntoPy<PyObject> for DexValue {
+    fn into_py(self, py: Python<'_>) -> PyObject {
         match self {
-            DexValue::Byte(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::Short(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::Char(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::Int(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::Long(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::Float(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::Double(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::MethodType(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::MethodHandle(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::String(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::Type(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::Field(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::Method(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::Enum(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::Array(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::Annotation(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::Null(val) => Bound::new(py, val).map(Bound::<_>::into_any),
-            DexValue::Boolean(val) => Bound::new(py, val).map(Bound::<_>::into_any),
+            DexValue::Byte(val) => val.into_py(py),
+            DexValue::Short(val) => val.into_py(py),
+            DexValue::Char(val) => val.into_py(py),
+            DexValue::Int(val) => val.into_py(py),
+            DexValue::Long(val) => val.into_py(py),
+            DexValue::Float(val) => val.into_py(py),
+            DexValue::Double(val) => val.into_py(py),
+            DexValue::MethodType(val) => val.into_py(py),
+            DexValue::MethodHandle(val) => val.into_py(py),
+            DexValue::String(val) => val.into_py(py),
+            DexValue::Type(val) => val.into_py(py),
+            DexValue::Field(val) => val.into_py(py),
+            DexValue::Method(val) => val.into_py(py),
+            DexValue::Enum(val) => val.into_py(py),
+            DexValue::Array(val) => val.into_py(py),
+            DexValue::Annotation(val) => val.into_py(py),
+            DexValue::Null(val) => val.into_py(py),
+            DexValue::Boolean(val) => val.into_py(py),
         }
     }
 }

androscalpel/src/visitor.rs

@@ -1,419 +0,0 @@
-//! The visitor trait and common implementations.
-
-use crate::{
-    ins::Instruction, scalar::*, Apk, CallSite, Class, Code, DexAnnotation, DexAnnotationItem,
-    DexFile, DexString, DexValue, Field, FieldVisibility, HiddenApiData, HiddenApiDomain,
-    HiddenApiPermission, IdEnum, IdField, IdMethod, IdMethodType, IdType, Method, MethodHandle,
-    MethodVisibility, Result,
-};
-use rayon::prelude::*;
-use std::collections::HashSet;
-
-pub trait Visitor: Sized {
-    fn visit_instruction(&mut self, ins: &Instruction) -> Result<()> {
-        ins.default_visit(self)
-    }
-    fn visit_string(&mut self, string: &DexString) -> Result<()> {
-        string.default_visit(self)
-    }
-    fn visit_type(&mut self, ty: &IdType) -> Result<()> {
-        ty.default_visit(self)
-    }
-    fn visit_method_type(&mut self, mty: &IdMethodType) -> Result<()> {
-        mty.default_visit(self)
-    }
-    fn visit_field_id(&mut self, id: &IdField) -> Result<()> {
-        id.default_visit(self)
-    }
-    fn visit_method_id(&mut self, id: &IdMethod) -> Result<()> {
-        id.default_visit(self)
-    }
-    fn visit_enum_id(&mut self, id: &IdEnum) -> Result<()> {
-        id.default_visit(self)
-    }
-    fn visit_method_handle(&mut self, handle: &MethodHandle) -> Result<()> {
-        handle.default_visit(self)
-    }
-    fn visit_call_site(&mut self, site: &CallSite) -> Result<()> {
-        site.default_visit(self)
-    }
-    fn visit_value(&mut self, val: &DexValue) -> Result<()> {
-        val.default_visit(self)
-    }
-    fn visit_byte(&mut self, _: &DexByte) -> Result<()> {
-        Ok(())
-    }
-    fn visit_short(&mut self, _: &DexShort) -> Result<()> {
-        Ok(())
-    }
-    fn visit_char(&mut self, _: &DexChar) -> Result<()> {
-        Ok(())
-    }
-    fn visit_int(&mut self, _: &DexInt) -> Result<()> {
-        Ok(())
-    }
-    fn visit_long(&mut self, _: &DexLong) -> Result<()> {
-        Ok(())
-    }
-    fn visit_float(&mut self, _: &DexFloat) -> Result<()> {
-        Ok(())
-    }
-    fn visit_double(&mut self, _: &DexDouble) -> Result<()> {
-        Ok(())
-    }
-    fn visit_array(&mut self, val: &DexArray) -> Result<()> {
-        val.default_visit(self)
-    }
-    fn visit_annotation(&mut self, val: &DexAnnotation) -> Result<()> {
-        val.default_visit(self)
-    }
-    fn visit_null(&mut self, _: &DexNull) -> Result<()> {
-        Ok(())
-    }
-    fn visit_bool(&mut self, _: &DexBoolean) -> Result<()> {
-        Ok(())
-    }
-    fn visit_class(&mut self, class: &Class) -> Result<()> {
-        class.default_visit(self)
-    }
-    fn visit_field(&mut self, field: &Field) -> Result<()> {
-        field.default_visit(self)
-    }
-    fn visit_method(&mut self, method: &Method) -> Result<()> {
-        method.default_visit(self)
-    }
-    fn visit_annotation_item(&mut self, annotation: &DexAnnotationItem) -> Result<()> {
-        annotation.default_visit(self)
-    }
-    fn visit_field_visibility(&mut self, _: &FieldVisibility) -> Result<()> {
-        Ok(())
-    }
-    fn visit_method_visibility(&mut self, _: &MethodVisibility) -> Result<()> {
-        Ok(())
-    }
-    fn visit_hidden_api_data(&mut self, hidden_api: &HiddenApiData) -> Result<()> {
-        hidden_api.default_visit(self)
-    }
-    fn visit_hidden_api_permission(&mut self, _: &HiddenApiPermission) -> Result<()> {
-        Ok(())
-    }
-    fn visit_hidden_api_domain(&mut self, _: &HiddenApiDomain) -> Result<()> {
-        Ok(())
-    }
-    fn visit_code(&mut self, code: &Code) -> Result<()> {
-        code.default_visit(self)
-    }
-    fn visit_dex_file(&mut self, dex_file: &DexFile) -> Result<()> {
-        dex_file.default_visit(self)
-    }
-    fn visit_apk(&mut self, apk: &Apk) -> Result<()> {
-        apk.default_visit(self)
-    }
-}
-
-pub trait VisitorMut: Sized {
-    fn visit_instruction(&mut self, ins: Instruction) -> Result<Instruction> {
-        ins.default_visit_mut(self)
-    }
-    fn visit_string(&mut self, string: DexString) -> Result<DexString> {
-        string.default_visit_mut(self)
-    }
-    fn visit_type(&mut self, ty: IdType) -> Result<IdType> {
-        ty.default_visit_mut(self)
-    }
-    fn visit_method_type(&mut self, mty: IdMethodType) -> Result<IdMethodType> {
-        mty.default_visit_mut(self)
-    }
-    fn visit_field_id(&mut self, id: IdField) -> Result<IdField> {
-        id.default_visit_mut(self)
-    }
-    fn visit_method_id(&mut self, id: IdMethod) -> Result<IdMethod> {
-        id.default_visit_mut(self)
-    }
-    fn visit_enum_id(&mut self, id: IdEnum) -> Result<IdEnum> {
-        id.default_visit_mut(self)
-    }
-    fn visit_method_handle(&mut self, handle: MethodHandle) -> Result<MethodHandle> {
-        handle.default_visit_mut(self)
-    }
-    fn visit_call_site(&mut self, site: CallSite) -> Result<CallSite> {
-        site.default_visit_mut(self)
-    }
-    fn visit_value(&mut self, val: DexValue) -> Result<DexValue> {
-        val.default_visit_mut(self)
-    }
-    fn visit_byte(&mut self, val: DexByte) -> Result<DexByte> {
-        Ok(val)
-    }
-    fn visit_short(&mut self, val: DexShort) -> Result<DexShort> {
-        Ok(val)
-    }
-    fn visit_char(&mut self, val: DexChar) -> Result<DexChar> {
-        Ok(val)
-    }
-    fn visit_int(&mut self, val: DexInt) -> Result<DexInt> {
-        Ok(val)
-    }
-    fn visit_long(&mut self, val: DexLong) -> Result<DexLong> {
-        Ok(val)
-    }
-    fn visit_float(&mut self, val: DexFloat) -> Result<DexFloat> {
-        Ok(val)
-    }
-    fn visit_double(&mut self, val: DexDouble) -> Result<DexDouble> {
-        Ok(val)
-    }
-    fn visit_array(&mut self, val: DexArray) -> Result<DexArray> {
-        val.default_visit_mut(self)
-    }
-    fn visit_annotation(&mut self, val: DexAnnotation) -> Result<DexAnnotation> {
-        val.default_visit_mut(self)
-    }
-    fn visit_null(&mut self, val: DexNull) -> Result<DexNull> {
-        Ok(val)
-    }
-    fn visit_bool(&mut self, val: DexBoolean) -> Result<DexBoolean> {
-        Ok(val)
-    }
-    fn visit_class(&mut self, class: Class) -> Result<Class> {
-        class.default_visit_mut(self)
-    }
-    fn visit_field(&mut self, field: Field) -> Result<Field> {
-        field.default_visit_mut(self)
-    }
-    fn visit_method(&mut self, method: Method) -> Result<Method> {
-        method.default_visit_mut(self)
-    }
-    fn visit_annotation_item(
-        &mut self,
-        annotation: DexAnnotationItem,
-    ) -> Result<DexAnnotationItem> {
-        annotation.default_visit_mut(self)
-    }
-    fn visit_field_visibility(&mut self, visibility: FieldVisibility) -> Result<FieldVisibility> {
-        Ok(visibility)
-    }
-    fn visit_method_visibility(
-        &mut self,
-        visibility: MethodVisibility,
-    ) -> Result<MethodVisibility> {
-        Ok(visibility)
-    }
-    fn visit_hidden_api_data(&mut self, hidden_api: HiddenApiData) -> Result<HiddenApiData> {
-        hidden_api.default_visit_mut(self)
-    }
-    fn visit_hidden_api_permission(
-        &mut self,
-        perm: HiddenApiPermission,
-    ) -> Result<HiddenApiPermission> {
-        Ok(perm)
-    }
-    fn visit_hidden_api_domain(&mut self, domain: HiddenApiDomain) -> Result<HiddenApiDomain> {
-        Ok(domain)
-    }
-    fn visit_code(&mut self, code: Code) -> Result<Code> {
-        code.default_visit_mut(self)
-    }
-    fn visit_dex_file(&mut self, dex_file: DexFile) -> Result<DexFile> {
-        dex_file.default_visit_mut(self)
-    }
-    fn visit_apk(&mut self, apk: Apk) -> Result<Apk> {
-        apk.default_visit_mut(self)
-    }
-}
-
-/// Trait for structures that can be visited.
-/// In theorie, this is not required, as the [`Visitor`] trait implement
-/// all tree traversal, however if the visitor do not do a complex
-/// operation it can be usefull to have a default tree traversal accessible
-/// to re-implemented trait methods.
-pub trait Visitable<V: Visitor> {
-    fn default_visit(&self, visitor: &mut V) -> Result<()>;
-}
-
-/// Trait for structures that can be modified by a visitor.
-/// In theorie, this is not required, as the [`VisitorMut`] trait implement
-/// all tree traversal, however if the visitor do not do a complex
-/// operation it can be usefull to have a default tree traversal accessible
-/// to re-implemented trait methods.
-pub trait VisitableMut<V: VisitorMut> {
-    fn default_visit_mut(self, visitor: &mut V) -> Result<Self>
-    where
-        Self: Sized;
-}
-
-#[derive(Debug, Default)]
-pub struct StringCollector {
-    pub strings: HashSet<DexString>,
-}
-
-impl StringCollector {
-    pub fn result(self) -> HashSet<DexString> {
-        self.strings
-    }
-}
-
-impl Visitor for StringCollector {
-    fn visit_string(&mut self, string: &DexString) -> Result<()> {
-        self.strings.insert(string.clone());
-        Ok(())
-    }
-}
-
-#[derive(Debug, Default)]
-pub struct TypeCollector {
-    pub types: HashSet<IdType>,
-}
-
-impl TypeCollector {
-    pub fn result(self) -> HashSet<IdType> {
-        self.types
-    }
-}
-
-impl Visitor for TypeCollector {
-    fn visit_type(&mut self, ty: &IdType) -> Result<()> {
-        self.types.insert(ty.clone());
-        Ok(())
-    }
-
-    fn visit_dex_file(&mut self, dex: &DexFile) -> Result<()> {
-        let results: Vec<_> = dex
-            .classes
-            .par_iter()
-            .map(|(ty, cls)| {
-                let mut v = Self::default();
-                v.visit_type(ty)
-                    .map(|_| v.visit_class(cls))
-                    .map(|_| v.result())
-            })
-            .collect::<Result<_, _>>()?;
-        for r in results.into_iter() {
-            self.types.extend(r);
-        }
-        Ok(())
-    }
-}
-
-#[derive(Debug, Default)]
-pub struct MethodTypeCollector {
-    pub method_types: HashSet<IdMethodType>,
-}
-
-impl MethodTypeCollector {
-    pub fn result(self) -> HashSet<IdMethodType> {
-        self.method_types
-    }
-}
-
-impl Visitor for MethodTypeCollector {
-    fn visit_method_type(&mut self, mty: &IdMethodType) -> Result<()> {
-        self.method_types.insert(mty.clone());
-        Ok(())
-    }
-
-    fn visit_dex_file(&mut self, dex: &DexFile) -> Result<()> {
-        let results: Vec<_> = dex
-            .classes
-            .par_iter()
-            .map(|(ty, cls)| {
-                let mut v = Self::default();
-                v.visit_type(ty)
-                    .map(|_| v.visit_class(cls))
-                    .map(|_| v.result())
-            })
-            .collect::<Result<_, _>>()?;
-        for r in results.into_iter() {
-            self.method_types.extend(r);
-        }
-        Ok(())
-    }
-}
-
-#[derive(Debug, Default)]
-pub struct FieldIdCollector {
-    pub field_ids: HashSet<IdField>,
-}
-
-impl FieldIdCollector {
-    pub fn result(self) -> HashSet<IdField> {
-        self.field_ids
-    }
-}
-
-impl Visitor for FieldIdCollector {
-    fn visit_field_id(&mut self, id: &IdField) -> Result<()> {
-        self.field_ids.insert(id.clone());
-        Ok(())
-    }
-
-    fn visit_dex_file(&mut self, dex: &DexFile) -> Result<()> {
-        let results: Vec<_> = dex
-            .classes
-            .par_iter()
-            .map(|(ty, cls)| {
-                let mut v = Self::default();
-                v.visit_type(ty)
-                    .map(|_| v.visit_class(cls))
-                    .map(|_| v.result())
-            })
-            .collect::<Result<_, _>>()?;
-        for r in results.into_iter() {
-            self.field_ids.extend(r);
-        }
-        Ok(())
-    }
-}
-
-#[derive(Debug, Default)]
-pub struct MethodIdCollector {
-    pub method_ids: HashSet<IdMethod>,
-}
-
-impl MethodIdCollector {
-    pub fn result(self) -> HashSet<IdMethod> {
-        self.method_ids
-    }
-}
-
-impl Visitor for MethodIdCollector {
-    fn visit_method_id(&mut self, id: &IdMethod) -> Result<()> {
-        self.method_ids.insert(id.clone());
-        Ok(())
-    }
-
-    fn visit_dex_file(&mut self, dex: &DexFile) -> Result<()> {
-        let results: Vec<_> = dex
-            .classes
-            .par_iter()
-            .map(|(ty, cls)| {
-                let mut v = Self::default();
-                v.visit_type(ty)
-                    .map(|_| v.visit_class(cls))
-                    .map(|_| v.result())
-            })
-            .collect::<Result<_, _>>()?;
-        for r in results.into_iter() {
-            self.method_ids.extend(r);
-        }
-        Ok(())
-    }
-}
-
-#[derive(Debug, Default)]
-pub struct MethodHandleCollector {
-    pub handles: HashSet<MethodHandle>,
-}
-
-impl MethodHandleCollector {
-    pub fn result(self) -> HashSet<MethodHandle> {
-        self.handles
-    }
-}
-
-impl Visitor for MethodHandleCollector {
-    fn visit_method_handle(&mut self, handle: &MethodHandle) -> Result<()> {
-        self.handles.insert(handle.clone());
-        Ok(())
-    }
-}

androscalpel_platform_api_list/Cargo.toml

@@ -1,10 +0,0 @@
-[package]
-name = "androscalpel_platform_api_list"
-version = "0.1.0"
-edition = "2024"
-license = "GPL-3.0-or-later"
-
-[dependencies]
-
-[features]
-sdk34 = []

androscalpel_platform_api_list/README.md

@@ -1,5 +0,0 @@
-# Android Platform API
-
-List of android platform API (API that are either in the android SDK or hidden API).
-
-The list is extracted from the default android emulator the respective SDKs.

