fix string order

androscalpel/src/dex_id.rs

@@ -227,6 +227,28 @@ impl IdType {
         Self(ty)
     }
 
+    /// Try to parse a smali representation of type into a IdType.
+    ///
+    /// In practice, it's juste wrapper arround `Self::new()` for consistancy with other
+    /// id types.
+    ///
+    /// ```
+    /// use androscalpel::IdType
+    ///
+    /// let id_type = IdType::from_smali(
+    ///   "Landroidx/core/util/Predicate;"
+    /// ).unwrap();
+    ///
+    /// assert_eq!(
+    ///   id_type,
+    ///   IdType::class("androidx/core/util/Predicate")
+    /// );
+    /// ```
+    #[staticmethod]
+    pub fn from_smali(smali_repr: &str) -> Result<Self> {
+        Self::new(smali_repr.into())
+    }
+
     /// Return a list of types from a string of concatenated types (like the ones between
     /// parentheses in the repr of a prototype)
     ///

androscalpel/src/tests/mod.rs

@@ -53,7 +53,7 @@ fn test_generated_apk_equivalence() {
 }
 
 #[test]
-fn test_string_order() {
+fn test_string_order_in_dex() {
     use std::collections::HashSet;
     let dex = DexFileReader::new(get_hello_world_dex()).unwrap();
     let new_dex = DexFileReader::new(get_hello_world_recompilled()).unwrap();
@@ -67,7 +67,28 @@ fn test_string_order() {
             println!("{}", string.__str__());
         }
     }
-    assert_eq!(dex.get_string_ids().len(), new_dex.get_string_ids().len());
+    assert_eq!(
+        dex.get_string_ids().len(),
+        new_dex.get_string_ids().len(),
+        "Not the same number of strings"
+    );
+    /* this is hell to debug
+    let strings: Vec<_> = (0..dex.get_string_ids().len())
+        .map(|idx| DexString(dex.get_string(idx as u32).unwrap()))
+        .collect();
+    let new_strings: Vec<_> = (0..new_dex.get_string_ids().len())
+        .map(|idx| DexString(new_dex.get_string(idx as u32).unwrap()))
+        .collect();
+    assert_eq!(strings, new_strings);
+    */
+    for idx in 0..dex.get_string_ids().len() {
+        let string = DexString(dex.get_string(idx as u32).unwrap());
+        let new_string = DexString(new_dex.get_string(idx as u32).unwrap());
+        assert_eq!(
+            string, new_string,
+            "Strings {idx} do not match, left original, right recompiled"
+        );
+    }
 }
 
 #[test]
@@ -141,3 +162,58 @@ fn test_parse_field_smali() {
         )
     );
 }
+
+#[test]
+fn test_sort_fields() {
+    let f1 = IdField::from_smali(
+        "Landroidx/lifecycle/WithLifecycleStateKt$suspendWithStateAtLeastUnchecked$2$observer$1;\
+        ->$co:Lkotlinx/coroutines/CancellableContinuation;",
+    )
+    .unwrap();
+    let f2 = IdField::from_smali(
+        "Lcom/google/android/material/search/SearchBarAnimationHelper$$ExternalSyntheticLambda4;\
+        ->f$0:Landroid/animation/Animator;",
+    )
+    .unwrap();list1
+    let mut list1 = vec![f2.clone(), f1.clone()];
+    let mut list2 = vec![f1.clone(), f2.clone()];
+    list1.sort();
+    list2.sort();
+    assert_eq!(list1, list2);
+    assert_eq!(list1, vec![f1.clone(), f2.clone()]);
+}
+
+#[test]
+fn test_sort_types() {
+    let t1 = IdType::from_smali(
+        "Landroidx/lifecycle/WithLifecycleStateKt$suspendWithStateAtLeastUnchecked$2$observer$1;",
+    )
+    .unwrap();
+    let t2 = IdType::from_smali(
+        "Lcom/google/android/material/search/SearchBarAnimationHelper$$ExternalSyntheticLambda4;",
+    )
+    .unwrap();
+    let mut list1 = vec![t2.clone(), t1.clone()];
+    let mut list2 = vec![t1.clone(), t2.clone()];
+    list1.sort();
+    list2.sort();
+    assert_eq!(list1, list2);
+    assert_eq!(list1, vec![t1.clone(), t2.clone()]);
+}
+
+#[test]
+fn test_sort_strings() {
+    let s1: DexString =
+        "Landroidx/lifecycle/WithLifecycleStateKt$suspendWithStateAtLeastUnchecked$2$observer$1;"
+            .into();
+    let s2: DexString =
+        "Lcom/google/android/material/search/SearchBarAnimationHelper$$ExternalSyntheticLambda4;"
+            .into();
+
+    let mut list1 = vec![s2.clone(), s1.clone()];
+    let mut list2 = vec![s1.clone(), s2.clone()];
+    list1.sort();
+    list2.sort();
+    assert_eq!(list1, list2);
+    assert_eq!(list1, vec![s1.clone(), s2.clone()]);
+}

