WIP

androscalpel/src/annotation.rs

@@ -7,7 +7,7 @@ use pyo3::prelude::*;
 use crate::hashmap_vectorize;
 use crate::{
     dex_id::IdType, value::DexValue, DexString, IdField, IdMethod, IdMethodType, MethodHandle,
-    Result,
+    Result, Visitable, VisitableMut, Visitor, VisitorMut,
 };
 
 /// Annotation with a visibility
@@ -217,3 +217,27 @@ impl DexAnnotation {
         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/code.rs

@@ -265,7 +265,7 @@ impl Code {
 
         let mut new_insns = vec![];
         last_ins_was_a_label = false;
-        for ins in self.insns.into_iter() {
+        for ins in self.insns.iter().cloned() {
             match ins {
                 Instruction::Goto { label } => {
                     last_ins_was_a_label = false;

androscalpel/src/dex_writer.rs

@@ -511,9 +511,7 @@ impl DexWriter {
         for ins in &code.insns {
             match ins {
                 Instruction::FillArrayData {
-                    data,
+                    data, elt_width, ..
-                    elt_width,
-                    arr,
                 } => {
                     let payload = InsFormat::FormatFillArrayDataPayload {
                         elt_width: *elt_width,

androscalpel/src/instructions.rs

@@ -3862,7 +3862,6 @@ impl Instruction {
             Self::FloatToDouble { dest, val } => {
                 sanity_check_4_bit_reg!("float-to-double", dest, val)
             }
-            Self::LongToInt { dest, val } => sanity_check_4_bit_reg!("long-to-int", dest, val),
             Self::DoubleToInt { dest, val } => sanity_check_4_bit_reg!("double-to-int", dest, val),
             Self::DoubleToLong { dest, val } => {
                 sanity_check_4_bit_reg!("double-to-long", dest, val)
@@ -6609,3 +6608,30 @@ impl CallSite {
         handles
     }
 }
+
+impl<V: Visitor> Visitable<V> for CallSite {
+    fn default_visit(&self, v: &mut V) -> Result<()> {
+        v.visit_method_handle(&self.method_handle)?;
+        v.visit_string(&self.name)?;
+        v.visit_method_type(&self.type_)?;
+        for arg in &self.args {
+            v.visit_value(&arg)?;
+        }
+        Ok(())
+    }
+}
+
+impl<V: VisitorMut> VisitableMut<V> for CallSite {
+    fn default_visit_mut(self, v: &mut V) -> Result<Self> {
+        Ok(Self {
+            method_handle: v.visit_method_handle(self.method_handle)?,
+            name: v.visit_string(self.name)?,
+            type_: v.visit_method_type(self.type_)?,
+            args: self
+                .args
+                .into_iter()
+                .map(|arg| v.visit_value(arg))
+                .collect::<Result<_>>()?,
+        })
+    }
+}

androscalpel/src/lib.rs

@@ -33,6 +33,7 @@ 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::*;
@@ -91,199 +92,6 @@ fn androscalpel(py: Python, m: &Bound<'_, PyModule>) -> PyResult<()> {
 /// Dalvik opcode for Androscalpel.
 fn androscalpel_ins(_py: Python, m: &Bound<'_, PyModule>) -> PyResult<()> {
     m.add_class::<ins::CallSite>()?;
-    m.add_class::<ins::Nop>()?;
+    m.add_class::<ins::Instruction>()?;
-    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/scalar.rs

@@ -4,7 +4,10 @@ use crate::Result;
 use serde::{Deserialize, Serialize};
 use std::collections::HashSet;
 
-use crate::{DexString, DexValue, IdField, IdMethod, IdMethodType, IdType, MethodHandle};
+use crate::{
+    DexString, DexValue, IdField, IdMethod, IdMethodType, IdType, MethodHandle, Visitable,
+    VisitableMut, Visitor, VisitorMut,
+};
 use pyo3::prelude::*;
 
 #[pyclass]
@@ -417,3 +420,23 @@ 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/value.rs

@@ -6,7 +6,10 @@ use std::collections::HashSet;
 use pyo3::exceptions::PyTypeError;
 use pyo3::prelude::*;
 
-use crate::{dex_id::*, scalar::*, DexAnnotation, DexString, MethodHandle};
+use crate::{
+    dex_id::*, scalar::*, DexAnnotation, DexString, MethodHandle, Result, Visitable, VisitableMut,
+    Visitor, VisitorMut,
+};
 
 #[derive(Debug, Clone, PartialEq, Deserialize, Serialize)]
 pub enum DexValue {
@@ -30,6 +33,56 @@ 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)?)),
+        }
+    }
+}
+
 impl<'source> FromPyObject<'source> for DexValue {
     fn extract(ob: &'source PyAny) -> PyResult<Self> {
         if let Ok(val) = DexByte::extract(ob) {

androscalpel/src/visitor.rs

@@ -1,8 +1,8 @@
 //! The visitor trait and common implementations.
 
 use crate::{
-    ins::Instruction, CallSite, DexString, IdEnum, IdField, IdMethod, IdMethodType, IdType,
+    ins::Instruction, scalar::*, CallSite, DexAnnotation, DexString, DexValue, IdEnum, IdField,
-    MethodHandle, Result,
+    IdMethod, IdMethodType, IdType, MethodHandle, Result,
 };
 use std::collections::HashSet;
 
@@ -34,6 +34,42 @@ pub trait Visitor: Sized {
     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(())
+    }
 }
 
 pub trait VisitorMut: Sized {
@@ -64,6 +100,42 @@ pub trait VisitorMut: Sized {
     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)
+    }
 }
 
 /// Trait for structures that can be visited.