androscalpel_platform_api_list/src/lib.rs

@@ -1,20 +0,0 @@
-pub mod sdk34;
-pub use sdk34::*;
-
-// TODO: automate addign a new SDK
-//  TODO: different version with different features
-
-/// Test if the class is part of the platform classes.
-pub fn is_platform_class(smali: &str) -> bool {
-    is_sdk34_platform_class(smali)
-}
-
-/// Test if the field is part of the platform classes.
-pub fn is_platform_field(smali: &str) -> bool {
-    is_sdk34_platform_field(smali)
-}
-
-/// Test if the method is part of the platform classes.
-pub fn is_platform_method(smali: &str) -> bool {
-    is_sdk34_platform_method(smali)
-}

androscalpel_platform_api_list/src/sdk34/mod.rs

@@ -1,60 +0,0 @@
-use std::collections::HashSet;
-use std::sync::LazyLock;
-
-pub static PLATFORM_CLASSES_SDK34: LazyLock<HashSet<&'static str>> =
-    LazyLock::new(|| include_str!("classes.txt").lines().collect());
-pub static PLATFORM_FIELDS_SDK34: LazyLock<HashSet<&'static str>> =
-    LazyLock::new(|| include_str!("fields.txt").lines().collect());
-pub static PLATFORM_METHODS_SDK34: LazyLock<HashSet<&'static str>> =
-    LazyLock::new(|| include_str!("methods.txt").lines().collect());
-
-/// Test if the class is part of the platform classes of the SDK 34.
-pub fn is_sdk34_platform_class(smali: &str) -> bool {
-    PLATFORM_CLASSES_SDK34.contains(smali)
-}
-
-/// Test if the field is part of the platform classes of the SDK 34.
-pub fn is_sdk34_platform_field(smali: &str) -> bool {
-    PLATFORM_FIELDS_SDK34.contains(smali)
-}
-
-/// Test if the method is part of the platform classes of the SDK 34.
-pub fn is_sdk34_platform_method(smali: &str) -> bool {
-    PLATFORM_METHODS_SDK34.contains(smali)
-}
-
-#[cfg(test)]
-mod test {
-    use super::*;
-
-    #[test]
-    fn test_classes_sdk34_set() {
-        assert!(PLATFORM_CLASSES_SDK34
-            .contains("Landroid/accessibilityservice/AccessibilityGestureEvent;"));
-        let s: String = "Landroid/accessibilityservice/AccessibilityInputMethodSession;".into();
-        assert!(PLATFORM_CLASSES_SDK34.contains(s.as_str()));
-    }
-
-    #[test]
-    fn test_fields_sdk34_set() {
-        assert!(PLATFORM_FIELDS_SDK34.contains(
-            "Landroid/accessibilityservice/AccessibilityButtonController$$\
-                ExternalSyntheticLambda1;->f$0:Landroid/accessibilityservice/\
-                AccessibilityButtonController;"
-        ));
-        let s: String = "Landroid/accessibilityservice/AccessibilityButtonController;->mLock:Ljava/lang/Object;".into();
-        assert!(PLATFORM_FIELDS_SDK34.contains(s.as_str()));
-    }
-
-    #[test]
-    fn test_methods_sdk34_set() {
-        assert!(PLATFORM_METHODS_SDK34
-            .contains("Landroid/accessibilityservice/AccessibilityButtonController$$ExternalSyntheticLambda1;->run()V"));
-        let s: String = "Landroid/accessibilityservice/AccessibilityButtonController;->\
-                         $r8$lambda$h5Na0_pkg6ffhlKQPqxrTXannSI(\
-                             Landroid/accessibilityservice/AccessibilityButtonController;\
-                             Landroid/accessibilityservice/AccessibilityButtonController$AccessibilityButtonCallback;\
-                             )V".into();
-        assert!(PLATFORM_METHODS_SDK34.contains(s.as_str()));
-    }
-}

androscalpel_serializer/Cargo.toml

@@ -1,8 +1,7 @@
 [package]
 name = "androscalpel_serializer"
 version = "0.1.0"
-edition = "2024"
-license = "GPL-3.0-or-later"
+edition = "2021"
 
 [dependencies]
 androscalpel_serializer_derive = { path = "../androscalpel_serializer_derive" }

androscalpel_serializer/src/core/leb.rs

@@ -19,7 +19,7 @@ pub struct Uleb128p1(pub u32);
 
 impl Sleb128 {
     fn get_serialized_bytes(&self) -> Vec<u8> {
-        let &Self(mut val) = self;
+        let Self(mut val) = self;
         let mut bytes = vec![];
         loop {
             let byte = val as u8 & 0b0111_1111;
@@ -81,7 +81,7 @@ impl Serializable for Sleb128 {
 
 impl Uleb128 {
     fn get_serialized_bytes(&self) -> Vec<u8> {
-        let &Self(mut val) = self;
+        let Self(mut val) = self;
         let mut bytes = vec![];
         loop {
             let byte = val as u8 & 0b0111_1111;

androscalpel_serializer/src/core/mod.rs

@@ -17,7 +17,6 @@ pub enum Error {
     DeserializationError(String),
     InvalidStringEncoding(String),
     InconsistantStruct(String),
-    OutOfBound(String),
 }
 
 pub type Result<T> = core::result::Result<T, Error>;
@@ -33,7 +32,6 @@ impl std::fmt::Display for Error {
             Self::DeserializationError(msg) => write!(f, "{}", msg),
             Self::InvalidStringEncoding(msg) => write!(f, "{}", msg),
             Self::InconsistantStruct(msg) => write!(f, "{}", msg),
-            Self::OutOfBound(msg) => write!(f, "{}", msg),
         }
     }
 }
@@ -390,9 +388,7 @@ mod test {
     fn serialize_u64() {
         assert_eq!(
             0x123456789ABCDEF0u64.serialize_to_vec().unwrap(),
-            vec![
-                0xF0u8, 0xDEu8, 0xBCu8, 0x9Au8, 0x78u8, 0x56u8, 0x34u8, 0x12u8
-            ]
+            vec![0xF0u8, 0xDEu8, 0xBCu8, 0x9Au8, 0x78u8, 0x56u8, 0x34u8, 0x12u8]
         );
     }
 

androscalpel_serializer/src/core/string.rs

@@ -143,7 +143,7 @@ impl StringDataItem {
             let two = self.data[i] as u32;
             i += 1;
             if one & 0x20 == 0 {
-                utf16_string.push(((one & 0x1f) << 6) | (two & 0x3f));
+                utf16_string.push((one & 0x1f) << 6 | (two & 0x3f));
                 continue;
             }
 
@@ -155,7 +155,7 @@ impl StringDataItem {
             let three = self.data[i] as u32;
             i += 1;
             if one & 0x10 == 0 {
-                utf16_string.push(((one & 0x0f) << 12) | ((two & 0x3f) << 6) | (three & 0x3f));
+                utf16_string.push((one & 0x0f) << 12 | (two & 0x3f) << 6 | (three & 0x3f));
                 continue;
             }
 
@@ -167,7 +167,7 @@ impl StringDataItem {
             let four = self.data[i] as u32;
             i += 1;
             let code_point =
-                ((one & 0x0f) << 18) | ((two & 0x3f) << 12) | ((three & 0x3f) << 6) | (four & 0x3f);
+                (one & 0x0f) << 18 | (two & 0x3f) << 12 | (three & 0x3f) << 6 | (four & 0x3f);
             let mut pair = ((code_point >> 10) + 0xd7c0) & 0xffff;
             pair |= ((code_point & 0x03ff) + 0xdc00) << 16;
             utf16_string.push(pair);
@@ -357,34 +357,26 @@ mod test {
     /// Test for bug found in <https://github.com/TkTech/mutf8/tree/master>:
     #[test]
     fn test_tktech_bad_mutf8() {
-        assert!(
-            TryInto::<String>::try_into(StringDataItem {
-                utf16_size: Uleb128(0),
-                data: vec![0x00]
-            })
-            .is_err()
-        );
-        assert!(
-            TryInto::<String>::try_into(StringDataItem {
-                utf16_size: Uleb128(0),
-                data: vec![0xC2]
-            })
-            .is_err()
-        );
-        assert!(
-            TryInto::<String>::try_into(StringDataItem {
-                utf16_size: Uleb128(0),
-                data: vec![0xED]
-            })
-            .is_err()
-        );
-        assert!(
-            TryInto::<String>::try_into(StringDataItem {
-                utf16_size: Uleb128(0),
-                data: vec![0xE2]
-            })
-            .is_err()
-        );
+        assert!(TryInto::<String>::try_into(StringDataItem {
+            utf16_size: Uleb128(0),
+            data: vec![0x00]
+        })
+        .is_err());
+        assert!(TryInto::<String>::try_into(StringDataItem {
+            utf16_size: Uleb128(0),
+            data: vec![0xC2]
+        })
+        .is_err());
+        assert!(TryInto::<String>::try_into(StringDataItem {
+            utf16_size: Uleb128(0),
+            data: vec![0xED]
+        })
+        .is_err());
+        assert!(TryInto::<String>::try_into(StringDataItem {
+            utf16_size: Uleb128(0),
+            data: vec![0xE2]
+        })
+        .is_err());
     }
 
     /// Test from <https://github.com/TkTech/mutf8/tree/master>, test 2 bytes

androscalpel_serializer/src/debug.rs

@@ -1,9 +1,7 @@
 //! Debug structs
 
 use crate as androscalpel_serializer;
-use crate::{
-    Error, NO_INDEX, ReadSeek, Result, Serializable, SerializableUntil, Sleb128, Uleb128, Uleb128p1,
-};
+use crate::{ReadSeek, Result, Serializable, SerializableUntil, Sleb128, Uleb128, Uleb128p1};
 use std::io::Write;
 
 /// <https://source.android.com/docs/core/runtime/dex-format#debug-info-item>
@@ -95,556 +93,6 @@ impl Serializable for DebugInfoItem {
     }
 }
 
