add support for data descriptor

androscalpel/examples/count_ins_and_genapk.rs

@@ -1,5 +1,9 @@
 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 {
@@ -16,7 +20,7 @@ impl VisitorMut for InsCounter {
 
 fn main() {
     let args: Vec<_> = env::args().collect();
-    assert!(args.len() > 1, "Need one argument");
+    assert!(args.len() > 2, "usage: {} input.apk output.apk", args[0]);
 
     let mut cnt = InsCounter::default();
     let apk = cnt
@@ -24,16 +28,29 @@ fn main() {
         .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;
-    /*apk_frauder::replace_dex(
+    loop {
-        (&args[1]).into(),
+        let name = if i == 0 {
-        (&args[1] + ".pathed").into(),
+            "classes.dex".into()
-        &mut dex_files,
+        } else {
-        cli.keystore,
+            format!("classes{}.dex", i + 1)
-        cli.zipalign,
+        };
-        cli.apksigner,
+        if let Some(file) = files.remove(&name) {
-        cli.keypassword.as_deref(),
+            dex_files.push(Cursor::new(file))
-        None::<HashMap<_, Option<Cursor<&[u8]>>>>,
+        } else {
-    )*/
+            break;
+        }
+        i += 1;
+    }
+
+    replace_dex_unsigned(
+        &args[1],
+        &args[2],
+        &mut dex_files,
+        None::<HashMap<_, Option<Cursor<&[u8]>>>>,
+    )
+    .unwrap();
 }

apk_frauder/src/lib.rs

@@ -143,7 +143,74 @@ impl FileInfo {
     }
 }
 
-/// Replace the dex files a an apk an resigned the apk.
+/// Replace the dex files in an apk and strip the old signature.
+#[allow(clippy::too_many_arguments)]
+pub fn replace_dex_unsigned(
+    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")?;
+    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")?)
+    .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");
+        } else {
+            file_info_ref.set_name(&format!("classes{}.dex", i + 1));
+        }
+        apk_out.insert_file(
+            &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
 ///
@@ -177,58 +244,8 @@ pub fn replace_dex(
     let aligned_path = tmp_dir.join("aligned.apk");
     fs::create_dir_all(&tmp_dir).context("Failed to create temporary directory")?;
 
-    let file = File::open(&apk).with_context(|| {
+    replace_dex_unsigned(apk, &unaligned_path, dexfiles, additionnal_files)?;
-        format!(
-            "failed to open file {}",
-            apk.as_ref().to_str().unwrap_or("")
-        )
-    })?;
-    let mut apk = ZipFileReader::new(file)?;
-    let file = File::create(&unaligned_path).context("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")?)
-    .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");
-        } else {
-            file_info_ref.set_name(&format!("classes{}.dex", i + 1));
-        }
-        apk_out.insert_file(
-            &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();
     // TODO: we can probably do that ourself an spare ourself the trouble of finding zipalign
     Command::new(zipalign)
         .arg("-v")

apk_frauder/src/zip_writer.rs

@@ -2,11 +2,13 @@ use std::io;
 use std::io::{Cursor, Read, Seek, SeekFrom, Write};
 
 use crate::compression::CompressionMethod;
-use crate::data_descriptor::DataDescriptor;
+use crate::data_descriptor::{DataDescriptor, DataDescriptor32, DataDescriptor64};
 use crate::end_of_central_directory::{
     EndCentralDirectory, Zip64EndCentralDirectory, Zip64EndCentralDirectoryLocator,
 };
-use crate::{FileHeader, FileInfo, LocalFileHeader, ZipFileReader, general_purpose_flags};
+use crate::{
+    general_purpose_flags, ExtraField, FileHeader, FileInfo, LocalFileHeader, ZipFileReader,
+};
 use androscalpel_serializer::Serializable;
 use flate2::write::DeflateEncoder;
 use flate2::{Compression, CrcWriter};
@@ -67,9 +69,13 @@ impl<T: Write> ZipFileWriter<T> {
         data_descriptor: Option<DataDescriptor>,
     ) {
         assert!(header.general_purpose_flags & general_purpose_flags::MASK_ENCRYPTED == 0);
-        assert!(
+        //assert!(
-            header.general_purpose_flags & general_purpose_flags::MASK_USE_DATA_DESCRIPTOR == 0
+        //    header.general_purpose_flags & general_purpose_flags::MASK_USE_DATA_DESCRIPTOR == 0,
-        ); // TODO
+        //    "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_STRONG_ENCRYPTION == 0);
         assert!(
             header.general_purpose_flags & general_purpose_flags::MASK_ENCRYPTED_CENTRAL_DIR == 0
@@ -88,9 +94,12 @@ 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!(
+            //assert!(
-                header.general_purpose_flags & general_purpose_flags::MASK_USE_DATA_DESCRIPTOR == 0
+            //    header.general_purpose_flags & general_purpose_flags::MASK_USE_DATA_DESCRIPTOR == 0,
-            ); // TODO
+            //    "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_STRONG_ENCRYPTION == 0
             );
@@ -110,15 +119,20 @@ impl<T: Write> ZipFileWriter<T> {
                 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: if header.compression_method == CompressionMethod::Deflated {
+                general_purpose_flags,
-                    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,
@@ -139,6 +153,10 @@ 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,
@@ -198,9 +216,74 @@ impl<T: Write> ZipFileWriter<T> {
         header.set_uncompressed_size(local_header.get_uncompressed_size());
         header.set_offset_local_header(header_offset);
 
-        if data_descriptor.is_some() {
+        let use_z64 = header
-            panic!("Writing file with data_descriptor is not yet implemented");
+            .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;
+            }
+            _ => {}
+        }
+
         let file_info = FileInfo {
             local_header,
             header,