androscalpel_serializer/src/core/string.rs

@@ -36,8 +36,9 @@ const VALUE_TRAYLING_BYTE_PREFIX: u8 = 0b1000_0000;
 
 impl Ord for StringDataItem {
     fn cmp(&self, other: &Self) -> Ordering {
-        self.data
+        self.get_aosp_utf16()
-            .cmp(&other.data)
+            .unwrap()
+            .cmp(&other.get_aosp_utf16().unwrap())
             .then(self.utf16_size.cmp(&other.utf16_size))
     }
 }
@@ -122,6 +123,57 @@ impl From<&str> for StringDataItem {
 }
 
 impl StringDataItem {
+    /// Return the utf-16 string used by google in the aosp for comparaison & other.
+    fn get_aosp_utf16(&self) -> Result<Vec<u32>> {
+        let mut utf16_string = vec![];
+        let mut i = 0;
+        while i < self.data.len() {
+            let one = self.data[i] as u32;
+            i += 1;
+            if one & 0x80 == 0 {
+                utf16_string.push(one);
+                continue;
+            }
+
+            if i >= self.data.len() {
+                return Err(Error::InvalidStringEncoding(
+                    "String contains invalid caracters".into(),
+                ));
+            }
+            let two = self.data[i] as u32;
+            i += 1;
+            if one & 0x20 == 0 {
+                utf16_string.push((one & 0x1f) << 6 | (two & 0x3f));
+                continue;
+            }
+
+            if i >= self.data.len() {
+                return Err(Error::InvalidStringEncoding(
+                    "String contains invalid caracters".into(),
+                ));
+            }
+            let three = self.data[i] as u32;
+            i += 1;
+            if one & 0x10 == 0 {
+                utf16_string.push((one & 0x0f) << 12 | (two & 0x3f) << 6 | (three & 0x3f));
+                continue;
+            }
+
+            if i >= self.data.len() {
+                return Err(Error::InvalidStringEncoding(
+                    "String contains invalid caracters".into(),
+                ));
+            }
+            let four = self.data[i] as u32;
+            i += 1;
+            let code_point =
+                (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);
+        }
+        Ok(utf16_string)
+    }
     fn get_string(&self) -> Result<String> {
         let mut string = String::new();
         let mut i = 0;
@@ -481,4 +533,25 @@ mod test {
             assert_eq!(encoded, expected.as_str().into());
         }
     }
+
+    /// Apparently I don't know how to code an order relation so here it is...
+    #[test]
+    fn test_ord_relation() {
+        let s1: StringDataItem = "Landroidx/lifecycle/WithLifecycleStateKt$suspendWithStateAtLeastUnchecked$2$observer$1;".into();
+        let s2: StringDataItem = "Lcom/google/android/material/search/SearchBarAnimationHelper$$ExternalSyntheticLambda4;".into();
+        let s1_utf16 = s1.get_aosp_utf16().unwrap();
+        let s2_utf16 = s2.get_aosp_utf16().unwrap();
+        assert_eq!(s1_utf16 < s2_utf16, true);
+        assert_eq!(s1_utf16 == s2_utf16, false);
+        assert_eq!(s1_utf16 > s2_utf16, false);
+        assert_eq!(s2_utf16 < s1_utf16, false);
+        assert_eq!(s2_utf16 == s1_utf16, false);
+        assert_eq!(s2_utf16 > s1_utf16, true);
+        assert_eq!(s1 < s2, true);
+        assert_eq!(s1 == s2, false);
+        assert_eq!(s1 > s2, false);
+        assert_eq!(s2 < s1, false);
+        assert_eq!(s2 == s1, false);
+        assert_eq!(s2 > s1, true);
+    }
 }

test.py

@@ -22,43 +22,43 @@ with z.ZipFile(APK_NAME) as zipf:
 apk = Apk()
 apk.add_dex_file(dex)
 