-/// The name and type of the variable in a register.
-#[derive(Debug, PartialEq, Eq, Clone, Copy)]
-pub struct DebugRegState {
-    pub type_idx: Option<u32>,
-    pub name_idx: Option<u32>,
-    pub sig_idx: Option<u32>,
-    pub in_scope: bool,
-}
-
-/// A simplified debug information
-#[derive(Debug, PartialEq, Eq, Clone)]
-pub enum DebugInfo {
-    DefLocal {
-        addr: u32,
-        reg: u32,
-        val: DebugRegState,
-    },
-    EndLocal {
-        addr: u32,
-        reg: u32,
-    },
-    PrologueEnd {
-        addr: u32,
-    },
-    EpilogueBegin {
-        addr: u32,
-    },
-    SetSourceFile {
-        addr: u32,
-        source_file_idx: Option<u32>,
-    },
-    SetLineNumber {
-        addr: u32,
-        line_num: u32,
-    },
-    EndOfData,
-}
-
-impl DebugInfo {
-    pub fn get_addr(&self) -> u32 {
-        match self {
-            Self::DefLocal { addr, .. } => *addr,
-            Self::EndLocal { addr, .. } => *addr,
-            Self::PrologueEnd { addr } => *addr,
-            Self::EpilogueBegin { addr } => *addr,
-            Self::SetSourceFile { addr, .. } => *addr,
-            Self::SetLineNumber { addr, .. } => *addr,
-            Self::EndOfData => u32::MAX, // TODO should be an Option::None?
-        }
-    }
-}
-
-/// A state machine that interpret a [`DebugInfoItem`].
-#[derive(Debug, PartialEq, Eq, Clone)]
-pub struct DebugInfoReader {
-    debug_info: DebugInfoItem,
-    pub pc: usize,
-    pub address: u32,
-    pub line: u32,
-    // Those are registers described in the doc but not necessary in the end
-    //pub source_file_idx: Option<u32>,
-    //pub prologue_end: bool,
-    //pub epilogue_begin: bool,
-    pub register_states: Vec<DebugRegState>,
-}
-
-impl DebugInfoReader {
-    pub fn new(
-        debug_info: DebugInfoItem,
-        //source_file_idx: Option<u32>,
-        //nb_reg: usize
-    ) -> Self {
-        let line = debug_info.line_start.0;
-        Self {
-            debug_info,
-            pc: 0,
-            address: 0,
-            line,
-            //source_file_idx,
-            //prologue_end: false,
-            //epilogue_begin: false,
-            //register_states: vec![
-            //    DebugRegState {
-            //        name_idx: None,
-            //        type_idx: None,
-            //        sig_idx: None,
-            //        in_scope: false,
-            //    };
-            //    nb_reg
-            //],
-            register_states: vec![], // In the end, it's easier to grow this on the fly
-        }
-    }
-
-    pub fn get_ins(&self) -> Result<DbgBytecode> {
-        if self.pc >= self.debug_info.bytecode.len() {
-            return Err(Error::OutOfBound(
-                "Try to read an instruction out of bound, maybe after the end of the debug sequence."
-                .into()
-            ));
-        }
-        Ok(self.debug_info.bytecode[self.pc])
-    }
-
-    pub fn tick(&mut self) -> Option<DebugInfo> {
-        let ins = if let Ok(ins) = self.get_ins() {
-            ins
-        } else {
-            return Some(DebugInfo::EndOfData);
-        };
-        self.pc += 1;
-        match ins {
-            DbgBytecode::EndSequence => {
-                self.pc = self.debug_info.bytecode.len();
-                Some(DebugInfo::EndOfData)
-            }
-            DbgBytecode::AdvancePC {
-                addr_diff: Uleb128(addr_diff),
-            } => {
-                self.address += addr_diff;
-                None
-            }
-            DbgBytecode::AdvanceLine {
-                line_diff: Sleb128(line_diff),
-            } => {
-                self.line = (self.line as i32 + line_diff) as u32;
-                None
-            }
-            DbgBytecode::StartLocal {
-                register_num: Uleb128(register_num),
-                name_idx,
-                type_idx,
-            } => {
-                while self.register_states.len() < (register_num + 1) as usize {
-                    self.register_states.push(DebugRegState {
-                        name_idx: None,
-                        type_idx: None,
-                        sig_idx: None,
-                        in_scope: false,
-                    })
-                }
-                self.register_states[register_num as usize] = DebugRegState {
-                    name_idx: if name_idx == NO_INDEX {
-                        None
-                    } else {
-                        Some(name_idx.0)
-                    },
-                    type_idx: if type_idx == NO_INDEX {
-                        None
-                    } else {
-                        Some(type_idx.0)
-                    },
-                    sig_idx: None,
-                    in_scope: true,
-                };
-                Some(DebugInfo::DefLocal {
-                    addr: self.address,
-                    reg: register_num,
-                    val: self.register_states[register_num as usize],
-                })
-            }
-            DbgBytecode::StartLocalExtended {
-                register_num: Uleb128(register_num),
-                name_idx,
-                type_idx,
-                sig_idx,
-            } => {
-                while self.register_states.len() < (register_num + 1) as usize {
-                    self.register_states.push(DebugRegState {
-                        name_idx: None,
-                        type_idx: None,
-                        sig_idx: None,
-                        in_scope: false,
-                    })
-                }
-                self.register_states[register_num as usize] = DebugRegState {
-                    name_idx: if name_idx == NO_INDEX {
-                        None
-                    } else {
-                        Some(name_idx.0)
-                    },
-                    type_idx: if type_idx == NO_INDEX {
-                        None
-                    } else {
-                        Some(type_idx.0)
-                    },
-                    sig_idx: if sig_idx == NO_INDEX {
-                        None
-                    } else {
-                        Some(sig_idx.0)
-                    },
-                    in_scope: true,
-                };
-                Some(DebugInfo::DefLocal {
-                    addr: self.address,
-                    reg: register_num,
-                    val: self.register_states[register_num as usize],
-                })
-            }
-            DbgBytecode::EndLocal {
-                register_num: Uleb128(register_num),
-            } => {
-                // Yes this can happen
-                while self.register_states.len() < (register_num + 1) as usize {
-                    self.register_states.push(DebugRegState {
-                        name_idx: None,
-                        type_idx: None,
-                        sig_idx: None,
-                        in_scope: false,
-                    })
-                }
-                self.register_states[register_num as usize].in_scope = false;
-                Some(DebugInfo::EndLocal {
-                    addr: self.address,
-                    reg: register_num,
-                })
-            }
-            DbgBytecode::RestartLocal {
-                register_num: Uleb128(register_num),
-            } => {
-                while self.register_states.len() < (register_num + 1) as usize {
-                    self.register_states.push(DebugRegState {
-                        name_idx: None,
-                        type_idx: None,
-                        sig_idx: None,
-                        in_scope: false,
-                    })
-                }
-                self.register_states[register_num as usize].in_scope = true;
-                Some(DebugInfo::DefLocal {
-                    addr: self.address,
-                    reg: register_num,
-                    val: self.register_states[register_num as usize],
-                })
-            }
-            DbgBytecode::SetPrologueEnd => {
-                //self.prologue_end = true;
-                Some(DebugInfo::PrologueEnd { addr: self.address })
-            }
-            DbgBytecode::SetEpilogueBegin => {
-                //self.epilogue_begin = true;
-                Some(DebugInfo::EpilogueBegin { addr: self.address })
-            }
-            DbgBytecode::SetFile { name_idx: NO_INDEX } => {
-                //self.source_file_idx = None;
-                Some(DebugInfo::SetSourceFile {
-                    addr: self.address,
-                    source_file_idx: None,
-                })
-            }
-            DbgBytecode::SetFile {
-                name_idx: Uleb128p1(name_idx),
-            } => {
-                //self.source_file_idx = Some(name_idx);
-                Some(DebugInfo::SetSourceFile {
-                    addr: self.address,
-                    source_file_idx: Some(name_idx),
-                })
-            }
-            DbgBytecode::SpecialOpcode(op) => {
-                //if op >= 0x0a {
-                //    self.prologue_end = false;
-                //    self.epilogue_begin = true;
-                //}
-                // See <https://source.android.com/docs/core/runtime/dex-format#opcodes>
-                let adjusted_opcode = op as u32 - 0x0a;
-                self.line = (self.line as i32 + (adjusted_opcode as i32 % 15) - 4) as u32;
-                self.address += adjusted_opcode / 15;
-                Some(DebugInfo::SetLineNumber {
-                    addr: self.address,
-                    line_num: self.line,
-                })
-            }
-        }
-    }
-
-    pub fn next_info(&mut self) -> DebugInfo {
-        loop {
-            if let Some(info) = self.tick() {
-                return info;
-            }
-        }
-    }
-}
-
-/// A state machine that generate a [`DebugInfoItem`].
-#[derive(Debug, PartialEq, Eq, Clone)]
-pub struct DebugInfoBuilder {
-    debug_infos: Vec<DbgBytecode>,
-    line_start: Option<u32>,
-    parameter_names: Vec<Uleb128p1>,
-
-    //pub pc: usize,
-    address: u32,
-    line: u32,
-    // Those are registers described in the doc but not necessary in the end
-    //pub source_file_idx: Option<u32>,
-    //pub prologue_end: bool,
-    //pub epilogue_begin: bool,
-    register_states: Vec<DebugRegState>,
-    finished: bool,
-}
-
-impl DebugInfoBuilder {
-    pub fn new(parameter_names: Vec<Uleb128p1>) -> Self {
-        Self {
-            debug_infos: vec![],
-            line_start: None,
-            parameter_names,
-            //pc: 0,
-            address: 0,
-            line: 0,
-            //source_file_idx,
-            //prologue_end: false,
-            //epilogue_begin: false,
-            //register_states: vec![
-            //    DebugRegState {
-            //        name_idx: None,
-            //        type_idx: None,
-            //        sig_idx: None,
-            //        in_scope: false,
-            //    };
-            //    nb_reg
-            //],
-            register_states: vec![], // In the end, it's easier to grow this on the fly
-            finished: false,
-        }
-    }
-
-    pub fn add_info(&mut self, info: &DebugInfo) -> Result<()> {
-        if self.finished {
-            return Err(Error::SerializationError(
-                "Cannot add more information: EndSequence has already been send".into(),
-            ));
-        }
-        match info {
-            DebugInfo::DefLocal { addr, reg, val } => {
-                if *addr < self.address {
-                    return Err(Error::SerializationError(format!(
-                        "The address register can only increase, \
-                        found 0x{addr:02x} while register is already \
-                        0x{:02x}",
-                        self.address
-                    )));
-                }
-                while self.register_states.len() < (reg + 1) as usize {
-                    self.register_states.push(DebugRegState {
-                        name_idx: None,
-                        type_idx: None,
-                        sig_idx: None,
-                        in_scope: false,
-                    });
-                }
-                if *addr != self.address {
-                    let addr_diff = *addr - self.address;
-                    self.debug_infos.push(DbgBytecode::AdvancePC {
-                        addr_diff: Uleb128(addr_diff),
-                    });
-                    self.address += addr_diff;
-                }
-                let mut old_val = self.register_states[*reg as usize];
-                let old_val_in_scope = old_val.in_scope;
-                old_val.in_scope = true;
-                if old_val_in_scope && old_val == *val {
-                    self.register_states[*reg as usize].in_scope = true;
-                    self.debug_infos.push(DbgBytecode::RestartLocal {
-                        register_num: Uleb128(*reg),
-                    });
-                } else {
-                    self.register_states[*reg as usize] = *val;
-                    if val.sig_idx.is_some() {
-                        self.debug_infos.push(DbgBytecode::StartLocalExtended {
-                            register_num: Uleb128(*reg),
-                            name_idx: if let Some(name_idx) = val.name_idx {
-                                Uleb128p1(name_idx)
-                            } else {
-                                NO_INDEX
-                            },
-                            type_idx: if let Some(type_idx) = val.type_idx {
-                                Uleb128p1(type_idx)
-                            } else {
-                                NO_INDEX
-                            },
-                            sig_idx: if let Some(sig_idx) = val.sig_idx {
-                                Uleb128p1(sig_idx)
-                            } else {
-                                NO_INDEX
-                            },
-                        })
-                    } else {
-                        self.debug_infos.push(DbgBytecode::StartLocal {
-                            register_num: Uleb128(*reg),
-                            name_idx: if let Some(name_idx) = val.name_idx {
-                                Uleb128p1(name_idx)
-                            } else {
-                                NO_INDEX
-                            },
-                            type_idx: if let Some(type_idx) = val.type_idx {
-                                Uleb128p1(type_idx)
-                            } else {
-                                NO_INDEX
-                            },
-                        })
-                    }
-                }
-                Ok(())
-            }
-            DebugInfo::EndLocal { addr, reg } => {
-                if *addr < self.address {
-                    return Err(Error::SerializationError(format!(
-                        "The address register can only increase, \
-                        found 0x{addr:02x} while register is already \
-                        0x{:02x}",
-                        self.address
-                    )));
-                }
-                while self.register_states.len() < (reg + 1) as usize {
-                    self.register_states.push(DebugRegState {
-                        name_idx: None,
-                        type_idx: None,
-                        sig_idx: None,
-                        in_scope: false,
-                    });
-                }
-                if *addr != self.address {
-                    let addr_diff = *addr - self.address;
-                    self.debug_infos.push(DbgBytecode::AdvancePC {
-                        addr_diff: Uleb128(addr_diff),
-                    });
-                    self.address += addr_diff;
-                }
-                self.debug_infos.push(DbgBytecode::EndLocal {
-                    register_num: Uleb128(*reg),
-                });
-                Ok(())
-            }
-            DebugInfo::PrologueEnd { addr } => {
-                if *addr < self.address {
-                    return Err(Error::SerializationError(format!(
-                        "The address register can only increase, \
-                        found 0x{addr:02x} while register is already \
-                        0x{:02x}",
-                        self.address
-                    )));
-                }
-                if *addr != self.address {
-                    let addr_diff = *addr - self.address;
-                    self.debug_infos.push(DbgBytecode::AdvancePC {
-                        addr_diff: Uleb128(addr_diff),
-                    });
-                    self.address += addr_diff;
-                }
-                self.debug_infos.push(DbgBytecode::SetPrologueEnd);
-                Ok(())
-            }
-            DebugInfo::EpilogueBegin { addr } => {
-                if *addr < self.address {
-                    return Err(Error::SerializationError(format!(
-                        "The address register can only increase, \
-                        found 0x{addr:02x} while register is already \
-                        0x{:02x}",
-                        self.address
-                    )));
-                }
-                if *addr != self.address {
-                    let addr_diff = *addr - self.address;
-                    self.debug_infos.push(DbgBytecode::AdvancePC {
-                        addr_diff: Uleb128(addr_diff),
-                    });
-                    self.address += addr_diff;
-                }
-                self.debug_infos.push(DbgBytecode::SetEpilogueBegin);
-                Ok(())
-            }
-            DebugInfo::SetLineNumber { addr, line_num } => {
-                if *addr < self.address {
-                    return Err(Error::SerializationError(format!(
-                        "The address register can only increase, \
-                        found 0x{addr:02x} while register is already \
-                        0x{:02x}",
-                        self.address
-                    )));
-                }
-                if self.line_start.is_none() {
-                    self.line_start = Some(*line_num);
-                    self.line = *line_num;
-                }
-                let mut line_diff = *line_num as i32 - self.line as i32;
-                let mut addr_diff = addr - self.address;
-                if !(-4..15 - 4).contains(&line_diff) {
-                    self.debug_infos.push(DbgBytecode::AdvanceLine {
-                        line_diff: Sleb128(line_diff),
-                    });
-                    self.line = *line_num;
-                    line_diff = 0;
-                }
-                if addr_diff as i32 * 15 + 0x0a + line_diff + 4 > 0xff {
-                    self.debug_infos.push(DbgBytecode::AdvancePC {
-                        addr_diff: Uleb128(addr_diff),
-                    });
-                    self.address = *addr;
-                    addr_diff = 0;
-                }
-                let op = 0x0a + addr_diff as u8 * 15 + (line_diff + 4) as u8;
-                self.debug_infos.push(DbgBytecode::SpecialOpcode(op));
-                self.address += addr_diff;
-                self.line = (self.line as i32 + line_diff) as u32;
-                Ok(())
-            }
-            DebugInfo::SetSourceFile {
-                addr,
-                source_file_idx,
-            } => {
-                if *addr != self.address {
-                    let addr_diff = *addr - self.address;
-                    self.debug_infos.push(DbgBytecode::AdvancePC {
-                        addr_diff: Uleb128(addr_diff),
-                    });
-                    self.address += addr_diff;
-                }
-                self.debug_infos.push(DbgBytecode::SetFile {
-                    name_idx: if let Some(source_file_idx) = source_file_idx {
-                        Uleb128p1(*source_file_idx)
-                    } else {
-                        NO_INDEX
-                    },
-                });
-                Ok(())
-            }
-            DebugInfo::EndOfData => {
-                self.finished = true;
-                Ok(())
-            }
-        }
-    }
-
-    /// If they are no debug information, return None, else compute and return the [`DebugInfoItem`].
-    pub fn build(self) -> Option<DebugInfoItem> {
-        if self.debug_infos.is_empty() && self.parameter_names.iter().all(|&idx| idx == NO_INDEX) {
-            None
-        } else {
-            Some(DebugInfoItem {
-                line_start: Uleb128(self.line_start.unwrap_or(0)),
-                parameter_names: self.parameter_names,
-                bytecode: self.debug_infos,
-            })
-        }
-    }
-}
-
 #[cfg(test)]
 mod test {
     use super::DbgBytecode::*;
@@ -664,10 +112,6 @@ mod test {
                     addr_diff: Uleb128(51),
                 },
                 SpecialOpcode(14),
-                // End a local that do not already exist
-                EndLocal {
-                    register_num: Uleb128(41),
-                },
             ],
         };
         assert_eq!(
@@ -686,27 +130,4 @@ mod test {
             DbgBytecode::deserialize_from_slice(&advance_line.serialize_to_vec().unwrap()).unwrap()
         );
     }
-
-    #[test]
-    fn test_get_expl_debug() {
-        const RAW_DEBUG: [u8; 10] = [23, 0, 14, 135, 3, 0, 16, 2, 150, 0];
-        let debug = DebugInfoItem::deserialize_from_slice(&RAW_DEBUG).unwrap();
-        let mut reader = DebugInfoReader::new(debug.clone());
-        let mut list_info = vec![];
-        loop {
-            list_info.push(reader.next_info());
-            if list_info.last() == Some(&DebugInfo::EndOfData) {
-                break;
-            }
-        }
-        let mut builder = DebugInfoBuilder::new(debug.parameter_names.clone());
-        for info in list_info {
-            builder.add_info(&info).unwrap();
-        }
-        let debug_computed = builder.build().unwrap();
-        assert_eq!(
-            &RAW_DEBUG,
-            &(debug_computed.serialize_to_vec().unwrap()).as_slice()
-        );
-    }
 }

androscalpel_serializer/src/file_reader.rs

@@ -1,9 +1,9 @@
 //! Parser for a .dex file.
 
