Haibo Shi

58 papers A* 2A 1B 4C 1Misc 1Journal 40Unranked 9
YearRankTypeTitle / Venue / Authors
2026 J jnl
J. Comput. Civ. Eng.
Yan Xing, Xu Han, Dong An, Gangyuan Yao, Haibo Shi, Weidong Liu
2025 J jnl
CoRR
Jun Li, Yijue Zhang, Haibo Shi, Minhong Li, Qiwei Li, Xiaohua Qian
2025 J jnl
IEEE Trans. Medical Imaging
Jun Li, Yijue Zhang, Haibo Shi, Minhong Li, Qiwei Li, Xiaohua Qian
2025 J jnl
IEEE Trans. Ind. Informatics
Mengqing Wu, Xiaofeng Zhou, Shuai Li, Haibo Shi
2025 J jnl
IEEE Trans Autom. Sci. Eng.
Naiju Zhai, Xiaofeng Zhou, Shuai Li, Haibo Shi
2025 J jnl
Int. J. Simul. Process. Model.
Yuehan Liu, Haibo Shi, Chang Liu
2025 J jnl
Neural Networks
Haoran Wang, Yaoru Sun, Zeshen Tang, Haibo Shi, Chenyuan Jiao
2024 A* conf
EMNLP
Junda Zhu, Lingyong Yan, Haibo Shi, Dawei Yin, Lei Sha
2024 J jnl
CoRR
Junda Zhu, Lingyong Yan, Haibo Shi, Dawei Yin, Lei Sha
2024 J jnl
CoRR
Zhengliang Shi, Shen Gao, Xiuyi Chen, Yue Feng, Lingyong Yan, Haibo Shi, Dawei Yin, Zhumin Chen, Suzan Verberne, Zhaochun Ren
2024 J jnl
Remote. Sens.
Haibo Shi, Peng Chen, Xianglei Liu, Zhonghua Hong, Zhen Ye, Yi Gao, Ziqi Liu, Xiaohua Tong
2024 conf
EMNLP (Industry Track)
Wenjie Zhou, Zhenxin Ding, Xiaodong Zhang, Haibo Shi, Junfeng Wang, Dawei Yin
2024 J jnl
CoRR
Wenjie Zhou, Zhenxin Ding, Xiaodong Zhang, Haibo Shi, Junfeng Wang, Dawei Yin
2024 A* conf
EMNLP
Yougang Lyu, Lingyong Yan, Shuaiqiang Wang, Haibo Shi, Dawei Yin, Pengjie Ren, Zhumin Chen, Maarten de Rijke, Zhaochun Ren
2024 J jnl
CoRR
Yougang Lyu, Lingyong Yan, Shuaiqiang Wang, Haibo Shi, Dawei Yin, Pengjie Ren, Zhumin Chen, Maarten de Rijke, Zhaochun Ren
2024 J jnl
IEEE Trans. Multim.
Siyu Zhang, Yeming Chen, Yaoru Sun, Fang Wang, Haibo Shi, Haoran Wang
2024 conf
EMNLP (Findings)
Zhengliang Shi, Shen Gao, Xiuyi Chen, Yue Feng, Lingyong Yan, Haibo Shi, Dawei Yin, Pengjie Ren, Suzan Verberne, Zhaochun Ren
2024 J jnl
CoRR
Zhengliang Shi, Shen Gao, Xiuyi Chen, Lingyong Yan, Haibo Shi, Dawei Yin, Zhumin Chen, Pengjie Ren, Suzan Verberne, Zhaochun Ren
2024 J jnl
CoRR
Guanying Jiang, Lingyong Yan, Haibo Shi, Dawei Yin
2023 conf
ICSI (1)
Yang Liu, TieJun Xing, Yuee Zhou, Nan Li, Lianbo Ma, Yingyou Wen, Chang Liu, Haibo Shi
2023 J jnl
Mach. Learn. Sci. Technol.
Dongni Jia, Xiaofeng Zhou, Shuai Li, Shurui Liu, Haibo Shi
2023 J jnl
CoRR
Siyu Zhang, Yeming Chen, Yaoru Sun, Fang Wang, Haibo Shi, Haoran Wang
2023 conf
ICSI (1)
Xiaojie An, Lianbo Ma, Yuee Zhou, Nan Li, TieJun Xing, Yingyou Wen, Chang Liu, Haibo Shi
2023 J jnl
IEEE Trans. Instrum. Meas.
Naiju Zhai, Xiaofeng Zhou, Shuai Li, Haibo Shi
2023 J jnl
IEEE Trans. Ind. Informatics
Xiaofeng Zhou, Naiju Zhai, Shuai Li, Haibo Shi
2023 J jnl
Appl. Intell.
Gang Wang, Haibo Shi, Yufei Chen, Bin Wu
2022 J jnl
Neurocomputing
Shiqing Gao, Haibo Shi, Fang Wang, Zijian Wang, Siyu Zhang, Yunxia Li, Yaoru Sun
2022 J jnl
Int. J. Intell. Syst.
Zijian Wang, Yanting Zhang, Haibo Shi, Lei Cao, Cairong Yan, Guangwei Xu
2021 J jnl
IEEE Access
Yang Cao, Haibo Shi
2021 B conf
ICTAI
Shiqing Gao, Fufei Yao, Yaoru Sun, Haibo Shi
2021 B conf
ICTAI
Gang Wang, Haibo Shi, Yufei Chen
2020 J jnl
IEEE Access
Daming Wang, Yaoru Sun, Haibo Shi, Fang Wang
2020 J jnl
Neurocomputing
Yaoru Sun, Haibo Shi, Fang Wang
2020 J jnl
Int. J. Comput. Appl. Technol.
Jingjing Yin, Haibo Shi, Xiaofeng Zhou, Liang Jin, Shuai Li, Yichi Zhang
2019 J jnl
IEEE Access
Haibo Shi, Yaoru Sun, Guangyuan Li, Fang Wang, Daming Wang, Jie Li
2019 J jnl
Int. J. Model. Identif. Control.
Zhonghua Han, Quan Zhang, Haibo Shi, Yuanwei Qi, Liangliang Sun
2018 J jnl
Comput. Intell. Neurosci.
Haibo Shi, Yaoru Sun, Jie Li
2018 J jnl
CoRR
Zhonghua Han, Quan Zhang, Haibo Shi, Yuanwei Qi, Liangliang Sun
2017 J jnl
Comput. Ind. Eng.
Shuai Li, Xiaofeng Zhou, Fucheng Pan, Haibo Shi, Kaituo Li, Zhongwei Wang
2016 J jnl
Inf. Sci.
Min Huang, Liang Ren, Loo Hay Lee, Xingwei Wang, Hanbin Kuang, Haibo Shi
2015 J jnl
Int. J. Model. Identif. Control.
Yang Cao, Haibo Shi
2015 J jnl
Int. J. Model. Identif. Control.
Yonglai Zhang, Xiaofeng Zhou, Haibo Shi, Zeyu Zheng, Shuai Li
2015 A conf
BMVC
Haibo Shi, Yong Luo, Chao Xu, Yonggang Wen
2015 conf
ICIA
Fucheng Pan, Haibo Shi, Bin Duan
2014 J jnl
Biomed. Signal Process. Control.
Xiaofeng Zhou, Yonglai Zhang, Mingrui Shi, Haibo Shi, Zeyu Zheng
2014 J jnl
Int. J. Model. Identif. Control.
Zhonghua Han, Haibo Shi, Lili Yao
2013 J jnl
Int. J. Model. Identif. Control.
Dechang Sun, Haibo Shi, Zhonghua Han, Chang Liu
2012 J jnl
Int. J. Model. Identif. Control.
Zhonghua Han, Haibo Shi, Feng Qiao, Lei Yue
2012 J jnl
Int. J. Model. Identif. Control.
Shuo Lin, Haibo Shi, Xisheng Lv, Zhonghua Han
2011 conf
ISNN (1)
Jinli Xie, Zhijie Wang, Haibo Shi
2011 conf
ISNN (1)
Haibo Shi, Zhijie Wang, Jinli Xie, Chongbin Guo
2010 Misc conf
ICNC
Jinli Xie, Zhijie Wang, Haibo Shi
2008 B conf
FUZZ-IEEE
Jie Yuan, Haibo Shi, Chang Liu, Wenli Shang
2008 conf
ISCSCT (2)
Wenli Shang, Bin Duan, Haibo Shi
2008 B conf
FUZZ-IEEE
Chang Liu, Haibo Shi, Jie Yuan
2007 C conf
KSEM
Hui Peng, Wenli Shang, Haibo Shi, Wei Peng
2007 J jnl
Int. J. Comput. Integr. Manuf.
Hong Song, Haibo Shi, Chunjie Zhao, Chang Liu
1992 conf
PROLAMAT
Haibo Shi, Jinsong Xue
tests/unit/test_apk_code_analyzer.py
← Index tests/unit/test_apk_code_analyzer.py python
"""
Unit tests for APK code analysis components.

