Add duplicate file analysis for Android

Author: runningcode

src/launchpad/analyzers/android.py

@@ -8,7 +8,9 @@
 from ..artifacts.android.zipped_aab import ZippedAAB
 from ..artifacts.android.zipped_apk import ZippedAPK
 from ..artifacts.artifact import AndroidArtifact
-from ..models.android import AndroidAnalysisResults, AndroidAppInfo
+from ..insights.common import DuplicateFilesInsight
+from ..insights.insight import InsightsInput
+from ..models.android import AndroidAnalysisResults, AndroidAppInfo, AndroidInsightResults
 from ..models.common import FileAnalysis, FileInfo
 from ..models.treemap import FILE_TYPE_TO_TREEMAP_TYPE, TreemapType
 from ..utils.file_utils import calculate_file_hash
@@ -25,6 +27,15 @@
 class AndroidAnalyzer:
     """Analyzer for Android apps (.apk, .aab files)."""
 
+    def __init__(
+        self,
+        skip_insights: bool = False,
+    ) -> None:
+        """Args:
+        skip_insights: Skip insights generation for faster analysis
+        """
+        self.skip_insights = skip_insights
+
     def analyze(self, artifact: AndroidArtifact) -> AndroidAnalysisResults:
         manifest_dict = artifact.get_manifest().model_dump()
         start_time = time.time()
@@ -61,13 +72,26 @@ def analyze(self, artifact: AndroidArtifact) -> AndroidAnalysisResults:
 
         treemap = treemap_builder.build_file_treemap(file_analysis)
 
+        insights: AndroidInsightResults | None = None
+        if not self.skip_insights:
+            logger.info("Generating insights from analysis results")
+            insights_input = InsightsInput(
+                app_info=app_info,
+                file_analysis=file_analysis,
+                treemap=treemap,
+            )
+            insights = AndroidInsightResults(
+                duplicate_files=DuplicateFilesInsight().generate(insights_input),
+            )
+
         analysis_duration = time.time() - start_time
         return AndroidAnalysisResults(
             generated_at=datetime.now(timezone.utc),
             analysis_duration=analysis_duration,
             app_info=app_info,
             treemap=treemap,
             file_analysis=file_analysis,
+            insights=insights,
         )
 
     def _get_file_analysis(self, apks: list[APK]) -> FileAnalysis:
@@ -114,7 +138,7 @@ def _get_file_analysis(self, apks: list[APK]) -> FileAnalysis:
                             size=merged_size,
                             file_type=file_type,
                             treemap_type=treemap_type,
-                            # Intentionally igoring hash of merged file
+                            # Intentionally ignoring hash of merged file
                             hash_md5="",
                         )
                         path_to_file_info[relative_path] = merged_file_info

src/launchpad/artifacts/__init__.py

@@ -0,0 +1,3 @@
+from .artifact_factory import ArtifactFactory
+
+__all__ = ["ArtifactFactory"]

src/launchpad/cli.py

@@ -217,6 +217,10 @@ def android(
         else:
             _print_android_table_output(results, quiet)
 
+        if not quiet:
+            console.print(f"\n[bold green]✓[/bold green] Analysis completed in {duration:.2f}s")
+            _print_android_summary(results)
+
     except Exception as e:
         if verbose:
             console.print_exception()
@@ -377,6 +381,24 @@ def _print_file_analysis_table(file_analysis: FileAnalysis) -> None:
     console.print()
 
 
+def _print_android_summary(results: AndroidAnalysisResults) -> None:
+    """Print a brief summary of the analysis."""
+    file_analysis = results.file_analysis
+    insights = results.insights
+
+    console.print("\n[bold]Summary:[/bold]")
+    console.print(f"• App name: [cyan]{results.app_info.name}[/cyan]")
+    console.print(f"• Package name: [cyan]{results.app_info.package_name}[/cyan]")
+    console.print(f"• Total app size: [cyan]{_format_bytes(file_analysis.total_size)}[/cyan]")
+    console.print(f"• File count: [cyan]{file_analysis.file_count:,}[/cyan]")
+
+    if insights and insights.duplicate_files and insights.duplicate_files.total_savings > 0:
+        console.print(
+            f"• Potential savings from duplicates: "
+            f"[yellow]{_format_bytes(insights.duplicate_files.total_savings)}[/yellow]"
+        )
+
+
 def _print_apple_summary(results: AppleAnalysisResults) -> None:
     """Print a brief summary of the analysis."""
     file_analysis = results.file_analysis

src/launchpad/insights/common.py

@@ -12,6 +12,9 @@
 
 class DuplicateFilesInsight(Insight[DuplicateFilesInsightResult]):
 