 use crate::{
-    CallSiteIdItem, ClassDataItem, ClassDefItem, EndianConstant, Error, FieldIdItem, HeaderItem,
-    HiddenApiFlags, HiddenapiClassDataItem, MapItemType, MapList, MethodHandleItem, MethodIdItem,
-    ProtoIdItem, Result, Serializable, StringDataItem, StringIdItem, TypeIdItem,
+    CallSiteIdItem, ClassDefItem, EndianConstant, Error, FieldIdItem, HeaderItem, MapItemType,
+    MapList, MethodHandleItem, MethodIdItem, ProtoIdItem, Result, Serializable, StringDataItem,
+    StringIdItem, TypeIdItem,
 };
 use log::{error, info, warn};
 use std::io::{Cursor, Seek, SeekFrom};
@@ -11,8 +11,6 @@ use std::sync::atomic::{AtomicBool, Ordering};
 
 #[derive(Debug)]
 pub struct DexFileReader<'a> {
-    // Ideally, this would be a Read+Seek, but Read+Seek is not thread safe, while we can
-    // internally instanciate multiple cursors on the same non mutable slice.
     data: &'a [u8],
     header: HeaderItem,
     string_ids: Vec<StringIdItem>,
@@ -30,7 +28,6 @@ pub struct DexFileReader<'a> {
     class_defs: Vec<ClassDefItem>,
     call_site_ids: Vec<CallSiteIdItem>,
     method_handles: Vec<MethodHandleItem>,
-    hiddenapi_class_data: Option<HiddenapiClassDataItem>,
     map_list: MapList,
 }
 
@@ -51,12 +48,9 @@ impl<'a> DexFileReader<'a> {
             class_defs: vec![],
             call_site_ids: vec![],
             method_handles: vec![],
-            hiddenapi_class_data: None,
             map_list: MapList { list: vec![] },
         };
-        if tmp_file.header.map_off != 0 {
-            tmp_file.map_list = tmp_file.get_struct_at_offset(tmp_file.header.map_off)?;
-        }
+        tmp_file.map_list = tmp_file.get_struct_at_offset(tmp_file.header.map_off)?;
         tmp_file.string_ids = tmp_file.get_item_list::<StringIdItem>(
             tmp_file.header.string_ids_off,
             tmp_file.header.string_ids_size,
@@ -103,15 +97,6 @@ impl<'a> DexFileReader<'a> {
             tmp_file.method_handles =
                 tmp_file.get_item_list::<MethodHandleItem>(item.offset, item.size)?
         }
-        if let Some(item) = tmp_file
-            .map_list
-            .list
-            .iter()
-            .find(|item| item.type_ == MapItemType::HiddenapiClassDataItem)
-        {
-            tmp_file.hiddenapi_class_data =
-                Some(tmp_file.get_struct_at_offset::<HiddenapiClassDataItem>(item.offset)?);
-        }
         tmp_file.sanity_check()?;
         Ok(tmp_file)
     }
@@ -120,39 +105,39 @@ impl<'a> DexFileReader<'a> {
     pub fn get_header(&self) -> &HeaderItem {
         &self.header
     }
-    /// Return the file [`StringIdItem`] list.
+    /// Retunr the file [`StringIdItem`] list.
     pub fn get_string_ids(&self) -> &[StringIdItem] {
         &self.string_ids
     }
-    /// Return the file [`TypeIdItem`] list.
+    /// Retunr the file [`TypeIdItem`] list.
     pub fn get_type_ids(&self) -> &[TypeIdItem] {
         &self.type_ids
     }
-    /// Return the file [`ProtoIdItem`] list.
+    /// Retunr the file [`ProtoIdItem`] list.
     pub fn get_proto_ids(&self) -> &[ProtoIdItem] {
         &self.proto_ids
     }
-    /// Return the file [`FieldIdItem`] list.
+    /// Retunr the file [`FieldIdItem`] list.
     pub fn get_field_ids(&self) -> &[FieldIdItem] {
         &self.field_ids
     }
-    /// Return the file [`MethodIdItem`] list.
+    /// Retunr the file [`MethodIdItem`] list.
     pub fn get_method_ids(&self) -> &[MethodIdItem] {
         &self.method_ids
     }
-    /// Return the file [`ClassDefItem`] list.
+    /// Retunr the file [`ClassDefItem`] list.
     pub fn get_class_defs(&self) -> &[ClassDefItem] {
         &self.class_defs
     }
-    /// Return the file [`CallSiteIdItem`] list.
+    /// Retunr the file [`CallSiteIdItem`] list.
     pub fn get_call_site_ids(&self) -> &[CallSiteIdItem] {
         &self.call_site_ids
     }
-    /// Return the file [`MethodHandleItem`] list.
+    /// Retunr the file [`MethodHandleItem`] list.
     pub fn get_method_handles(&self) -> &[MethodHandleItem] {
         &self.method_handles
     }
-    /// Return the file [`MapList`].
+    /// Retunr the file [`MapList`].
     pub fn get_map_list(&self) -> &MapList {
         &self.map_list
     }
@@ -237,11 +222,9 @@ impl<'a> DexFileReader<'a> {
     }
 
     fn sanity_check(&self) -> Result<()> {
-        let version = self.header.magic.version;
-        let version = (version[0] - 0x30) * 100 + (version[1] - 0x30) * 10 + (version[2] - 0x30);
-        if version > 39 {
+        if self.header.magic.version != [0x30, 0x33, 0x39] {
             warn!(
-                "Only version <= DEX 039 are currently supported, found {}",
+                "DEX 039 is the only version currently supported, found {}",
                 std::str::from_utf8(self.header.magic.version.as_slice())
                     .unwrap_or(&format!("{:x?}", self.header.magic.version))
             );
@@ -272,7 +255,7 @@ impl<'a> DexFileReader<'a> {
                 MapItemType::HeaderItem if item.offset != 0 || item.size != 1 => {
                     return Err(Error::InconsistantStruct(format!(
                         "Inconsistant Header Mapping info found in map_list: {item:x?}"
-                    )));
+                    )))
                 }
                 MapItemType::StringIdItem
                     if item.offset != self.header.string_ids_off
@@ -282,7 +265,7 @@ impl<'a> DexFileReader<'a> {
                         "Inconsistant MapList Mapping info found in map_list: {item:x?}, \
                         header.string_ids_off: 0x{:x}, header.string_ids_size: {}",
                         self.header.string_ids_off, self.header.string_ids_size
-                    )));
+                    )))
                 }
                 MapItemType::TypeIdItem
                     if item.offset != self.header.type_ids_off
@@ -292,7 +275,7 @@ impl<'a> DexFileReader<'a> {
                         "Inconsistant MapList Mapping info found in map_list: {item:x?}, \
                         header.type_ids_off: 0x{:x}, header.type_ids_size: {}",
                         self.header.type_ids_off, self.header.type_ids_size
-                    )));
+                    )))
                 }
                 MapItemType::ProtoIdItem
                     if item.offset != self.header.proto_ids_off
@@ -302,7 +285,7 @@ impl<'a> DexFileReader<'a> {
                         "Inconsistant MapList Mapping info found in map_list: {item:x?}, \
                         header.proto_ids_off: 0x{:x}, header.proto_ids_size: {}",
                         self.header.proto_ids_off, self.header.proto_ids_size
-                    )));
+                    )))
                 }
                 MapItemType::FieldIdItem
                     if item.offset != self.header.field_ids_off
@@ -312,7 +295,7 @@ impl<'a> DexFileReader<'a> {
                         "Inconsistant MapList Mapping info found in map_list: {item:x?}, \
                         header.field_ids_off: 0x{:x}, header.field_ids_size: {}",
                         self.header.field_ids_off, self.header.field_ids_size
-                    )));
+                    )))
                 }
                 MapItemType::MethodIdItem
                     if item.offset != self.header.method_ids_off
@@ -322,7 +305,7 @@ impl<'a> DexFileReader<'a> {
                         "Inconsistant MapList Mapping info found in map_list: {item:x?}, \
                         header.method_ids_off: 0x{:x}, header.method_ids_size: {}",
                         self.header.method_ids_off, self.header.method_ids_size
-                    )));
+                    )))
                 }
                 MapItemType::ClassDefItem
                     if item.offset != self.header.class_defs_off
@@ -332,14 +315,14 @@ impl<'a> DexFileReader<'a> {
                         "Inconsistant MapList Mapping info found in map_list: {item:x?}, \
                         header.class_defs_off: 0x{:x}, header.class_defs_size: {}",
                         self.header.class_defs_off, self.header.class_defs_size
-                    )));
+                    )))
                 }
                 MapItemType::MapList if item.offset != self.header.map_off || item.size != 1 => {
                     return Err(Error::InconsistantStruct(format!(
                         "Inconsistant MapList Mapping info found in map_list: {item:x?}, \
                         header.map_list_off: 0x{:x}",
                         self.header.map_off
-                    )));
+                    )))
                 }
                 /*
                 MapItemType::CallSiteIdItem => todo!(),
@@ -382,9 +365,6 @@ impl<'a> DexFileReader<'a> {
     }
 
     fn get_item_list<T: Serializable>(&self, offset: u32, size: u32) -> Result<Vec<T>> {
-        if offset == 0 {
-            return Ok(vec![]);
-        }
         let mut buffer = Cursor::new(self.data);
         buffer.seek(SeekFrom::Start(offset as u64)).map_err(|err| {
             Error::DeserializationError(format!("Failed to seek 0x{offset:x} position: {err}"))
@@ -436,45 +416,6 @@ impl<'a> DexFileReader<'a> {
         r
     }
 
-    /// Return the hiddenapi flags list for the given class.
-    ///
-    /// The list of flags is composed of one [`HiddenApiFlags`] for each static field, instance
-    /// field, direct method and virtual method of the class, in that order.
-    ///
-    /// `class_def_item_idx` if the idx of the `class_def_item`, **not** the `class_idx` (contrary
-    /// to what <https://source.android.com/docs/core/runtime/dex-format#hiddenapi-class-data-item>
-    /// says)
-    pub fn get_class_hiddenapi_flags(
-        &self,
-        class_def_item_idx: usize,
-    ) -> Result<Option<Vec<HiddenApiFlags>>> {
-        if class_def_item_idx >= self.class_defs.len() {
-            return Err(Error::InconsistantStruct(format!(
-                "idx 0x{class_def_item_idx:x} is out of bound of class_defs"
-            )));
-        }
-        let class_def = self.class_defs[class_def_item_idx];
-        if class_def.class_data_off == 0 {
-            if self.hiddenapi_class_data.is_some() {
-                return Ok(Some(vec![]));
-            } else {
-                return Ok(None);
-            }
-        }
-        let class_data = self.get_struct_at_offset::<ClassDataItem>(class_def.class_data_off)?;
-        let nb_flags = class_data.static_fields.len()
-            + class_data.instance_fields.len()
-            + class_data.direct_methods.len()
-            + class_data.virtual_methods.len();
-        if let Some(hidden_api_data) = &self.hiddenapi_class_data {
-            hidden_api_data
-                .get_flags(nb_flags, class_def_item_idx)
-                .map(Some)
-        } else {
-            Ok(None)
-        }
-    }
-
     /// Return the strings that where not referenced.
     pub fn get_not_resolved_strings(&mut self) -> Result<Vec<StringDataItem>> {
         // use `&mut self` because using this method at the same time as performing

androscalpel_serializer/src/items/code.rs

@@ -100,8 +100,8 @@ impl CodeItem {
             addresses.push(try_.start_addr);
             addresses.push(try_.start_addr + try_.insn_count as u32);
         }
-        if let Some(handlers) = &self.handlers {
-            for catch in &handlers.list {
+        for handler in &self.handlers {
+            for catch in &handler.list {
                 for EncodedTypeAddrPair {
                     addr: Uleb128(addr),
                     ..
@@ -977,9 +977,7 @@ mod test {
             }
             .serialize_to_vec()
             .unwrap(),
-            vec![
-                0x03, 0xb4, 0x01, 0x80, 0x02, 0xc7, 0x08, 0x8e, 0x02, 0xd9, 0x09, 0x87, 0x02,
-            ]
+            vec![0x03, 0xb4, 0x01, 0x80, 0x02, 0xc7, 0x08, 0x8e, 0x02, 0xd9, 0x09, 0x87, 0x02,]
         );
         assert_eq!(
             EncodedCatchHandler {
@@ -1001,9 +999,7 @@ mod test {
             }
             .serialize_to_vec()
             .unwrap(),
-            vec![
-                0x03, 0xe9, 0x46, 0x56, 0xc7, 0x08, 0x8e, 0x02, 0xd9, 0x09, 0x87, 0x02,
-            ]
+            vec![0x03, 0xe9, 0x46, 0x56, 0xc7, 0x08, 0x8e, 0x02, 0xd9, 0x09, 0x87, 0x02,]
         );
         assert_eq!(
             EncodedCatchHandler {

androscalpel_serializer/src/items/hiddenapi.rs

@@ -1,12 +1,13 @@
 //! Hidden api items.
 
+use crate as androscalpel_serializer;
 use crate::{Error, ReadSeek, Result, Serializable, Uleb128};
-use std::io::{Cursor, Seek, SeekFrom, Write};
+use std::io::{Cursor, Write};
 
 /// <https://source.android.com/docs/core/runtime/dex-format#hiddenapi-class-data-item>
-/// Hard to serialize/deserialize without additional data like the number of classes defs
+/// Hard to serialize/deserialize without additional data like the number of classes
 /// or the method/field of the classes.
-#[derive(Clone, PartialEq, Eq, Debug)]
+#[derive(Clone, PartialEq, Eq)]
 pub struct HiddenapiClassDataItem {
     //pub size: u32,
     pub data: Vec<u8>,
@@ -17,7 +18,7 @@ impl HiddenapiClassDataItem {
         self.data.len() as u32
     }
 