Tests APKCodeAnalyzer, LibraryFilter, SmaliParser, method enumeration,
library filtering, and method key matching. All external tools are mocked.
"""
import os
import tempfile
import textwrap

import pytest
from unittest.mock import MagicMock, patch, PropertyMock

pytestmark = [pytest.mark.unit, pytest.mark.apk, pytest.mark.decompile]


def _has_mmh3():
    try:
        import mmh3  # noqa: F401
        return True
    except ImportError:
        return False


# ============================================================================
# LibraryFilter Tests
# ============================================================================

class TestLibraryFilter:
    """Tests for package-based library filtering."""

    def test_default_prefixes_loaded(self):
        from redb.extractors.decompiler.apk.library_filter import LibraryFilter
        lf = LibraryFilter()
        assert lf.is_library("android.app.Activity")
        assert lf.is_library("androidx.core.app.NotificationCompat")

    def test_user_class_not_filtered(self):
        from redb.extractors.decompiler.apk.library_filter import LibraryFilter
        lf = LibraryFilter()
        assert not lf.is_library("com.example.myapp.MainActivity")
        assert not lf.is_library("org.myorg.MyClass")

    def test_dalvik_descriptor_format(self):
        from redb.extractors.decompiler.apk.library_filter import LibraryFilter
        lf = LibraryFilter()
        assert lf.is_library("Landroid/app/Activity;")
        assert lf.is_library("Landroidx/core/app/NotificationCompat;")
        assert not lf.is_library("Lcom/example/myapp/MainActivity;")

    def test_custom_prefixes(self):
        from redb.extractors.decompiler.apk.library_filter import LibraryFilter
        lf = LibraryFilter(prefixes=["com.custom."])
        assert lf.is_library("com.custom.SomeClass")
        assert not lf.is_library("android.app.Activity")

    def test_env_var_override(self):
        from redb.extractors.decompiler.apk.library_filter import LibraryFilter
        with patch.dict(os.environ, {"APK_LIBRARY_PREFIXES": "my.lib.,other.lib."}):
            lf = LibraryFilter()
            assert lf.is_library("my.lib.SomeClass")
            assert lf.is_library("other.lib.AnotherClass")
            assert not lf.is_library("android.app.Activity")

    def test_filter_stats(self):
        from redb.extractors.decompiler.apk.library_filter import LibraryFilter
        lf = LibraryFilter()
        lf.is_library("android.app.Activity")
        lf.is_library("com.example.MyClass")
        lf.is_library("kotlin.Unit")
        stats = lf.get_filter_stats()
        assert stats["library"] == 2
        assert stats["user"] == 1