-# clazz_id = IdType("Lcom/example/testapplication/ui/home/HomeViewModel;")
+clazz_id = IdType("Lcom/example/testapplication/ui/home/HomeViewModel;")
-# proto_id = IdMethodType(IdType("Ljava/lang/String;"), [])
+proto_id = IdMethodType(IdType("Ljava/lang/String;"), [])
-# method_id = IdMethod("text_gen", proto_id, clazz_id)
+method_id = IdMethod("text_gen", proto_id, clazz_id)
-#
+
-# clazz = apk.classes[clazz_id]
+clazz = apk.classes[clazz_id]
-# method = clazz.virtual_methods[method_id]
+method = clazz.virtual_methods[method_id]
-# code = method.code
+code = method.code
-#
+
 logging.getLogger().setLevel(logging.WARNING)
-#
+
-# print(f"[+] Code of {method_id} ")
+print(f"[+] Code of {method_id} ")
-# for i in code.insns:
+for i in code.insns:
-#     print(f"    {i}")
+    print(f"    {i}")
-# print("[+] Modify code")
+print("[+] Modify code")
-#
+
-# new_insns = []
+new_insns = []
-# for i in code.insns:
+for i in code.insns:
-#     if isinstance(i, ins.ConstString):
+    if isinstance(i, ins.ConstString):
-#         if i.lit == "Hello":
+        if i.lit == "Hello":
-#             i = ins.ConstString(i.reg, DexString("Degemer Mat"))
+            i = ins.ConstString(i.reg, DexString("Degemer Mat"))
-#         elif i.lit == "Bye":
+        elif i.lit == "Bye":
-#             i = ins.ConstString(i.reg, DexString("Kenavo"))
+            i = ins.ConstString(i.reg, DexString("Kenavo"))
-#     new_insns.append(i)
+    new_insns.append(i)
-#
+
-# # This need improving!
+# This need improving!
-# code = Code(code.registers_size, code.ins_size, code.outs_size, new_insns)
+code = Code(code.registers_size, code.ins_size, code.outs_size, new_insns)
-# apk.set_method_code(method_id, code)
+apk.set_method_code(method_id, code)
-# # apk.set_method_code(method.descriptor, code)
+# apk.set_method_code(method.descriptor, code)
-#
+
-#
+
-# clazz = apk.classes[clazz_id]
+clazz = apk.classes[clazz_id]
-# method = clazz.virtual_methods[method_id]
+method = clazz.virtual_methods[method_id]
-# code = method.code
+code = method.code
-# print(f"[+] New code of {method_id} ")
+print(f"[+] New code of {method_id} ")
-# for i in code.insns:
+for i in code.insns:
-#     print(f"    {i}")
+    print(f"    {i}")
-#
+
 # # Strip class for debugging
 # classes = list(
 #     filter(
@@ -80,22 +80,22 @@ print("[+] Recompile")
 
 dex_raw = apk.gen_raw_dex()
 
-new_apk = Apk()
+# new_apk = Apk()
-for dex in dex_raw:
+# for dex in dex_raw:
-    new_apk.add_dex_file(dex)
+#     new_apk.add_dex_file(dex)
 
 
-# print("[+] Repackage")
+print("[+] Repackage")
-#
+
-# utils.replace_dex(
+utils.replace_dex(
-#     APK_NAME,
+    APK_NAME,
-#     APK_NAME.parent / (APK_NAME.name.removesuffix(".apk") + "-instrumented.apk"),
+    APK_NAME.parent / (APK_NAME.name.removesuffix(".apk") + "-instrumented.apk"),
-#     dex_raw,
+    dex_raw,
-#     Path().parent / "my-release-key.jks",
+    Path().parent / "my-release-key.jks",
-#     zipalign=Path.home() / "Android" / "Sdk" / "build-tools" / "34.0.0" / "zipalign",
+    zipalign=Path.home() / "Android" / "Sdk" / "build-tools" / "34.0.0" / "zipalign",
-#     apksigner=Path.home() / "Android" / "Sdk" / "build-tools" / "34.0.0" / "apksigner",
+    apksigner=Path.home() / "Android" / "Sdk" / "build-tools" / "34.0.0" / "apksigner",
-# )
+)
-#
+
 last_id = None
 
 MAX_REQ = 1
@@ -167,8 +167,8 @@ def cmp_list(a, b, req=0):
             cmp(a[i], b[i], req + 1)
 
 
-apk_eq = new_apk == apk
+# apk_eq = new_apk == apk
-print(f"[+] apk are equals: {nice_bool(apk_eq)}")
+# print(f"[+] apk are equals: {nice_bool(apk_eq)}")
 # if not apk_eq:
 #    cmp(new_apk, apk)