-    /// Return `hiddenapi_class_data_item.offsets[class_def_item_idx]`.
+    /// Return `hiddenapi_class_data_item.offsets[class_idx]`.
     ///
     /// If `0`: Either no data for this class or all API flags are zero.
     /// Else: offset from the begining of the [`HiddenapiClassDataItem`]
@@ -25,17 +26,13 @@ impl HiddenapiClassDataItem {
     ///
     /// # Warning
     ///
-    /// `class_def_item_idx` is **NOT** `class_idx` (Contrary to what
-    /// <https://source.android.com/docs/core/runtime/dex-format#hiddenapi-class-data-item> says).
-    /// `class_def_item_idx` is the index of the `class_def_item`.
-    ///
-    /// They are no check weither the `class_def_item_idx` is valid one or not.
-    /// Giving an invalid idx (like an idx >= nb class def) is UB.
-    pub fn get_offset(&self, class_def_item_idx: usize) -> Result<u32> {
-        let index = class_def_item_idx * 4;
-        if self.data.len() <= index {
+    /// They are no check weither the `class_idx` is valid one or not.
+    /// Giving an invalid idx (like an idx >= nb class) is UB.
+    pub fn get_offset(&self, class_idx: u32) -> Result<u32> {
+        let index = (class_idx as usize) * 4;
+        if self.data.len() < index - 4 {
             Err(Error::InconsistantStruct(format!(
-                "class index 0x{class_def_item_idx:x} out of bound of HiddenapiClassDataItem data"
+                "class index 0x{class_idx:x} out of bound of HiddenapiClassDataItem data"
             )))
         } else {
             u32::deserialize_from_slice(&self.data[index..])
@@ -47,7 +44,7 @@ impl HiddenapiClassDataItem {
     /// # Warning
     ///
     /// They are no check weither the `nb_class` is valid one or not.
-    /// Giving an invalid `nb_class` is UB.
+    /// Giving an invalid `nb_class`.
     pub fn get_offsets(&self, nb_class: u32) -> Result<Vec<u32>> {
         let mut offsets = vec![];
         let mut buffer = Cursor::new(self.data.as_slice());
@@ -63,37 +60,19 @@ impl HiddenapiClassDataItem {
     ///
     /// # Warning
     ///
-    /// `class_def_item_idx` is **NOT** `class_idx` (Contrary to what
-    /// <https://source.android.com/docs/core/runtime/dex-format#hiddenapi-class-data-item> says).
-    /// `class_def_item_idx` is the index of the `class_def_item`.
-    ///
-    /// They are no check weither the `nb_flags` or `class_def_item_idx`
+    /// They are no check weither the `nb_flags` or `offset`
     /// are valid. Providing invalid values is UB.
-    pub fn get_flags(
-        &self,
-        nb_flags: usize,
-        class_def_item_idx: usize,
-    ) -> Result<Vec<HiddenApiFlags>> {
-        let offset = self.get_offset(class_def_item_idx)?;
+    pub fn get_flags(&self, nb_flags: usize, offset: u32) -> Result<Vec<Uleb128>> {
         if offset == 0 {
-            Ok(vec![HiddenApiFlags::DEFAULT; nb_flags])
+            Ok(vec![Uleb128(0); nb_flags])
         } else if offset < 4 {
             // < 8 is almost certainly false
             panic!()
         } else {
             let mut buffer = Cursor::new(self.data.as_slice());
-            buffer
-                .seek(SeekFrom::Start(offset as u64 - 4))
-                .map_err(|_| {
-                    Error::InconsistantStruct(format!(
-                        "{offset} if out of data bound for HiddenApiClassData"
-                    ))
-                })?;
             let mut flags = vec![];
             for _ in 0..nb_flags {
-                flags.push(HiddenApiFlags::from_uleb128(Uleb128::deserialize(
-                    &mut buffer,
-                )?));
+                flags.push(Uleb128::deserialize(&mut buffer)?);
             }
             Ok(flags)
         }
@@ -123,139 +102,40 @@ impl Serializable for HiddenapiClassDataItem {
     }
 }
 
-#[derive(Clone, PartialEq, Eq)]
-pub struct HiddenApiFlags {
-    pub value: HiddenApiValue,
-    pub domain_api: HiddenApiDomain,
-}
-
-impl HiddenApiFlags {
-    pub const DEFAULT: Self = Self {
-        value: HiddenApiValue::Whitelist,
-        domain_api: HiddenApiDomain {
-            is_test_api: false,
-            is_core_platform_api: false,
-            unknown_flags: 0,
-        },
-    };
-
-    pub fn from_uleb128(Uleb128(val): Uleb128) -> Self {
-        Self {
-            value: HiddenApiValue::from_u32(val),
-            domain_api: HiddenApiDomain::from_u32(val),
-        }
-    }
-
-    pub fn to_uleb128(&self) -> Uleb128 {
-        Uleb128(self.value.to_u32() | self.domain_api.to_u32())
-    }
-}
-
-/// Flags for hidden api flag value
-#[derive(Debug, PartialEq, Eq, Copy, Clone)]
-pub enum HiddenApiValue {
+/// Flags for hidden api
+#[derive(Serializable, Debug, PartialEq, Eq, Copy, Clone)]
+#[prefix_type(Uleb128)]
+pub enum HiddenApiFlag {
     /// Interfaces that can be freely used and are supported as
     /// part of the officially documented Android framework Package Index
+    #[prefix(Uleb128(0x00))]
     Whitelist,
     /// Non-SDK interfaces that can be used regardless of the
     /// application's target API level
+    #[prefix(Uleb128(0x01))]
     Greylist,
     /// Non-SDK interfaces that cannot be used regardless of the
     /// application's target API level. Accessing one of these
     /// interfaces causes a runtime error.
+    #[prefix(Uleb128(0x02))]
     Blacklist,
     /// Non-SDK interfaces that can be used for Android 8.x and
-    /// below unless they are restricted (targetSdkVersion <= 27 (O_MR1)).
+    /// below unless they are restricted.
+    #[prefix(Uleb128(0x03))]
     GreylistMaxO,
     /// Non-SDK interfaces that can be used for Android 9.x unless
     /// they are restricted.
+    #[prefix(Uleb128(0x04))]
     GreylistMaxP,
     /// Non-SDK interfaces that can be used for Android 10.x unless
     /// they are restricted.
+    #[prefix(Uleb128(0x05))]
     GreylistMaxQ,
     /// Non-SDK interfaces that can be used for Android 11.x unless
     /// they are restricted.
+    #[prefix(Uleb128(0x06))]
     GreylistMaxR,
-    GreylistMaxS,
     /// Unknown flag, either an error or this crate is out of date.
-    Unknwon(u8),
-}
-
-impl HiddenApiValue {
-    pub const MASK: u32 = 0b111;
-    pub fn from_u32(flags: u32) -> Self {
-        match flags & Self::MASK {
-            0x00 => Self::Whitelist,
-            0x01 => Self::Greylist,
-            0x02 => Self::Blacklist,
-            0x03 => Self::GreylistMaxO,
-            0x04 => Self::GreylistMaxP,
-            0x05 => Self::GreylistMaxQ,
-            0x06 => Self::GreylistMaxR,
-            0x07 => Self::GreylistMaxS,
-            other => Self::Unknwon(other as u8),
-        }
-    }
-    pub fn to_u32(&self) -> u32 {
-        match self {
-            Self::Whitelist => 0x00,
-            Self::Greylist => 0x01,
-            Self::Blacklist => 0x02,
-            Self::GreylistMaxO => 0x03,
-            Self::GreylistMaxP => 0x04,
-            Self::GreylistMaxQ => 0x05,
-            Self::GreylistMaxR => 0x06,
-            Self::GreylistMaxS => 0x07,
-            Self::Unknwon(other) => {
-                if (0b1111_1000 & *other) != 0 {
-                    panic!("HiddenApiValue is encoded on 3 bits but found value {other}");
-                }
-                *other as u32
-            }
-        }
-    }
-}
-
-/// Flags for hidden api flag domain
-#[derive(Debug, PartialEq, Eq, Copy, Clone)]
-pub struct HiddenApiDomain {
-    pub is_core_platform_api: bool,
-    pub is_test_api: bool,
-    pub unknown_flags: u32,
-}
-
-impl HiddenApiDomain {
-    pub const CORE_PLATFORM_API_FLAG: u32 = 0x08;
-    pub const TEST_API_FLAG: u32 = 0x10;
-    pub fn from_u32(flags: u32) -> Self {
-        let flags = flags & !HiddenApiValue::MASK;
-        Self {
-            is_core_platform_api: (flags & Self::CORE_PLATFORM_API_FLAG) != 0,
-            is_test_api: (flags & Self::TEST_API_FLAG) != 0,
-            unknown_flags: flags & (!Self::CORE_PLATFORM_API_FLAG) & (!Self::TEST_API_FLAG),
-        }
-    }
-    pub fn to_u32(&self) -> u32 {
-        if ((0b111 | Self::CORE_PLATFORM_API_FLAG | Self::CORE_PLATFORM_API_FLAG)
-            & self.unknown_flags)
-            != 0
-        {
-            panic!(
-                "The first 5 bits of HiddenApiDomain are reserved and \
-                        shoud be set to 0, but value 0x{:x} was found",
-                self.unknown_flags
-            );
-        }
-        self.unknown_flags
-            | if self.is_core_platform_api {
-                Self::CORE_PLATFORM_API_FLAG
-            } else {
-                0
-            }
-            | if self.is_test_api {
-                Self::TEST_API_FLAG
-            } else {
-                0
-            }
-    }
+    #[default_variant]
+    Unknwon(Uleb128),
 }

androscalpel_serializer/src/items/map.rs

@@ -71,36 +71,6 @@ pub enum MapItemType {
     UnkownType(u16),
 }
 
-impl MapItemType {
-    /// If data of this type is stored in the data section
-    pub fn is_data_section_type(&self) -> bool {
-        match self {
-            Self::HeaderItem => false,
-            Self::StringIdItem => false,
-            Self::TypeIdItem => false,
-            Self::ProtoIdItem => false,
-            Self::FieldIdItem => false,
-            Self::MethodIdItem => false,
-            Self::ClassDefItem => false,
-            Self::CallSiteIdItem => true,
-            Self::MethodHandleItem => true,
-            Self::MapList => true,
-            Self::TypeList => true,
-            Self::AnnotationSetRefList => true,
-            Self::AnnotationSetItem => true,
-            Self::ClassDataItem => true,
-            Self::CodeItem => true,
-            Self::StringDataItem => true,
-            Self::DebugInfoItem => true,
-            Self::AnnotationItem => true,
-            Self::EncodedArrayItem => true,
-            Self::AnnotationsDirectoryItem => true,
-            Self::HiddenapiClassDataItem => true,
-            Self::UnkownType(_) => true, // Most likely
-        }
-    }
-}
-
 impl MapList {
     /// The size field of a MapList.
     pub fn size_field(&self) -> u32 {

androscalpel_serializer_derive/Cargo.toml

@@ -1,8 +1,7 @@
 [package]
 name = "androscalpel_serializer_derive"
 version = "0.1.0"
-edition = "2024"
-license = "GPL-3.0-or-later"
+edition = "2021"
 
 [lib]
 proc-macro = true

androscalpel_serializer_derive/src/lib.rs

@@ -3,8 +3,8 @@ use quote::{format_ident, quote, quote_spanned};
 use syn::parse::{Parse, ParseStream};
 use syn::spanned::Spanned;
 use syn::{
-    Attribute, Data, DeriveInput, Field, Fields, Ident, Index, Meta, MetaList, Token, Type,
-    Variant, parse_macro_input,
+    parse_macro_input, Attribute, Data, DeriveInput, Field, Fields, Ident, Index, Meta, MetaList,
+    Token, Type, Variant,
 };
 
 /// Derive the type Serializable.
@@ -55,7 +55,7 @@ use syn::{
 /// An enum can define ONE variant as the default variant. This variant is a catch all,
 /// and MUST be have named field or unnamed field (no unit variant!) and:
 /// - The first field of named fields variant must be named `prefix` and of the type
-///     `prefix_type`.
+/// `prefix_type`.
 /// - The first field of unnamed fields variant must be of the type `prefix_type`.
 ///
 /// The first field of the default variant store the prefix of the variant.
@@ -318,25 +318,19 @@ fn get_enum_match(variant: &Variant) -> TokenStream {
 /// for a specific field `f` accessible using `field_ref`.
 fn get_implem_size_for_field(f: &Field, field_ref: TokenStream) -> TokenStream {
     let params = ParamsField::parse(&f.attrs);
-    let prefix_stream = match params.prefix {
-        Some(PrefixParamsField(ref stream)) => {
-            quote_spanned! { f.span() =>
-                #stream.iter().collect::<Vec<_>>().len()
-            }
-        }
-        _ => {
-            quote_spanned! { f.span() => 0 }
+    let prefix_stream = if let Some(PrefixParamsField(ref stream)) = params.prefix {
+        quote_spanned! { f.span() =>
+            #stream.iter().collect::<Vec<_>>().len()
         }
+    } else {
+        quote_spanned! { f.span() => 0 }
     };
-    let suffix_stream = match params.suffix {
-        Some(SuffixParamsField(ref stream)) => {
-            quote_spanned! { f.span() =>
-                #stream.iter().collect::<Vec<_>>().len()
-            }
-        }
-        _ => {
-            quote_spanned! { f.span() => 0 }
+    let suffix_stream = if let Some(SuffixParamsField(ref stream)) = params.suffix {
+        quote_spanned! { f.span() =>
+            #stream.iter().collect::<Vec<_>>().len()
         }
+    } else {
+        quote_spanned! { f.span() => 0 }
     };
     let main_stream = match (&f.ty, params) {
         (
@@ -442,29 +436,23 @@ fn get_implem_size(data: &Data, params: &ParamsStruct) -> TokenStream {
 fn get_implem_serialize_for_field(f: &Field, field_ref: TokenStream) -> TokenStream {
     let params = ParamsField::parse(&f.attrs);
     // TODO: Improve error handling
-    let prefix_stream = match params.prefix {
-        Some(PrefixParamsField(ref stream)) => {
-            quote_spanned! { f.span() =>
-                for byte in #stream {
-                    <u8 as androscalpel_serializer::Serializable>::serialize(&byte, output)?;
-                }
+    let prefix_stream = if let Some(PrefixParamsField(ref stream)) = params.prefix {
+        quote_spanned! { f.span() =>
+            for byte in #stream {
+                <u8 as androscalpel_serializer::Serializable>::serialize(&byte, output)?;
             }
         }
-        _ => {
-            quote_spanned! { f.span() => }
-        }
+    } else {
+        quote_spanned! { f.span() => }
     };
-    let suffix_stream = match params.suffix {
-        Some(SuffixParamsField(ref stream)) => {
-            quote_spanned! { f.span() =>
-                for byte in #stream {
-                    <u8 as androscalpel_serializer::Serializable>::serialize(&byte, output)?;
-                }
+    let suffix_stream = if let Some(SuffixParamsField(ref stream)) = params.suffix {
+        quote_spanned! { f.span() =>
+            for byte in #stream {
+                <u8 as androscalpel_serializer::Serializable>::serialize(&byte, output)?;
             }
         }
-        _ => {
-            quote_spanned! { f.span() => }
-        }
+    } else {
+        quote_spanned! { f.span() => }
     };
     let main_stream = match (&f.ty, params) {
         (
@@ -550,6 +538,7 @@ fn get_implem_serialize(data: &Data, params: &ParamsStruct) -> TokenStream {
                                 "The first field of a named field variant must be named `prefix` and of the type of the prefix of the enum"
                             );
                         }
+                        
                         let recurse = fields.named.iter().map(|f| {
                             let name = &f.ident;
                             get_implem_serialize_for_field(f, quote! { *#name })
@@ -598,37 +587,31 @@ fn get_implem_deserialize_for_field(f: &Field, field_ref: TokenStream) -> TokenS
     let params = ParamsField::parse(&f.attrs);
     let ty = &f.ty;
     // TODO: Improve error handling
-    let prefix_stream = match params.prefix {
-        Some(PrefixParamsField(ref stream)) => {
-            quote_spanned! { f.span() =>
-                for byte in #stream {
-                    if <u8 as androscalpel_serializer::Serializable>::deserialize(input)? != byte {
-                        return Err(androscalpel_serializer::Error::DeserializationError(
-                            "Prefix do not match #stream".into()
-                        ));
-                    }
+    let prefix_stream = if let Some(PrefixParamsField(ref stream)) = params.prefix {
+        quote_spanned! { f.span() =>
+            for byte in #stream {
+                if <u8 as androscalpel_serializer::Serializable>::deserialize(input)? != byte {
+                    return Err(androscalpel_serializer::Error::DeserializationError(
+                        "Prefix do not match #stream".into()
+                    ));
                 }
             }
         }
-        _ => {
-            quote_spanned! { f.span() => }
-        }
+    } else {
+        quote_spanned! { f.span() => }
     };
-    let suffix_stream = match params.suffix {
-        Some(SuffixParamsField(ref stream)) => {
-            quote_spanned! { f.span() =>
-                for byte in #stream {
-                    if <u8 as androscalpel_serializer::Serializable>::deserialize(input)? != byte {
-                        return Err(androscalpel_serializer::Error::DeserializationError(
-                            "Suffix do not match #stream".into()
-                        ));
-                    }
+    let suffix_stream = if let Some(SuffixParamsField(ref stream)) = params.suffix {
+        quote_spanned! { f.span() =>
+            for byte in #stream {
+                if <u8 as androscalpel_serializer::Serializable>::deserialize(input)? != byte {
+                    return Err(androscalpel_serializer::Error::DeserializationError(
+                        "Suffix do not match #stream".into()
+                    ));
                 }
             }
         }
-        _ => {
-            quote_spanned! { f.span() => }
-        }
+    } else {
+        quote_spanned! { f.span() => }
     };
     match (ty, params) {
         (

apk_frauder/Cargo.toml

@@ -1,14 +1,11 @@
 [package]
 name = "apk_frauder"
 version = "0.1.0"
-edition = "2024"
-license = "GPL-3.0-or-later"
+edition = "2021"
 