    def test_normalize_class_name(self):
        from redb.extractors.decompiler.apk.library_filter import LibraryFilter
        assert LibraryFilter._normalize_class_name("Lcom/example/Foo;") == "com.example.Foo"
        assert LibraryFilter._normalize_class_name("com.example.Foo") == "com.example.Foo"
        assert LibraryFilter._normalize_class_name("com/example/Foo") == "com.example.Foo"

    def test_google_libraries_filtered(self):
        from redb.extractors.decompiler.apk.library_filter import LibraryFilter
        lf = LibraryFilter()
        assert lf.is_library("com.google.android.gms.ads.AdView")
        assert lf.is_library("com.google.firebase.messaging.FirebaseMessagingService")
        assert lf.is_library("com.google.gson.Gson")

    def test_kotlin_filtered(self):
        from redb.extractors.decompiler.apk.library_filter import LibraryFilter
        lf = LibraryFilter()
        assert lf.is_library("kotlin.Unit")
        assert lf.is_library("kotlinx.coroutines.CoroutineScope")

    def test_third_party_libs_filtered(self):
        from redb.extractors.decompiler.apk.library_filter import LibraryFilter
        lf = LibraryFilter()
        assert lf.is_library("com.squareup.okhttp3.OkHttpClient")
        assert lf.is_library("io.reactivex.Observable")
        assert lf.is_library("org.apache.commons.io.IOUtils")

    def test_empty_class_name(self):
        from redb.extractors.decompiler.apk.library_filter import LibraryFilter
        lf = LibraryFilter()
        assert not lf.is_library("")


# ============================================================================
# SmaliParser Tests
# ============================================================================

class TestSmaliParser:
    """Tests for smali file parsing."""

    SAMPLE_SMALI = textwrap.dedent("""\
        .class public Lcom/example/MyClass;
        .super Ljava/lang/Object;

        .method public constructor <init>()V
            .registers 1
            invoke-direct {p0}, Ljava/lang/Object;-><init>()V
            return-void
        .end method

        .method public onCreate(Landroid/os/Bundle;)V
            .registers 4
            .param p1, "savedInstanceState"

            .line 10
            invoke-super {p0, p1}, Landroid/app/Activity;->onCreate(Landroid/os/Bundle;)V
            const/high16 v0, 0x7f090000
            invoke-virtual {p0, v0}, Lcom/example/MyClass;->setContentView(I)V
            const-string v1, "hello"
            invoke-virtual {p0, v1}, Lcom/example/MyClass;->log(Ljava/lang/String;)V
            return-void
        .end method

        .method public abstract doSomething()V
        .end method

        .method public native nativeMethod()V
        .end method
    """)

    def test_parse_smali_content(self):
        from redb.extractors.decompiler.apk.smali_parser import SmaliParser
        methods = SmaliParser._parse_smali_content(self.SAMPLE_SMALI)
        # abstract and native should be skipped
        assert len(methods) == 2

    def test_method_names_extracted(self):
        from redb.extractors.decompiler.apk.smali_parser import SmaliParser
        methods = SmaliParser._parse_smali_content(self.SAMPLE_SMALI)
        names = {m.method_name for m in methods}
        assert "<init>" in names
        assert "onCreate" in names

    def test_class_name_extracted(self):
        from redb.extractors.decompiler.apk.smali_parser import SmaliParser
        methods = SmaliParser._parse_smali_content(self.SAMPLE_SMALI)
        for m in methods:
            assert m.class_name == "Lcom/example/MyClass;"

    def test_method_signature(self):
        from redb.extractors.decompiler.apk.smali_parser import SmaliParser
        methods = SmaliParser._parse_smali_content(self.SAMPLE_SMALI)
        oncreate = [m for m in methods if m.method_name == "onCreate"][0]
        assert oncreate.method_signature == "(Landroid/os/Bundle;)V"

    def test_instruction_count(self):
        from redb.extractors.decompiler.apk.smali_parser import SmaliParser
        methods = SmaliParser._parse_smali_content(self.SAMPLE_SMALI)
        oncreate = [m for m in methods if m.method_name == "onCreate"][0]
        # invoke-super, const/high16, invoke-virtual, const-string, invoke-virtual, return-void
        assert oncreate.instruction_count == 6

    def test_register_count(self):
        from redb.extractors.decompiler.apk.smali_parser import SmaliParser
        methods = SmaliParser._parse_smali_content(self.SAMPLE_SMALI)
        oncreate = [m for m in methods if m.method_name == "onCreate"][0]
        assert oncreate.register_count == 4

    def test_register_count_locals_directive(self):
        """apktool outputs .locals by default, not .registers."""
        from redb.extractors.decompiler.apk.smali_parser import SmaliParser
        smali_with_locals = textwrap.dedent("""\
            .class public Lcom/example/Locals;
            .super Ljava/lang/Object;