+    def __call__(self, input: InsightsInput) -> DuplicateFilesInsightResult:
+        return self.generate(input)
+
     def generate(self, input: InsightsInput) -> DuplicateFilesInsightResult:
         # Group files by hash
         files_by_hash: Dict[str, List[FileInfo]] = defaultdict(list)

src/launchpad/insights/insight.py

@@ -1,20 +1,18 @@
-from abc import abstractmethod
 from dataclasses import dataclass
-from typing import Protocol, TypeVar
+from typing import Protocol, Sequence, TypeVar
 
-from ..models.apple import AppleAppInfo, MachOBinaryAnalysis
-from ..models.common import FileAnalysis
-from ..models.treemap import TreemapResults
+from launchpad.models.common import BaseAppInfo, BaseBinaryAnalysis, FileAnalysis
+from launchpad.models.treemap import TreemapResults
 
 T_co = TypeVar("T_co", covariant=True)
 
 
 @dataclass
 class InsightsInput:
-    app_info: AppleAppInfo
+    app_info: BaseAppInfo
     file_analysis: FileAnalysis
     treemap: TreemapResults | None
-    binary_analysis: list[MachOBinaryAnalysis]
+    binary_analysis: Sequence[BaseBinaryAnalysis] | None = None
 
 
 class Insight(Protocol[T_co]):
@@ -24,8 +22,7 @@ class Insight(Protocol[T_co]):
     All data needed for the insight must be collected during the main analysis phase.
     """
 
-    @abstractmethod
-    def generate(self, input: InsightsInput) -> T_co:
+    def __call__(self, input: InsightsInput) -> T_co:
         """Generate insights from analysis results.
 
         Args:
@@ -34,4 +31,4 @@ def generate(self, input: InsightsInput) -> T_co:
         Returns:
             Typed insight results
         """
-        raise NotImplementedError("Not implemented")
+        ...

src/launchpad/models/android.py

@@ -1,6 +1,7 @@
-from pydantic import ConfigDict, Field
+from pydantic import BaseModel, ConfigDict, Field
 
 from .common import BaseAnalysisResults, BaseAppInfo
+from .insights import DuplicateFilesInsightResult
 
 
 class AndroidAppInfo(BaseAppInfo):
@@ -9,8 +10,15 @@ class AndroidAppInfo(BaseAppInfo):
     package_name: str = Field(..., description="Android package name")
 
 
-class AndroidAnalysisResults(BaseAnalysisResults):
-    """Complete Android analysis results."""
+class AndroidInsightResults(BaseModel):
+    model_config = ConfigDict(frozen=True)
+
+    duplicate_files: DuplicateFilesInsightResult | None = Field(None, description="Duplicate files analysis")
 
+
+class AndroidAnalysisResults(BaseAnalysisResults):
     model_config = ConfigDict(frozen=True)
     app_info: AndroidAppInfo = Field(..., description="Android app information")
+    insights: AndroidInsightResults | None = Field(
+        description="Generated insights from the analysis",
+    )

tests/unit/test_android_analyzer.py

@@ -0,0 +1,51 @@
+"""Tests for Android analyzer with duplicate file detection."""
+
+from pathlib import Path
+
+import pytest
+
+from launchpad.analyzers.android import AndroidAnalyzer
+from launchpad.artifacts import ArtifactFactory
+
+
+@pytest.fixture
+def test_apk_path() -> Path:
+    return Path("tests/_fixtures/android/hn.apk")
+
+
+@pytest.fixture
+def android_analyzer() -> AndroidAnalyzer:
+    return AndroidAnalyzer()
+
+
+class TestAndroidAnalyzer:
+    def test_analyze_with_duplicate_detection(self, test_apk_path: Path, android_analyzer: AndroidAnalyzer) -> None:
+        """Test that Android analyzer includes duplicate file detection."""
+        artifact = ArtifactFactory.from_path(test_apk_path)
+        results = android_analyzer.analyze(artifact)
+
+        assert results.app_info.name == "Hacker News"
+        assert results.app_info.package_name == "com.emergetools.hackernews"
+        assert results.file_analysis is not None
+        assert len(results.file_analysis.files) > 0
+
+        assert results.insights is not None
+        assert results.insights.duplicate_files is not None
+
+        duplicate_insight = results.insights.duplicate_files
+        assert hasattr(duplicate_insight, "files")
+        assert hasattr(duplicate_insight, "total_savings")
+        assert hasattr(duplicate_insight, "duplicate_count")
+        assert isinstance(duplicate_insight.total_savings, int)
+        assert isinstance(duplicate_insight.duplicate_count, int)
+        assert duplicate_insight.total_savings == 51709
+        assert duplicate_insight.duplicate_count == 52
+
+    def test_duplicate_files_have_hashes(self, test_apk_path: Path, android_analyzer: AndroidAnalyzer) -> None:
+        """Test that all files have MD5 hashes for duplicate detection."""
+        artifact = ArtifactFactory.from_path(test_apk_path)
+        results = android_analyzer.analyze(artifact)
+
+        for file_info in results.file_analysis.files:
+            assert file_info.hash_md5 is not None
+            assert len(file_info.hash_md5) > 0