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dependencies]
 androscalpel_serializer = { version = "0.1.0", path = "../androscalpel_serializer" }
-anyhow = "1.0.98"
 flate2 = { version = "1.0.28", features = ["rust_backend"] }
-log = "0.4.25"
 rand = "0.8.5"

apk_frauder/TODO.md

@@ -1,3 +0,0 @@
-- replace panic/unwrap/expect with results
-- tests
-- write zip with data descriptor

apk_frauder/src/data_descriptor.rs

@@ -1,117 +0,0 @@
-use crate::Signature;
-use androscalpel_serializer::{ReadSeek, Result, Serializable};
-use std::io::{SeekFrom, Write};
-
-#[derive(Debug, Clone, PartialEq, Eq)]
-pub enum DataDescriptor {
-    Zip32(DataDescriptor32),
-    Zip64(DataDescriptor64),
-}
-
-/// DataDescriptor for 32bit zip format.
-#[derive(Debug, Clone, PartialEq, Eq)]
-pub struct DataDescriptor32 {
-    // signature: Option<Signature(0x08074b50)>
-    pub crc_32: u32,
-    pub compressed_size: u32,
-    pub uncompressed_size: u32,
-    /// For reproducibility. This should be true when writting a zip, but can be false when
-    /// reading one. If we want to stick to the original, we need to keep this information.
-    pub use_signature: bool,
-}
-
-/// DataDescriptor for 64bit zip format.
-#[derive(Debug, Clone, PartialEq, Eq)]
-pub struct DataDescriptor64 {
-    // signature: Option<Signature(0x08074b50)>
-    pub crc_32: u32,
-    pub compressed_size: u64,
-    pub uncompressed_size: u64,
-    /// For reproducibility. This should be true when writting a zip, but can be false when
-    /// reading one. If we want to stick to the original, we need to keep this information.
-    pub use_signature: bool,
-}
-
-impl DataDescriptor32 {
-    const SIGNATURE: Signature = Signature(0x08074b50);
-}
-
-impl Serializable for DataDescriptor32 {
-    fn serialize(&self, output: &mut dyn Write) -> Result<()> {
-        if self.use_signature {
-            Self::SIGNATURE.serialize(output)?;
-        }
-        self.crc_32.serialize(output)?;
-        self.compressed_size.serialize(output)?;
-        self.uncompressed_size.serialize(output)
-    }
-
-    fn deserialize(input: &mut dyn ReadSeek) -> Result<Self> {
-        let pos = input.stream_position().unwrap(); //TODO
-        let signature = Signature::deserialize(input)?;
-        let use_signature = if signature != Self::SIGNATURE {
-            input.seek(SeekFrom::Start(pos)).unwrap(); //TODO
-            false
-        } else {
-            true
-        };
-        let crc_32 = u32::deserialize(input)?;
-        let compressed_size = u32::deserialize(input)?;
-        let uncompressed_size = u32::deserialize(input)?;
-        Ok(Self {
-            crc_32,
-            compressed_size,
-            uncompressed_size,
-            use_signature,
-        })
-    }
-    fn size(&self) -> usize {
-        12 + if self.use_signature {
-            Self::SIGNATURE.size()
-        } else {
-            0
-        }
-    }
-}
-
-impl DataDescriptor64 {
-    const SIGNATURE: Signature = Signature(0x08074b50);
-}
-
-impl Serializable for DataDescriptor64 {
-    fn serialize(&self, output: &mut dyn Write) -> Result<()> {
-        if self.use_signature {
-            Self::SIGNATURE.serialize(output)?;
-        }
-        self.crc_32.serialize(output)?;
-        self.compressed_size.serialize(output)?;
-        self.uncompressed_size.serialize(output)
-    }
-
-    fn deserialize(input: &mut dyn ReadSeek) -> Result<Self> {
-        let pos = input.stream_position().unwrap(); //TODO
-        let signature = Signature::deserialize(input)?;
-        let use_signature = if signature != Self::SIGNATURE {
-            input.seek(SeekFrom::Start(pos)).unwrap(); //TODO
-            false
-        } else {
-            true
-        };
-        let crc_32 = u32::deserialize(input)?;
-        let compressed_size = u64::deserialize(input)?;
-        let uncompressed_size = u64::deserialize(input)?;
-        Ok(Self {
-            crc_32,
-            compressed_size,
-            uncompressed_size,
-            use_signature,
-        })
-    }
-    fn size(&self) -> usize {
-        20 + if self.use_signature {
-            Self::SIGNATURE.size()
-        } else {
-            0
-        }
-    }
-}

apk_frauder/src/file_header.rs

@@ -1,4 +1,3 @@
-use log::warn;
 use std::io::{SeekFrom, Write};
 
 use crate::compression::CompressionMethod;
@@ -123,7 +122,7 @@ impl Serializable for FileHeader {
             let field = ExtraField::deserialize(input);
 
             if let Err(err) = field {
-                warn!(
+                println!(
                     "Failed to parsed extra field in {}: {err:?}",
                     header.get_name()
                 );
@@ -136,15 +135,8 @@ impl Serializable for FileHeader {
             }
         }
         if extra_size_read > extra_field_length as usize {
-            let last_extra = header.extra_field.pop().unwrap();
-            let size = last_extra.size();
-            warn!(
-                "Failed to parse last extra field in {}, last field ({} bytes long) is {} bytes too long",
-                header.get_name(),
-                size,
-                extra_size_read - extra_field_length as usize
-            );
-            //warn!("Forgetting last extra field: {:?}", last_extra);
+            println!("Failed to parse last extra field in {}", header.get_name());
+            let size = header.extra_field.pop().unwrap().size();
             input.seek(SeekFrom::Current(-(size as i64))).unwrap();
         }
         let mut extra_size_read = input.stream_position().unwrap() - extra_field_off;

apk_frauder/src/lib.rs

@@ -1,6 +1,4 @@
 use androscalpel_serializer::Serializable;
-use anyhow::{Context, Result};
-use std::collections::HashMap;
 use std::env;
 use std::fs;
 use std::fs::File;
@@ -11,7 +9,6 @@ use std::process::Command;
 pub mod apk_signing_block;
 pub mod compression;
 mod cp437;
-pub mod data_descriptor;
 pub mod end_of_central_directory;
 pub mod error;
 pub mod extra_fields;
@@ -20,7 +17,6 @@ pub mod local_file_header;
 pub mod zip_reader;
 pub mod zip_writer;
 
-use data_descriptor::DataDescriptor;
 use extra_fields::{ExtraField, Zip64ExtraField};
 use file_header::FileHeader;
 use local_file_header::LocalFileHeader;
@@ -72,8 +68,8 @@ pub mod external_file_attributes {
 pub struct FileInfo {
     pub local_header: LocalFileHeader,
     pub header: FileHeader,
-    pub data_descriptor: Option<DataDescriptor>,
 }
+// TODO: support data descriptor (MASK_USE_DATA_DESCRIPTOR)
 
 impl FileInfo {
     pub fn get_name(&self) -> String {
@@ -143,42 +139,55 @@ impl FileInfo {
     }
 }
 
-/// Replace the dex files in an apk and strip the old signature.
-#[allow(clippy::too_many_arguments)]
-pub fn replace_dex_unsigned(
+/// Replace the dex files a an apk an resigned the apk.
+///
+/// # Warning
+///
+/// For now, only jks keystore are allowed.
+///
+/// The `zipalign` and `apksigner` args allow to use a specific version of the
+/// tools instead of the one in the PATH (if it even exist)
+pub fn replace_dex(
     apk: impl AsRef<Path>,
     dst: impl AsRef<Path>,
     dexfiles: &mut [impl Read + Seek],
-    additionnal_files: Option<HashMap<String, Option<impl Read + Seek>>>,
-) -> Result<()> {
-    let file = File::open(&apk).with_context(|| {
-        format!(
-            "failed to open file {}",
-            apk.as_ref().to_str().unwrap_or("")
-        )
-    })?;
-    let mut apk = ZipFileReader::new(file)?;
-    let file = File::create(&dst).context("failed to create file")?;
+    keystore: impl AsRef<Path>, // TODO enum for handling p11 and generating a random cert
+    // `keytool -genkey -v -keystore KEYSTORE.jks -keyalg RSA -keysize
+    // 2048 -validity 10000 -alias ALIAS`
+    zipalign: Option<impl AsRef<Path>>,
+    apksigner: Option<impl AsRef<Path>>,
+) {
+    let zipalign = if let Some(path) = &zipalign {
+        path.as_ref().as_os_str()
+    } else {
+        "zipalign".as_ref()
+    };
+    let apksigner = if let Some(path) = &apksigner {
+        path.as_ref().as_os_str()
+    } else {
+        "apksigner".as_ref()
+    };
+    let tmp_dir = env::temp_dir().join(format!("apk_frauder_{:x}", rand::random::<u128>()));
+    let unaligned_path = tmp_dir.join("stripped.apk");
+    let aligned_path = tmp_dir.join("aligned.apk");
+    fs::create_dir_all(&tmp_dir).expect("Failed to create temporary directory");
+
+    let file = File::open(apk).expect("failed to open file");
+    let mut apk = ZipFileReader::new(file);
+    let file = File::create(&unaligned_path).expect("failed to create file");
     let mut apk_out = ZipFileWriter::new(file, apk.zip64_end_of_central_directory.clone());
 
     let mut file_info_ref = (*apk
         .get_classes_file_info()
         .first()
-        .context("No dex file found in apk")?)
+        .expect("No dex file found in apk"))
     .clone();
 
     apk.unlink_signature_files();
     apk.unlink_bytecode_files();
-
-    if let Some(additionnal_files) = &additionnal_files {
-        apk.files
-            .retain(|file| additionnal_files.get(&file.get_name()).is_none());
-    }
-
     for f in apk.files.clone() {
         apk_out.insert_file_from_zip(f, &mut apk);
     }
-
     for (i, mut dex) in dexfiles.iter_mut().enumerate() {
         if i == 0 {
             file_info_ref.set_name("classes.dex");
@@ -189,63 +198,9 @@ pub fn replace_dex_unsigned(
             &mut dex,
             file_info_ref.header.clone(),
             Some(file_info_ref.local_header.clone()),
-            file_info_ref.data_descriptor.clone(),
         );
     }
-
-    if let Some(mut additionnal_files) = additionnal_files {
-        for (name, data) in &mut additionnal_files {
-            file_info_ref.set_name(name);
-            if let Some(data) = data.as_mut() {
-                apk_out.insert_file(
-                    data,
-                    file_info_ref.header.clone(),
-                    Some(file_info_ref.local_header.clone()),
-                    file_info_ref.data_descriptor.clone(),
-                );
-            }
-        }
-    }
     apk_out.write_central_directory();
-    Ok(())
-}
-
-/// Replace the dex files in an apk an resigned the apk.
-///
-/// # Warning
-///
-/// For now, only jks keystore are allowed.
-///
-/// The `zipalign` and `apksigner` args allow to use a specific version of the
-/// tools instead of the one in the PATH (if it even exist)
-#[allow(clippy::too_many_arguments)]
-pub fn replace_dex(
-    apk: impl AsRef<Path>,
-    dst: impl AsRef<Path>,
-    dexfiles: &mut [impl Read + Seek],
-    keystore: impl AsRef<Path>, // TODO enum for handling p11 and generating a random cert
-    // `keytool -genkey -v -keystore KEYSTORE.jks -keyalg RSA -keysize
-    // 2048 -validity 10000 -alias ALIAS`
-    zipalign: Option<impl AsRef<Path>>,
-    apksigner: Option<impl AsRef<Path>>,
-    keypassword: Option<&str>,
-    additionnal_files: Option<HashMap<String, Option<impl Read + Seek>>>,
-) -> Result<()> {
-    let zipalign = match &zipalign {
-        Some(path) => path.as_ref().as_os_str(),
-        _ => "zipalign".as_ref(),
-    };
-    let apksigner = match &apksigner {
-        Some(path) => path.as_ref().as_os_str(),
-        _ => "apksigner".as_ref(),
-    };
-    let tmp_dir = env::temp_dir().join(format!("apk_frauder_{:x}", rand::random::<u128>()));
-    let unaligned_path = tmp_dir.join("stripped.apk");
-    let aligned_path = tmp_dir.join("aligned.apk");
-    fs::create_dir_all(&tmp_dir).context("Failed to create temporary directory")?;
-
-    replace_dex_unsigned(apk, &unaligned_path, dexfiles, additionnal_files)?;
-
     // TODO: we can probably do that ourself an spare ourself the trouble of finding zipalign
     Command::new(zipalign)
         .arg("-v")
@@ -254,23 +209,16 @@ pub fn replace_dex(
         .arg(unaligned_path.as_os_str())
         .arg(aligned_path.as_os_str())
         .status()
-        .context("Failed to run zipalign")?;
+        .unwrap();
 
-    let mut cmd = Command::new(apksigner);
-    let cmd = cmd
+    Command::new(apksigner)
         .arg("sign")
         .arg("--ks")
         .arg(keystore.as_ref().as_os_str())
         .arg("--out")
         .arg(dst.as_ref().as_os_str())
-        .arg("--in")
-        .arg(aligned_path.as_os_str());
-    let cmd = if let Some(pwd) = keypassword {
-        cmd.arg("--ks-pass").arg(format!("pass:{pwd}"))
-    } else {
-        cmd
-    };
-    cmd.status().context("Failled to run apksigne")?;
-    fs::remove_dir_all(tmp_dir).context("Failled to remove tmp dir")?;
-    Ok(())
+        .arg(aligned_path.as_os_str())
+        .status()
+        .unwrap();
+    fs::remove_dir_all(tmp_dir).expect("Failled to remove tmp dir");
 }

apk_frauder/src/local_file_header.rs

@@ -1,7 +1,5 @@
 use std::io::{SeekFrom, Write};
 