            .method public doWork()V
                .locals 3
                const/4 v0, 0x0
                const/4 v1, 0x1
                add-int v2, v0, v1
                return-void
            .end method
        """)
        methods = SmaliParser._parse_smali_content(smali_with_locals)
        assert len(methods) == 1
        assert methods[0].register_count == 3

    def test_abstract_native_skipped(self):
        from redb.extractors.decompiler.apk.smali_parser import SmaliParser
        methods = SmaliParser._parse_smali_content(self.SAMPLE_SMALI)
        names = {m.method_name for m in methods}
        assert "doSomething" not in names
        assert "nativeMethod" not in names

    def test_normalize_smali_body(self):
        from redb.extractors.decompiler.apk.smali_parser import SmaliParser
        body = textwrap.dedent("""\
            .registers 4
            .line 10
            # comment
            invoke-super {p0, p1}, Landroid/app/Activity;->onCreate(Landroid/os/Bundle;)V

            const-string v1, "hello"
        """)
        normalized = SmaliParser.normalize_smali_body(body)
        assert ".line" not in normalized
        assert "# comment" not in normalized
        assert "invoke-super" in normalized
        assert "const-string" in normalized

    def test_count_instructions(self):
        from redb.extractors.decompiler.apk.smali_parser import SmaliParser
        body = textwrap.dedent("""\
            .registers 4
            .param p1, "x"
            invoke-virtual {p0}, Lcom/example/Foo;->bar()V
            const/4 v0, 0x0
            :label_0
            if-eqz v0, :label_1
            return-void
            :label_1
            goto :label_0
        """)
        assert SmaliParser.count_instructions(body) == 5

    def test_make_method_key(self):
        from redb.extractors.decompiler.apk.smali_parser import SmaliParser
        key = SmaliParser.make_method_key(
            "Lcom/example/Foo;", "bar", "(I)V"
        )
        assert key == "Lcom/example/Foo;->bar(I)V"

    def test_parse_smali_file(self):
        from redb.extractors.decompiler.apk.smali_parser import SmaliParser
        with tempfile.NamedTemporaryFile(
            mode="w", suffix=".smali", delete=False
        ) as f:
            f.write(self.SAMPLE_SMALI)
            f.flush()
            try:
                methods = SmaliParser.parse_smali_file(f.name)
                assert len(methods) == 2
            finally:
                os.unlink(f.name)

    def test_parse_smali_directory(self):
        from redb.extractors.decompiler.apk.smali_parser import SmaliParser
        with tempfile.TemporaryDirectory() as tmpdir:
            # Create nested dir structure
            pkg_dir = os.path.join(tmpdir, "com", "example")
            os.makedirs(pkg_dir)
            with open(os.path.join(pkg_dir, "MyClass.smali"), "w") as f:
                f.write(self.SAMPLE_SMALI)

            result = SmaliParser.parse_smali_directory(tmpdir)
            assert len(result) == 2
            # Check keys are in expected format
            for key in result:
                assert "->" in key

    def test_empty_smali_file(self):
        from redb.extractors.decompiler.apk.smali_parser import SmaliParser
        methods = SmaliParser._parse_smali_content("")
        assert methods == []

    def test_class_only_no_methods(self):
        from redb.extractors.decompiler.apk.smali_parser import SmaliParser
        content = ".class public Lcom/example/Empty;\n.super Ljava/lang/Object;\n"
        methods = SmaliParser._parse_smali_content(content)
        assert methods == []


# ============================================================================
# MethodExtractor Tests (Hashing, Obfuscation, Type Conversion)
# ============================================================================

class TestMethodExtractor:
    """Tests for method-level content extraction and hashing."""

    def test_compute_sha256(self):
        from redb.extractors.decompiler.apk.method_extractor import compute_sha256
        h = compute_sha256("test content")
        assert len(h) == 64
        assert h == compute_sha256("test content")  # deterministic

    def test_compute_sha256_different_input(self):
        from redb.extractors.decompiler.apk.method_extractor import compute_sha256
        assert compute_sha256("a") != compute_sha256("b")

    def test_compute_ssdeep_short_input(self):
        from redb.extractors.decompiler.apk.method_extractor import compute_ssdeep
        result = compute_ssdeep("short")
        assert result is None

    def test_compute_ssdeep_long_input(self):
        from redb.extractors.decompiler.apk.method_extractor import compute_ssdeep
        result = compute_ssdeep("x" * 200)
        # May be None depending on entropy, but shouldn't crash
        assert result is None or isinstance(result, str)

    def test_compute_tlsh_short_input(self):
        from redb.extractors.decompiler.apk.method_extractor import compute_tlsh
        result = compute_tlsh("short")
        assert result is None

    def test_compute_tlsh_long_input(self):
        from redb.extractors.decompiler.apk.method_extractor import compute_tlsh
        data = "".join(chr(i % 256) for i in range(200))
        result = compute_tlsh(data)
        assert result is None or isinstance(result, str)

    def test_compute_minhash_returns_none_without_mmh3(self):
        from unittest.mock import patch
        from redb.extractors.decompiler.apk.method_extractor import compute_minhash
        with patch.dict("sys.modules", {"mmh3": None}):
            # Force reimport to hit ImportError
            import importlib
            import redb.extractors.decompiler.apk.method_extractor as mod
            importlib.reload(mod)
            result = mod.compute_minhash("line1\nline2\nline3\nline4")
            # mmh3 may or may not be available; just check it doesn't crash
            assert result is None or isinstance(result, list)
            importlib.reload(mod)  # restore

    def test_compute_minhash_too_few_lines(self):
        from redb.extractors.decompiler.apk.method_extractor import compute_minhash
        result = compute_minhash("invoke-direct {v0}, Lfoo;->bar()V")
        # Only 1 instruction line, need at least 3 for 3-grams
        assert result is None

    @pytest.mark.skipif(
        not _has_mmh3(), reason="mmh3 not installed"
    )
    def test_compute_minhash_deterministic(self):
        from redb.extractors.decompiler.apk.method_extractor import compute_minhash
        smali = (
            "invoke-direct {v1}, Ljava/lang/Object;-><init>()V\n"
            "const-string v0, \"hello\"\n"
            "iput-object v0, v1, LA;->a:Ljava/lang/String;\n"
            "return-void"
        )
        sig1 = compute_minhash(smali)
        sig2 = compute_minhash(smali)
        assert sig1 is not None
        assert sig1 == sig2

    @pytest.mark.skipif(
        not _has_mmh3(), reason="mmh3 not installed"
    )
    def test_compute_minhash_signature_length(self):
        from redb.extractors.decompiler.apk.method_extractor import compute_minhash
        smali = (
            "invoke-direct {v1}, Ljava/lang/Object;-><init>()V\n"
            "const-string v0, \"hello\"\n"
            "iput-object v0, v1, LA;->a:Ljava/lang/String;\n"
            "return-void"
        )
        sig = compute_minhash(smali)
        assert sig is not None
        assert len(sig) == 64  # SIGNATURE_LENGTH

    @pytest.mark.skipif(
        not _has_mmh3(), reason="mmh3 not installed"
    )
    def test_compute_minhash_value_range(self):
        from redb.extractors.decompiler.apk.method_extractor import compute_minhash
        smali = (
            "invoke-direct {v1}, Ljava/lang/Object;-><init>()V\n"
            "const-string v0, \"hello\"\n"
            "iput-object v0, v1, LA;->a:Ljava/lang/String;\n"
            "return-void"
        )
        sig = compute_minhash(smali)
        assert sig is not None
        for val in sig:
            assert 0 <= val < 256  # 8-bit values

    @pytest.mark.skipif(
        not _has_mmh3(), reason="mmh3 not installed"
    )
    def test_compute_minhash_different_inputs_differ(self):
        from redb.extractors.decompiler.apk.method_extractor import compute_minhash
        smali_a = (
            "invoke-direct {v1}, Ljava/lang/Object;-><init>()V\n"
            "const-string v0, \"hello\"\n"
            "iput-object v0, v1, LA;->a:Ljava/lang/String;\n"
            "return-void"
        )
        smali_b = (
            "sget-object v0, Ljava/lang/System;->out:Ljava/io/PrintStream;\n"
            "const-string v1, \"world\"\n"
            "invoke-virtual {v0, v1}, Ljava/io/PrintStream;->println(Ljava/lang/String;)V\n"
            "return-void"
        )
        sig_a = compute_minhash(smali_a)
        sig_b = compute_minhash(smali_b)
        assert sig_a is not None
        assert sig_b is not None
        assert sig_a != sig_b

    @pytest.mark.skipif(
        not _has_mmh3(), reason="mmh3 not installed"
    )
    def test_compute_minhash_skips_directives_and_labels(self):
        from redb.extractors.decompiler.apk.method_extractor import compute_minhash
        # Directives and labels should be filtered out
        smali = (
            ".registers 3\n"
            ":start\n"
            "invoke-direct {v1}, Ljava/lang/Object;-><init>()V\n"
            ".line 10\n"
            "const-string v0, \"hello\"\n"
            "iput-object v0, v1, LA;->a:Ljava/lang/String;\n"
            "return-void"
        )
        smali_clean = (
            "invoke-direct {v1}, Ljava/lang/Object;-><init>()V\n"
            "const-string v0, \"hello\"\n"
            "iput-object v0, v1, LA;->a:Ljava/lang/String;\n"
            "return-void"
        )
        sig_with_directives = compute_minhash(smali)
        sig_clean = compute_minhash(smali_clean)
        assert sig_with_directives == sig_clean

    def test_dalvik_to_java_class(self):
        from redb.extractors.decompiler.apk.method_extractor import dalvik_to_java_class
        assert dalvik_to_java_class("Lcom/example/Foo;") == "com.example.Foo"
        assert dalvik_to_java_class("com.example.Foo") == "com.example.Foo"

    def test_dalvik_type_to_java(self):
        from redb.extractors.decompiler.apk.method_extractor import dalvik_type_to_java
        assert dalvik_type_to_java("V") == "void"
        assert dalvik_type_to_java("I") == "int"
        assert dalvik_type_to_java("Z") == "boolean"
        assert dalvik_type_to_java("Ljava/lang/String;") == "String"
        assert dalvik_type_to_java("[I") == "int[]"
        assert dalvik_type_to_java("[Ljava/lang/String;") == "String[]"

    def test_dalvik_to_java_prototype(self):
        from redb.extractors.decompiler.apk.method_extractor import dalvik_to_java_prototype
        result = dalvik_to_java_prototype("onCreate", "(Landroid/os/Bundle;)V")
        assert result == "void onCreate(Bundle)"

    def test_dalvik_to_java_prototype_no_params(self):
        from redb.extractors.decompiler.apk.method_extractor import dalvik_to_java_prototype
        result = dalvik_to_java_prototype("toString", "()Ljava/lang/String;")
        assert result == "String toString()"

    def test_dalvik_to_java_prototype_multiple_params(self):
        from redb.extractors.decompiler.apk.method_extractor import dalvik_to_java_prototype
        result = dalvik_to_java_prototype("foo", "(ILjava/lang/String;Z)V")
        assert result == "void foo(int, String, boolean)"

    def test_detect_obfuscation_short_name(self):
        from redb.extractors.decompiler.apk.method_extractor import detect_obfuscation_indicators
        result = detect_obfuscation_indicators("a", "Lcom/example/Foo;", "", 10)
        assert result["short_method_name"] is True
        result = detect_obfuscation_indicators("onCreate", "Lcom/example/Foo;", "", 10)
        assert result["short_method_name"] is False

    def test_detect_obfuscation_short_class(self):
        from redb.extractors.decompiler.apk.method_extractor import detect_obfuscation_indicators
        result = detect_obfuscation_indicators("foo", "Lcom/example/a;", "", 10)
        assert result["short_class_name"] is True
        result = detect_obfuscation_indicators("foo", "Lcom/example/MainActivity;", "", 10)
        assert result["short_class_name"] is False