-use log::warn;
-
 use crate::compression::CompressionMethod;
 use crate::error::Error;
 use crate::extra_fields::{ExtraField, GenericExtraField, Zip64ExtraField};
@@ -101,9 +99,9 @@ impl Serializable for LocalFileHeader {
             let field = ExtraField::deserialize(input);
 
             if let Err(err) = field {
-                warn!(
+                println!(
                     "Failed to parsed extra field in {}: {err:?}",
-                    header.get_name(),
+                    header.get_name()
                 );
                 input.seek(SeekFrom::Start(field_off)).unwrap();
                 break;
@@ -113,32 +111,25 @@ impl Serializable for LocalFileHeader {
                 header.extra_field.push(field);
             }
         }
-        let mut failed_last_extra_field_data = None;
+        let mut failed_last_extra_field = false;
         if extra_size_read > extra_field_length as usize {
             //println!("Failed to parse last extra field in {}", header.get_name());
+            failed_last_extra_field = true;
             let size = header.extra_field.pop().unwrap().size();
             input.seek(SeekFrom::Current(-(size as i64))).unwrap();
-            failed_last_extra_field_data =
-                Some((size, extra_size_read - extra_field_length as usize));
         }
         let mut extra_size_read = input.stream_position().unwrap() - extra_field_off;
         while extra_size_read < extra_field_length as u64 {
             header.malformed_extra_field.push(u8::deserialize(input)?);
             extra_size_read += 1;
         }
-        if let Some((size, size_diff)) = failed_last_extra_field_data {
-            // If it is not padding from zipalign
-            if header.malformed_extra_field != vec![0]
-                && header.malformed_extra_field != vec![0, 0]
-                && header.malformed_extra_field != vec![0, 0, 0]
-            {
-                warn!(
-                    "Failed to parse last extra field in {}, last field ({} bytes long) is {} bytes too long", 
-                    header.get_name(), 
-                    size, 
-                    size_diff
-                );
-            }
+        // If it is not padding from zipalign
+        if failed_last_extra_field
+            && header.malformed_extra_field != vec![0]
+            && header.malformed_extra_field != vec![0, 0]
+            && header.malformed_extra_field != vec![0, 0, 0]
+        {
+            println!("Failed to parse last extra field in {}", header.get_name());
         }
         //input.seek(SeekFrom::Start(end_of_extra_field)).unwrap();
         for field in &mut header.extra_field {

apk_frauder/src/main.rs

@@ -1,11 +1,8 @@
-use apk_frauder::ZipFileReader;
-//use std::collections::HashMap;
-//use std::env;
+use std::env;
 use std::fs::File;
-//use std::io::Cursor;
 
+/*
 fn main() {
-    /*
     apk_frauder::replace_dex(
         "app-release.apk",
         "app-instrumented.apk",
@@ -19,9 +16,78 @@ fn main() {
             "{}/Android/Sdk/build-tools/34.0.0/apksigner",
             env::var("HOME").expect("$HOME not set")
         )),
-        None::<HashMap<String, Option<Cursor<&[u8]>>>>,
-    );*/
-    let file = File::open("zagruski.apk").unwrap();
-    let reader = ZipFileReader::new(file).unwrap();
-    println!("{:#?}", &reader.files[..2]);
+    );
+}*/
+
+fn main() {
+    let args: Vec<String> = env::args().collect();
+    if args.len() != 3 {
+        panic!("Usage: {} <folder to zip> <header file>", args[0]);
+    }
+    let folder = &args[1];
+    let header_file = &args[2];
+    let archive_name = format!("{folder}.zip");
+    let file = File::create(archive_name).expect("failed to create file");
+    let mut zip = apk_frauder::zip_writer::ZipFileWriter::new(file, None);
+    zip.insert_untracked_file(
+        &mut File::open(header_file)
+            .unwrap_or_else(|_| panic!("failed to open file {header_file}")),
+    );
+    insert_to_zip(std::path::Path::new(folder), &mut zip, None);
+    zip.write_central_directory();
+}
+
+fn insert_to_zip(
+    file: &std::path::Path,
+    zip: &mut apk_frauder::zip_writer::ZipFileWriter<File>,
+    root: Option<&std::path::Path>,
+) {
+    let root = if let Some(root) = root { root } else { file };
+    let internal_path =
+        &std::path::Path::new(root.file_name().unwrap()).join(file.strip_prefix(root).unwrap());
+    if file.is_dir() {
+        zip.insert_file(
+            &mut std::io::Cursor::new(vec![]),
+            apk_frauder::file_header::FileHeader {
+                external_file_attributes: apk_frauder::external_file_attributes::DIR_FILE
+                    | apk_frauder::external_file_attributes::PERM_UR
+                    | apk_frauder::external_file_attributes::PERM_UW
+                    | apk_frauder::external_file_attributes::PERM_GR
+                    | apk_frauder::external_file_attributes::PERM_OR
+                    | apk_frauder::external_file_attributes::PERM_UX
+                    | apk_frauder::external_file_attributes::PERM_GX
+                    | apk_frauder::external_file_attributes::PERM_OX,
+                ..apk_frauder::file_header::FileHeader::new_default(internal_path.to_str().unwrap())
+            },
+            None,
+        );
+        for entry in std::fs::read_dir(file).unwrap() {
+            let entry = entry.unwrap();
+            let path = entry.path();
+            insert_to_zip(path.as_ref(), zip, Some(root));
+        }
+    } else {
+        //use std::os::unix::fs::PermissionsExt;
+        let mut f = File::open(file)
+            .unwrap_or_else(|_| panic!("failed to open file {}", file.to_str().unwrap()));
+        //let metadata = f.metadata().unwrap();
+        //let permissions = metadata.permissions();
+        //let external_file_attributes = permissions.mode(); does not work?
+        let external_file_attributes = apk_frauder::external_file_attributes::REGULAR_FILE
+            | apk_frauder::external_file_attributes::PERM_UR
+            | apk_frauder::external_file_attributes::PERM_UW
+            | apk_frauder::external_file_attributes::PERM_GR
+            | apk_frauder::external_file_attributes::PERM_OR
+            | apk_frauder::external_file_attributes::PERM_UX
+            | apk_frauder::external_file_attributes::PERM_GX
+            | apk_frauder::external_file_attributes::PERM_OX;
+        zip.insert_file(
+            &mut f,
+            apk_frauder::file_header::FileHeader {
+                external_file_attributes,
+                ..apk_frauder::file_header::FileHeader::new_default(internal_path.to_str().unwrap())
+            },
+            None,
+        );
+    }
 }

apk_frauder/src/zip_reader.rs

@@ -1,20 +1,13 @@
+use std::io::{Read, Seek, SeekFrom};
+
 use crate::{
-    FileHeader, FileInfo, LocalFileHeader, Signature,
-    apk_signing_block::ApkSigningBlock,
-    apk_signing_block::Magic,
-    compression::CompressionMethod,
-    data_descriptor::{DataDescriptor, DataDescriptor32, DataDescriptor64},
+    apk_signing_block::ApkSigningBlock, apk_signing_block::Magic,
     end_of_central_directory::EndCentralDirectory,
     end_of_central_directory::Zip64EndCentralDirectory,
-    end_of_central_directory::Zip64EndCentralDirectoryLocator,
-    general_purpose_flags,
+    end_of_central_directory::Zip64EndCentralDirectoryLocator, general_purpose_flags, FileHeader,
+    FileInfo, LocalFileHeader, Signature,
 };
 use androscalpel_serializer::Serializable;
-use anyhow::{Context, Result, bail};
-use flate2::read::DeflateDecoder;
-use log::{info, warn};
-use std::collections::HashMap;
-use std::io::{Read, Seek, SeekFrom};
 
 #[derive(Debug, PartialEq, Eq)]
 pub struct ZipFileReader<T: Read + Seek> {
@@ -26,37 +19,29 @@ pub struct ZipFileReader<T: Read + Seek> {
 }
 
 impl<T: Read + Seek> ZipFileReader<T> {
-    pub fn new(mut reader: T) -> Result<Self> {
+    pub fn new(mut reader: T) -> Self {
         let end_of_central_directory_off =
-            Self::get_end_of_central_directory_offset(&mut reader)
-                .context("end of centrall directory not found, probably not a zip")?;
+            Self::get_end_of_central_directory_offset(&mut reader).unwrap();
         reader
             .seek(SeekFrom::Start(end_of_central_directory_off))
-            .context("Failed to seek to end of central directory")?;
-        let end_of_central_directory = EndCentralDirectory::deserialize(&mut reader)
-            .context("Failed to deserialize end of central directory")?;
-        let zip64_end_of_central_directory = if reader
+            .unwrap();
+        let end_of_central_directory = EndCentralDirectory::deserialize(&mut reader).unwrap();
+        reader
             .seek(SeekFrom::Start(
                 end_of_central_directory_off - Zip64EndCentralDirectoryLocator::SIZE as u64,
             ))
-            .is_ok()
-        {
-            let zip64_ecd_locator = Zip64EndCentralDirectoryLocator::deserialize(&mut reader).ok();
-            if let Some(zip64_ecd_locator) = zip64_ecd_locator {
-                assert_eq!(
-                    zip64_ecd_locator.disk_number_of_zip64_end_central_directory_start,
-                    0
-                );
-                assert!(zip64_ecd_locator.total_number_of_disks <= 1);
-                let zip64_edc_record_off =
-                    zip64_ecd_locator.offset_zip64_end_of_central_directory_record;
-                reader
-                    .seek(SeekFrom::Start(zip64_edc_record_off))
-                    .context("Failed to seek to end of zip64 central directory")?;
-                Zip64EndCentralDirectory::deserialize(&mut reader).ok()
-            } else {
-                None
-            }
+            .unwrap();
+        let zip64_ecd_locator = Zip64EndCentralDirectoryLocator::deserialize(&mut reader).ok();
+        let zip64_end_of_central_directory = if let Some(zip64_ecd_locator) = zip64_ecd_locator {
+            assert_eq!(
+                zip64_ecd_locator.disk_number_of_zip64_end_central_directory_start,
+                0
+            );
+            assert!(zip64_ecd_locator.total_number_of_disks <= 1);
+            let zip64_edc_record_off =
+                zip64_ecd_locator.offset_zip64_end_of_central_directory_record;
+            reader.seek(SeekFrom::Start(zip64_edc_record_off)).unwrap();
+            Zip64EndCentralDirectory::deserialize(&mut reader).ok()
         } else {
             None
         };
@@ -75,62 +60,36 @@ impl<T: Read + Seek> ZipFileReader<T> {
         zip_file
             .data
             .seek(SeekFrom::Start(zip_file.get_cd_offset()))
-            .context("Failed to seek to central directory")?;
+            .unwrap();
 
         let mut size_read = 0;
         let cd_size = zip_file.get_cd_size();
         while size_read < cd_size {
-            let header = FileHeader::deserialize(&mut zip_file.data)
-                .context("Failed to deserialize file header")?;
+            let header = FileHeader::deserialize(&mut zip_file.data).unwrap();
             //println!("{:#?}", header);
             size_read += header.size() as u64;
-            let pos_in_dir = zip_file
-                .data
-                .stream_position()
-                .context("Failed to get stream position")?;
+            let pos_in_dir = zip_file.data.stream_position().unwrap();
             if header.general_purpose_flags & general_purpose_flags::MASK_ENCRYPTED_CENTRAL_DIR != 0
             {
-                bail!("Central directory encryption not supported");
+                panic!("Central directory encryption not supported");
             }
             zip_file
                 .data
                 .seek(SeekFrom::Start(header.get_offset_local_header()))
-                .context("Failled to seek to local header")?;
-            let local_header = LocalFileHeader::deserialize(&mut zip_file.data)
-                .context("Failed to deserialize local file header")?;
-            let data_descriptor = if (local_header.general_purpose_flags
+                .unwrap();
+            let local_header = LocalFileHeader::deserialize(&mut zip_file.data).unwrap();
+            zip_file.data.seek(SeekFrom::Start(pos_in_dir)).unwrap();
+            if (local_header.general_purpose_flags
                 & general_purpose_flags::MASK_USE_DATA_DESCRIPTOR
                 != 0)
                 || (header.general_purpose_flags & general_purpose_flags::MASK_USE_DATA_DESCRIPTOR
                     != 0)
             {
-                warn!("Data Descriptor support is experimental");
-                zip_file
-                    .data
-                    .seek(SeekFrom::Current(header.compressed_size as i64))
-                    .context("failed to seek to after the file data")?;
-                if zip_file.zip64_end_of_central_directory.is_some() {
-                    Some(DataDescriptor::Zip64(
-                        DataDescriptor64::deserialize(&mut zip_file.data)
-                            .context("Failed to deserialize data descriptor 64")?,
-                    ))
-                } else {
-                    Some(DataDescriptor::Zip32(
-                        DataDescriptor32::deserialize(&mut zip_file.data)
-                            .context("Failed to deserialize data descriptor")?,
-                    ))
-                }
-            } else {
-                None
-            };
-            zip_file
-                .data
-                .seek(SeekFrom::Start(pos_in_dir))
-                .context("Failed to seek to position in directory")?;
+                panic!("Data Descriptor not yet suported");
+            }
             zip_file.files.push(FileInfo {
                 local_header,
                 header,
-                data_descriptor,
             });
         }
         assert_eq!(size_read, cd_size);
@@ -138,26 +97,24 @@ impl<T: Read + Seek> ZipFileReader<T> {
             zip_file
                 .data
                 .seek(SeekFrom::Start(zip_file.get_cd_offset() - 16))
-                .context("Failed to seek to central directory")?;
-            let magic =
-                Magic::deserialize(&mut zip_file.data).context("Failed to deserialize Magic")?;
+                .unwrap();
+            let magic = Magic::deserialize(&mut zip_file.data).unwrap();
             if magic == ApkSigningBlock::MAGIC {
                 zip_file
                     .data
                     .seek(SeekFrom::Start(zip_file.get_cd_offset() - 16 - 8))
-                    .context("Failed to seek to central directory")?;
-                let block_size = u64::deserialize(&mut zip_file.data)
-                    .context("Failed to deserialize block size")?;
+                    .unwrap();
+                let block_size = u64::deserialize(&mut zip_file.data).unwrap();
                 zip_file
                     .data
                     .seek(SeekFrom::Start(zip_file.get_cd_offset() - block_size - 8))
-                    .context("Failed to seek to central directory")?;
+                    .unwrap();
 
                 zip_file.apk_sign_block = ApkSigningBlock::deserialize(&mut zip_file.data).ok();
             }
         }
 
-        Ok(zip_file)
+        zip_file
     }
 
     pub fn is_zip64(&self) -> bool {
@@ -216,7 +173,7 @@ impl<T: Read + Seek> ZipFileReader<T> {
     }
 
     pub fn get_end_of_central_directory_offset(reader: &mut T) -> Option<u64> {
-        let file_size = reader.seek(SeekFrom::End(0)).ok()?;
+        let file_size = reader.seek(SeekFrom::End(0)).unwrap();
         let mut sig = Signature::default();
         let mut comment_size = 0;
         while sig != EndCentralDirectory::SIGNATURE {
@@ -224,8 +181,8 @@ impl<T: Read + Seek> ZipFileReader<T> {
                 .seek(SeekFrom::End(
                     -(EndCentralDirectory::MIN_SIZE as i64) - comment_size,
                 ))
-                .ok()?;
-            sig = Signature::deserialize(reader).ok()?;
+                .unwrap();
+            sig = Signature::deserialize(reader).unwrap();
             comment_size += 1;
             if comment_size > 65536
                 || comment_size as usize + EndCentralDirectory::MIN_SIZE > file_size as usize
@@ -296,8 +253,8 @@ impl<T: Read + Seek> ZipFileReader<T> {
         let mut lst_offset = 0;
         for file in files.iter() {
             if file.get_offset_local_header() != lst_offset {
-                info!(
-                    "Hole in zip before {} between 0x{:x} and 0x{:x}",
+                println!(
+                    "Hole before {} between 0x{:x} and 0x{:x}",
                     file.get_name(),
                     lst_offset,
                     file.get_offset_local_header()
@@ -309,8 +266,8 @@ impl<T: Read + Seek> ZipFileReader<T> {
         if let Some(apk_sign_block) = &self.apk_sign_block {
             let apk_sb_off = self.get_cd_offset() - apk_sign_block.size() as u64;
             if apk_sb_off != lst_offset {
-                info!(
-                    "Hole in zip before apk signing block, between 0x{:x} and 0x{:x}",
+                println!(
+                    "Hole before apk signing block, between 0x{:x} and 0x{:x}",
                     lst_offset, apk_sb_off
                 );
             }
@@ -318,58 +275,21 @@ impl<T: Read + Seek> ZipFileReader<T> {
             lst_offset = self.get_cd_offset();
         }
         if self.get_cd_offset() != lst_offset {
-            info!(
-                "Hole in zip before central directory between 0x{:x} and 0x{:x}",
+            println!(
+                "Hole before central directory between 0x{:x} and 0x{:x}",
                 lst_offset,
                 self.get_cd_offset()
             );
         }
     }
 