    def test_detect_reflection_calls(self):
        from redb.extractors.decompiler.apk.method_extractor import detect_obfuscation_indicators
        smali = "invoke-virtual {v0}, Ljava/lang/reflect/Method;->invoke(Ljava/lang/Object;[Ljava/lang/Object;)Ljava/lang/Object;"
        result = detect_obfuscation_indicators("foo", "Lcom/Foo;", smali, 10)
        assert result["has_reflection_calls"] is True

    def test_detect_no_reflection(self):
        from redb.extractors.decompiler.apk.method_extractor import detect_obfuscation_indicators
        smali = "invoke-virtual {p0}, Lcom/example/Foo;->bar()V"
        result = detect_obfuscation_indicators("foo", "Lcom/Foo;", smali, 10)
        assert result["has_reflection_calls"] is False

    def test_detect_string_encryption(self):
        from redb.extractors.decompiler.apk.method_extractor import detect_obfuscation_indicators
        smali = 'const-string v0, "encrypted"\n    invoke-static {v0}, Lcom/example/Crypto;->decrypt(Ljava/lang/String;)Ljava/lang/String;'
        result = detect_obfuscation_indicators("foo", "Lcom/Foo;", smali, 10)
        assert result["has_string_encryption"] is True

    def test_detect_excessive_goto(self):
        from redb.extractors.decompiler.apk.method_extractor import detect_obfuscation_indicators
        smali = "\n".join(["goto :label_0"] * 20)
        result = detect_obfuscation_indicators("foo", "Lcom/Foo;", smali, 30)
        # threshold = max(5, 30*0.15) = 5, 20 > 5
        assert result["excessive_goto_count"] is True

    def test_detect_no_excessive_goto(self):
        from redb.extractors.decompiler.apk.method_extractor import detect_obfuscation_indicators
        smali = "goto :label_0\nreturn-void"
        result = detect_obfuscation_indicators("foo", "Lcom/Foo;", smali, 100)
        assert result["excessive_goto_count"] is False

    def test_parse_dalvik_params_empty(self):
        from redb.extractors.decompiler.apk.method_extractor import _parse_dalvik_params
        assert _parse_dalvik_params("") == []

    def test_parse_dalvik_params_primitives(self):
        from redb.extractors.decompiler.apk.method_extractor import _parse_dalvik_params
        result = _parse_dalvik_params("IZJ")
        assert result == ["I", "Z", "J"]

    def test_parse_dalvik_params_object(self):
        from redb.extractors.decompiler.apk.method_extractor import _parse_dalvik_params
        result = _parse_dalvik_params("Ljava/lang/String;")
        assert result == ["Ljava/lang/String;"]

    def test_parse_dalvik_params_mixed(self):
        from redb.extractors.decompiler.apk.method_extractor import _parse_dalvik_params
        result = _parse_dalvik_params("ILjava/lang/String;Z")
        assert result == ["I", "Ljava/lang/String;", "Z"]

    def test_parse_dalvik_params_array(self):
        from redb.extractors.decompiler.apk.method_extractor import _parse_dalvik_params
        result = _parse_dalvik_params("[I[Ljava/lang/String;")
        assert result == ["[I", "[Ljava/lang/String;"]


# ============================================================================
# APKCodeAnalyzer Tests (mocked tools)
# ============================================================================

class TestAPKCodeAnalyzer:
    """Tests for the APKCodeAnalyzer orchestration."""

    def test_init_defaults(self):
        from redb.extractors.decompiler.apk.analyzer import APKCodeAnalyzer
        analyzer = APKCodeAnalyzer("/fake/path.apk")
        assert analyzer.filepath == "/fake/path.apk"
        assert analyzer.timeout == 600
        assert analyzer.min_instructions == 5

    def test_init_custom_timeout(self):
        from redb.extractors.decompiler.apk.analyzer import APKCodeAnalyzer
        analyzer = APKCodeAnalyzer("/fake/path.apk", timeout=300)
        assert analyzer.timeout == 300

    def test_init_min_instructions_env(self):
        from redb.extractors.decompiler.apk.analyzer import APKCodeAnalyzer
        with patch.dict(os.environ, {"APK_MIN_METHOD_INSTRUCTIONS": "10"}):
            analyzer = APKCodeAnalyzer("/fake/path.apk")
            assert analyzer.min_instructions == 10

    @patch("redb.extractors.decompiler.apk.analyzer.APKCodeAnalyzer._run_androguard")
    @patch.object(
        __import__("redb.extractors.decompiler.apk.apktool_wrapper", fromlist=["ApktoolDisassembler"]).ApktoolDisassembler,
        "disassemble",
        return_value=False,
    )
    @patch.object(
        __import__("redb.extractors.decompiler.apk.jadx_wrapper", fromlist=["JADXDecompiler"]).JADXDecompiler,
        "decompile",
        return_value=False,
    )
    def test_extract_returns_dict_keys(self, mock_jadx, mock_apktool, mock_androguard):
        from redb.extractors.decompiler.apk.analyzer import APKCodeAnalyzer
        mock_analysis = MagicMock()
        mock_analysis.get_methods.return_value = []
        mock_androguard.return_value = (MagicMock(), [], mock_analysis)

        analyzer = APKCodeAnalyzer("/fake/path.apk", log=MagicMock())
        results = analyzer.extract()

        assert "decompiled_content" in results
        assert "decompiled_refs" in results
        assert "smali_content" in results
        assert "smali_refs" in results
        assert "similarity_metrics" in results
        assert "strings" in results
        assert "analysis_errors" in results
        analyzer.cleanup()

    @patch("redb.extractors.decompiler.apk.analyzer.APKCodeAnalyzer._run_androguard")
    def test_extract_androguard_failure(self, mock_androguard):
        from redb.extractors.decompiler.apk.analyzer import APKCodeAnalyzer
        mock_androguard.side_effect = Exception("Analysis failed")

        analyzer = APKCodeAnalyzer("/fake/path.apk", log=MagicMock())
        results = analyzer.extract()

        assert len(results["analysis_errors"]) > 0
        analyzer.cleanup()

    def test_cleanup_removes_temp_dirs(self):
        from redb.extractors.decompiler.apk.analyzer import APKCodeAnalyzer
        analyzer = APKCodeAnalyzer("/fake/path.apk")
        tmpdir = tempfile.mkdtemp()
        analyzer._temp_dirs.append(tmpdir)
        assert os.path.isdir(tmpdir)
        analyzer.cleanup()
        assert not os.path.isdir(tmpdir)

    def test_normalize_java(self):
        from redb.extractors.decompiler.apk.analyzer import _normalize_java
        source = "  public void foo() {\n    int x = 1;\n  }\n"
        normalized = _normalize_java(source)
        assert normalized == "public void foo() {\nint x = 1;\n}"


class TestNormalizeAndroguardOperands:
    """Tests for androguard-to-apktool output normalization."""

    def test_invoke_adds_braces(self):
        from redb.extractors.decompiler.apk.analyzer import _normalize_androguard_operands
        result = _normalize_androguard_operands(
            "invoke-direct",
            "v1, Ljava/lang/Object;-><init>()V"
        )
        assert result == "{v1}, Ljava/lang/Object;-><init>()V"

    def test_invoke_virtual_multiple_regs(self):
        from redb.extractors.decompiler.apk.analyzer import _normalize_androguard_operands
        result = _normalize_androguard_operands(
            "invoke-virtual",
            "v0, v1, Ljava/lang/String;->equals(Ljava/lang/Object;)Z"
        )
        assert result == "{v0, v1}, Ljava/lang/String;->equals(Ljava/lang/Object;)Z"

    def test_invoke_static_no_regs(self):
        from redb.extractors.decompiler.apk.analyzer import _normalize_androguard_operands
        # invoke-static with no register args (rare but possible)
        result = _normalize_androguard_operands(
            "invoke-static",
            "Ljava/lang/System;->gc()V"
        )
        # First part starts with L, so no registers to wrap
        assert result == "Ljava/lang/System;->gc()V"

    def test_invoke_range(self):
        from redb.extractors.decompiler.apk.analyzer import _normalize_androguard_operands
        result = _normalize_androguard_operands(
            "invoke-virtual/range",
            "v3, v4, v5, Landroid/util/Log;->d(Ljava/lang/String; Ljava/lang/String;)I"
        )
        assert result == "{v3, v4, v5}, Landroid/util/Log;->d(Ljava/lang/String; Ljava/lang/String;)I"

    def test_iget_field_colon(self):
        from redb.extractors.decompiler.apk.analyzer import _normalize_androguard_operands
        result = _normalize_androguard_operands(
            "iget-object",
            "v0, v4, LF;->a Ljava/lang/String;"
        )
        assert result == "v0, v4, LF;->a:Ljava/lang/String;"

    def test_iput_field_colon(self):
        from redb.extractors.decompiler.apk.analyzer import _normalize_androguard_operands
        result = _normalize_androguard_operands(
            "iput-object",
            "v0, v1, LA;->b Ljava/lang/String;"
        )
        assert result == "v0, v1, LA;->b:Ljava/lang/String;"

    def test_sget_primitive_field(self):
        from redb.extractors.decompiler.apk.analyzer import _normalize_androguard_operands
        result = _normalize_androguard_operands(
            "sget",
            "v0, Lcom/Foo;->count I"
        )
        assert result == "v0, Lcom/Foo;->count:I"

    def test_sput_array_field(self):
        from redb.extractors.decompiler.apk.analyzer import _normalize_androguard_operands
        result = _normalize_androguard_operands(
            "sput-object",
            "v0, Lcom/Foo;->data [B"
        )
        assert result == "v0, Lcom/Foo;->data:[B"

    def test_non_invoke_non_field_passthrough(self):
        from redb.extractors.decompiler.apk.analyzer import _normalize_androguard_operands
        result = _normalize_androguard_operands(
            "const-string",
            'v0, "hello world"'
        )
        assert result == 'v0, "hello world"'

    def test_move_passthrough(self):
        from redb.extractors.decompiler.apk.analyzer import _normalize_androguard_operands
        result = _normalize_androguard_operands("move", "v0, v4")
        assert result == "v0, v4"

    def test_goto_passthrough(self):
        from redb.extractors.decompiler.apk.analyzer import _normalize_androguard_operands
        result = _normalize_androguard_operands("goto", "+005h")
        assert result == "+005h"


class TestSmaliCFG:
    """Tests for smali CFG construction and metrics computation."""

    def test_empty_body(self):
        from redb.extractors.decompiler.apk.smali_cfg import compute_cfg_metrics
        m = compute_cfg_metrics("")
        assert m.block_count == 0
        assert m.cyclomatic_complexity == 1

    def test_linear_method_apktool(self):
        """Single basic block — no branches."""
        from redb.extractors.decompiler.apk.smali_cfg import compute_cfg_metrics
        smali = textwrap.dedent("""\
            .registers 2
            invoke-direct {v1}, Ljava/lang/Object;-><init>()V
            const-string v0, "hello"
            iput-object v0, v1, LA;->a:Ljava/lang/String;
            return-void
        """)
        m = compute_cfg_metrics(smali)
        assert m.block_count == 1
        assert m.edge_count == 0
        assert m.cyclomatic_complexity == 1
        assert m.loop_count == 0
        assert m.max_depth == 0
        assert m.max_fan_out == 0