-    pub fn get_bin(&mut self, offset: u64, size: usize) -> Result<Vec<u8>> {
-        self.data
-            .seek(SeekFrom::Start(offset))
-            .context("Failed to seek to data")?;
-        let mut data = vec![0u8; size];
-        self.data
-            .read_exact(&mut data)
-            .context("failed to read data")?;
-        /*
+    pub fn get_bin(&mut self, offset: u64, size: usize) -> Vec<u8> {
+        self.data.seek(SeekFrom::Start(offset)).unwrap();
+        let mut data = vec![];
         for _ in 0..size {
             data.push(u8::deserialize(&mut self.data).unwrap());
         }
-        */
-        Ok(data)
-    }
-
-    pub fn read_file_as_vec(&mut self, name: &str) -> Result<Vec<u8>> {
-        let file = self
-            .get_file_info(name)
-            .with_context(|| format!("Failed to get info for {name}"))?;
-        let offset = file.get_file_offset();
-        let size_c = file.header.compressed_size as usize;
-        let size = file.header.uncompressed_size as usize;
-        let compression_method = file.header.compression_method;
-        let mut data = vec![0u8; size_c];
-        self.data
-            .seek(SeekFrom::Start(offset))
-            .with_context(|| format!("Failed to seek to start of file {name} (at 0x{offset:x})"))?;
-        self.data
-            .read_exact(&mut data)
-            .with_context(|| format!("Failed to read data for file {name}"))?;
-        match compression_method {
-            CompressionMethod::Stored => {}
-            CompressionMethod::Deflated => {
-                let mut decomp_data = vec![0u8; size];
-                let mut deflater = DeflateDecoder::new(&data[..]);
-                deflater
-                    .read_exact(&mut decomp_data)
-                    .with_context(|| format!("Failed to decompress data for file {name}"))?;
-                data = decomp_data
-            }
-            _ => unimplemented!(),
-        }
-        Ok(data)
+        data
     }
 
     pub fn get_file_info(&self, name: &str) -> Option<&FileInfo> {
@@ -377,29 +297,10 @@ impl<T: Read + Seek> ZipFileReader<T> {
     }
 
     pub fn get_classes_file_info(&self) -> Vec<&FileInfo> {
-        let files_map: HashMap<String, &FileInfo> = self
-            .files
+        self.files
             .iter()
-            .by_ref()
             .filter(|&file| match_dexfile_name(&file.get_name()))
-            .map(|file| (file.get_name(), file))
-            .collect();
-        let mut files = vec![];
-        let mut i = 0;
-        loop {
-            let name = if i == 0 {
-                "classes.dex".into()
-            } else {
-                format!("classes{}.dex", i + 1)
-            };
-            if let Some(file) = files_map.get(&name) {
-                files.push(*file);
-            } else {
-                break;
-            }
-            i += 1;
-        }
-        files
+            .collect()
     }
 }
 

apk_frauder/src/zip_writer.rs

@@ -2,13 +2,10 @@ use std::io;
 use std::io::{Cursor, Read, Seek, SeekFrom, Write};
 
 use crate::compression::CompressionMethod;
-use crate::data_descriptor::{DataDescriptor, DataDescriptor32, DataDescriptor64};
 use crate::end_of_central_directory::{
     EndCentralDirectory, Zip64EndCentralDirectory, Zip64EndCentralDirectoryLocator,
 };
-use crate::{
-    general_purpose_flags, ExtraField, FileHeader, FileInfo, LocalFileHeader, ZipFileReader,
-};
+use crate::{general_purpose_flags, FileHeader, FileInfo, LocalFileHeader, ZipFileReader};
 use androscalpel_serializer::Serializable;
 use flate2::write::DeflateEncoder;
 use flate2::{Compression, CrcWriter};
@@ -57,6 +54,24 @@ impl<T: Write> ZipFileWriter<T> {
         self.files.push(file_info);
     }
 
+    /// Insert a file at the begining of archive that will not be tracked.
+    pub fn insert_untracked_file<U: Read + Seek>(&mut self, file: &mut U) {
+        let pos = file.stream_position().unwrap();
+        let pos_end = file.seek(SeekFrom::End(0)).unwrap();
+        file.seek(SeekFrom::Start(pos)).unwrap();
+        let file_size = pos_end - pos;
+        let mut remaining_size = file_size;
+        let mut buffer = [0u8; 4096];
+        while remaining_size != 0 {
+            let read = file
+                .read(&mut buffer[0..core::cmp::min(4096, remaining_size as usize)])
+                .unwrap();
+            self.data.write_all(&buffer[0..read]).unwrap();
+            remaining_size -= read as u64;
+        }
+        self.current_offset += file_size;
+    }
+
     /// Insert a file
     /// `header` provide for the values that are not computed from the file.
     /// `local_header` can be provided is information in the local header differ
@@ -66,16 +81,11 @@ impl<T: Write> ZipFileWriter<T> {
         file: &mut U,
         mut header: FileHeader,
         local_header: Option<LocalFileHeader>,
-        data_descriptor: Option<DataDescriptor>,
     ) {
         assert!(header.general_purpose_flags & general_purpose_flags::MASK_ENCRYPTED == 0);
-        //assert!(
-        //    header.general_purpose_flags & general_purpose_flags::MASK_USE_DATA_DESCRIPTOR == 0,
-        //    "Writing file with data_descriptor is not yet implemented"
-        //); // TODO
-        let mut use_data_descriptor = data_descriptor.is_some()
-            || ((header.general_purpose_flags & general_purpose_flags::MASK_USE_DATA_DESCRIPTOR)
-                != 0);
+        assert!(
+            header.general_purpose_flags & general_purpose_flags::MASK_USE_DATA_DESCRIPTOR == 0
+        ); // TODO
         assert!(header.general_purpose_flags & general_purpose_flags::MASK_STRONG_ENCRYPTION == 0);
         assert!(
             header.general_purpose_flags & general_purpose_flags::MASK_ENCRYPTED_CENTRAL_DIR == 0
@@ -83,10 +93,9 @@ impl<T: Write> ZipFileWriter<T> {
         assert!(
             header.general_purpose_flags
                 & (general_purpose_flags::MASK_COMPRESS_OPTION_1
-                    | general_purpose_flags::MASK_COMPRESS_OPTION_2)
+                    | general_purpose_flags::MASK_COMPRESS_OPTION_2
+                    | general_purpose_flags::MASK_UTF8_FILENAME)
                 == 0
-                || header.compression_method == CompressionMethod::Deflated
-                || header.compression_method == CompressionMethod::Stored
         );
         assert!(
             header.compression_method == CompressionMethod::Deflated
@@ -94,12 +103,9 @@ impl<T: Write> ZipFileWriter<T> {
         );
         let mut local_header = if let Some(header) = local_header {
             assert!(header.general_purpose_flags & general_purpose_flags::MASK_ENCRYPTED == 0);
-            //assert!(
-            //    header.general_purpose_flags & general_purpose_flags::MASK_USE_DATA_DESCRIPTOR == 0,
-            //    "Writing file with data_descriptor is not yet implemented"
-            //); // TODO
-            use_data_descriptor |=
-                header.general_purpose_flags & general_purpose_flags::MASK_USE_DATA_DESCRIPTOR != 0;
+            assert!(
+                header.general_purpose_flags & general_purpose_flags::MASK_USE_DATA_DESCRIPTOR == 0
+            ); // TODO
             assert!(
                 header.general_purpose_flags & general_purpose_flags::MASK_STRONG_ENCRYPTION == 0
             );
@@ -110,29 +116,23 @@ impl<T: Write> ZipFileWriter<T> {
             assert!(
                 header.general_purpose_flags
                     & (general_purpose_flags::MASK_COMPRESS_OPTION_1
-                        | general_purpose_flags::MASK_COMPRESS_OPTION_2)
+                        | general_purpose_flags::MASK_COMPRESS_OPTION_2
+                        | general_purpose_flags::MASK_UTF8_FILENAME)
                     == 0
-                    || header.compression_method == CompressionMethod::Deflated
-                    || header.compression_method == CompressionMethod::Stored
             );
             assert!(
                 header.compression_method == CompressionMethod::Deflated
                     || header.compression_method == CompressionMethod::Stored
             );
-            let mut general_purpose_flags = header.general_purpose_flags;
-            // We only support options for Deflate parameter and data descriptor
-            if header.compression_method == CompressionMethod::Deflated {
-                general_purpose_flags &= general_purpose_flags::MASK_COMPRESS_OPTION_1
-                    | general_purpose_flags::MASK_COMPRESS_OPTION_2;
-            } else {
-                general_purpose_flags = 0
-            }
-            if use_data_descriptor {
-                general_purpose_flags |= general_purpose_flags::MASK_USE_DATA_DESCRIPTOR;
-            }
             LocalFileHeader {
                 version_needed_to_extract: header.version_needed_to_extract,
-                general_purpose_flags,
+                general_purpose_flags: if header.compression_method == CompressionMethod::Deflated {
+                    header.general_purpose_flags
+                        & (general_purpose_flags::MASK_COMPRESS_OPTION_1
+                            | general_purpose_flags::MASK_COMPRESS_OPTION_2)
+                } else {
+                    0
+                },
                 compression_method: header.compression_method,
                 last_mod_file_time: header.last_mod_file_time,
                 last_mod_file_data: header.last_mod_file_data,
@@ -153,10 +153,6 @@ impl<T: Write> ZipFileWriter<T> {
                             | general_purpose_flags::MASK_COMPRESS_OPTION_2)
                 } else {
                     0
-                } | if use_data_descriptor {
-                    general_purpose_flags::MASK_USE_DATA_DESCRIPTOR
-                } else {
-                    0
                 },
                 compression_method: header.compression_method,
                 last_mod_file_time: header.last_mod_file_time,
@@ -188,17 +184,9 @@ impl<T: Write> ZipFileWriter<T> {
             CompressionMethod::Deflated => {
                 // TODO: find a way to do this in place, the compressed data can be large, and storing it
                 // in memory is bad, but Deflate consume the Reader, so we cannot juste give it self.data
-                let option = match header.general_purpose_flags
-                    & (general_purpose_flags::MASK_COMPRESS_OPTION_1
-                        | general_purpose_flags::MASK_COMPRESS_OPTION_2)
-                {
-                    0b000 => Compression::default(),
-                    0b010 => Compression::best(),
-                    0b100 => Compression::fast(),
-                    0b110 => panic!("Super Fast deflate compression not supported"),
-                    flag => panic!("Something is verry wrong: {flag:b}"),
-                };
-                let mut compressor = CrcWriter::new(DeflateEncoder::new(Vec::new(), option));
+                // TODO: Compression::default -> use flag?
+                let mut compressor =
+                    CrcWriter::new(DeflateEncoder::new(Vec::new(), Compression::default()));
                 io::copy(file, &mut compressor).unwrap();
                 local_header.crc_32 = compressor.crc().sum();
                 compressor.into_inner().finish().unwrap()
@@ -216,78 +204,15 @@ impl<T: Write> ZipFileWriter<T> {
         header.set_uncompressed_size(local_header.get_uncompressed_size());
         header.set_offset_local_header(header_offset);
 
-        let use_z64 = header
-            .extra_field
-            .iter()
-            .any(|f| matches!(f, ExtraField::Zip64(_)));
-
-        let data_descriptor = if use_data_descriptor {
-            match data_descriptor {
-                None if use_z64 => Some(DataDescriptor::Zip64(DataDescriptor64 {
-                    crc_32: local_header.crc_32,
-                    compressed_size: local_header.get_compressed_size(),
-                    uncompressed_size: local_header.get_uncompressed_size(),
-                    use_signature: true,
-                })),
-                None => Some(DataDescriptor::Zip32(DataDescriptor32 {
-                    crc_32: local_header.crc_32,
-                    compressed_size: local_header.get_compressed_size() as u32,
-                    uncompressed_size: local_header.get_uncompressed_size() as u32,
-                    use_signature: true,
-                })),
-                Some(DataDescriptor::Zip32(DataDescriptor32 { use_signature, .. })) if use_z64 => {
-                    Some(DataDescriptor::Zip64(DataDescriptor64 {
-                        crc_32: local_header.crc_32,
-                        compressed_size: local_header.get_compressed_size(),
-                        uncompressed_size: local_header.get_uncompressed_size(),
-                        use_signature,
-                    }))
-                }
-                Some(DataDescriptor::Zip64(DataDescriptor64 { use_signature, .. })) if !use_z64 => {
-                    Some(DataDescriptor::Zip32(DataDescriptor32 {
-                        crc_32: local_header.crc_32,
-                        compressed_size: local_header.get_compressed_size() as u32,
-                        uncompressed_size: local_header.get_uncompressed_size() as u32,
-                        use_signature,
-                    }))
-                }
-                Some(DataDescriptor::Zip64(DataDescriptor64 { use_signature, .. })) => {
-                    Some(DataDescriptor::Zip64(DataDescriptor64 {
-                        crc_32: local_header.crc_32,
-                        compressed_size: local_header.get_compressed_size(),
-                        uncompressed_size: local_header.get_uncompressed_size(),
-                        use_signature,
-                    }))
-                }
-                Some(DataDescriptor::Zip32(DataDescriptor32 { use_signature, .. })) => {
-                    Some(DataDescriptor::Zip32(DataDescriptor32 {
-                        crc_32: local_header.crc_32,
-                        compressed_size: local_header.get_compressed_size() as u32,
-                        uncompressed_size: local_header.get_uncompressed_size() as u32,
-                        use_signature,
-                    }))
-                }
-            }
-        } else {
-            None
-        };
-
-        match &data_descriptor {
-            Some(DataDescriptor::Zip32(data_descriptor)) => {
-                data_descriptor.serialize(&mut self.data).unwrap();
-                self.current_offset += data_descriptor.size() as u64;
-            }
-            Some(DataDescriptor::Zip64(data_descriptor)) => {
-                data_descriptor.serialize(&mut self.data).unwrap();
-                self.current_offset += data_descriptor.size() as u64;
-            }
-            _ => {}
-        }
+        // TODO:  compression flags are not used right now, set to zero
+        local_header.general_purpose_flags &= !(general_purpose_flags::MASK_COMPRESS_OPTION_1
+            | general_purpose_flags::MASK_COMPRESS_OPTION_2);
+        header.general_purpose_flags &= !(general_purpose_flags::MASK_COMPRESS_OPTION_1
+            | general_purpose_flags::MASK_COMPRESS_OPTION_2);
 
         let file_info = FileInfo {
             local_header,
             header,
-            data_descriptor,
         };
         self.files.push(file_info);
     }