    def test_linear_method_androguard(self):
        """Single basic block — androguard format (no labels/directives)."""
        from redb.extractors.decompiler.apk.smali_cfg import compute_cfg_metrics
        smali = (
            "    invoke-direct {v1}, Ljava/lang/Object;-><init>()V\n"
            "    const-string v0, \"hello\"\n"
            "    iput-object v0, v1, LA;->a:Ljava/lang/String;\n"
            "    return-void"
        )
        m = compute_cfg_metrics(smali)
        assert m.block_count == 1
        assert m.edge_count == 0
        assert m.cyclomatic_complexity == 1

    def test_if_branch_apktool(self):
        """Diamond pattern: if-else with two paths merging."""
        from redb.extractors.decompiler.apk.smali_cfg import compute_cfg_metrics
        smali = textwrap.dedent("""\
            .registers 2
            const/4 v0, 0x0
            if-eqz v0, :cond_0
            const/4 v0, 0x1
            goto :goto_0
            :cond_0
            const/4 v0, 0x2
            :goto_0
            return v0
        """)
        m = compute_cfg_metrics(smali)
        # Blocks: [const,if] [const,goto] [const] [return]
        assert m.block_count == 4
        assert m.cyclomatic_complexity == 2  # E - N + 2
        assert m.loop_count == 0
        assert m.max_fan_out == 2  # if-branch has 2 successors

    def test_loop_apktool(self):
        """Simple loop: goto back to earlier label."""
        from redb.extractors.decompiler.apk.smali_cfg import compute_cfg_metrics
        smali = textwrap.dedent("""\
            .registers 2
            const/4 v0, 0x0
            :loop_start
            add-int/lit8 v0, v0, 0x1
            const/16 v1, 0xa
            if-lt v0, v1, :loop_start
            return v0
        """)
        m = compute_cfg_metrics(smali)
        assert m.loop_count == 1  # back edge from if-lt to loop_start
        assert m.cyclomatic_complexity >= 2

    def test_if_branch_androguard(self):
        """If-else in androguard offset format."""
        from redb.extractors.decompiler.apk.smali_cfg import compute_cfg_metrics
        smali = (
            "    const/4 v0, 0x0\n"
            "    if-eqz v0, +2h\n"        # skip 2 instructions ahead
            "    const/4 v0, 0x1\n"
            "    goto +1h\n"               # skip 1 ahead
            "    const/4 v0, 0x2\n"
            "    return v0"
        )
        m = compute_cfg_metrics(smali)
        assert m.block_count >= 3
        assert m.cyclomatic_complexity >= 2

    def test_multiple_if_branches(self):
        """Multiple if-branches increase cyclomatic complexity."""
        from redb.extractors.decompiler.apk.smali_cfg import compute_cfg_metrics
        smali = textwrap.dedent("""\
            .registers 3
            if-eqz v0, :cond_0
            if-eqz v1, :cond_1
            const/4 v2, 0x0
            goto :end
            :cond_0
            const/4 v2, 0x1
            goto :end
            :cond_1
            const/4 v2, 0x2
            :end
            return v2
        """)
        m = compute_cfg_metrics(smali)
        # Two if-branches: CC should be >= 3
        assert m.cyclomatic_complexity >= 3

    def test_return_terminates_block(self):
        """Return instruction terminates block with no successor."""
        from redb.extractors.decompiler.apk.smali_cfg import compute_cfg_metrics
        smali = textwrap.dedent("""\
            .registers 1
            const/4 v0, 0x0
            return v0
        """)
        m = compute_cfg_metrics(smali)
        assert m.block_count == 1
        assert m.edge_count == 0

    def test_throw_terminates_block(self):
        """Throw instruction terminates block with no successor."""
        from redb.extractors.decompiler.apk.smali_cfg import compute_cfg_metrics
        smali = textwrap.dedent("""\
            .registers 1
            new-instance v0, Ljava/lang/RuntimeException;
            invoke-direct {v0}, Ljava/lang/RuntimeException;-><init>()V
            throw v0
        """)
        m = compute_cfg_metrics(smali)
        assert m.block_count == 1
        assert m.edge_count == 0

    def test_max_depth_linear_chain(self):
        """Chain of blocks gives increasing depth."""
        from redb.extractors.decompiler.apk.smali_cfg import compute_cfg_metrics
        # 3 blocks chained: entry -> if -> fall-through -> return
        smali = textwrap.dedent("""\
            .registers 2
            const/4 v0, 0x0
            if-nez v0, :cond_0
            const/4 v0, 0x1
            :cond_0
            return v0
        """)
        m = compute_cfg_metrics(smali)
        assert m.max_depth >= 1

    def test_nested_loop_apktool(self):
        """Nested loop: two back edges."""
        from redb.extractors.decompiler.apk.smali_cfg import compute_cfg_metrics
        smali = textwrap.dedent("""\
            .registers 3
            const/4 v0, 0x0
            :outer
            const/4 v1, 0x0
            :inner
            add-int/lit8 v1, v1, 0x1
            if-lt v1, v2, :inner
            add-int/lit8 v0, v0, 0x1
            if-lt v0, v2, :outer
            return-void
        """)
        m = compute_cfg_metrics(smali)
        assert m.loop_count == 2

    def test_consistency_formula(self):
        """Verify E - N + 2 formula matches cyclomatic_complexity."""
        from redb.extractors.decompiler.apk.smali_cfg import compute_cfg_metrics
        smali = textwrap.dedent("""\
            .registers 2
            const/4 v0, 0x0
            if-eqz v0, :cond_0
            const/4 v0, 0x1
            goto :goto_0
            :cond_0
            const/4 v0, 0x2
            :goto_0
            return v0
        """)
        m = compute_cfg_metrics(smali)
        expected_cc = m.edge_count - m.block_count + 2
        assert m.cyclomatic_complexity == max(1, expected_cc)