Another batch of insight fixes

.cursor/rules/overview.mdc

@@ -22,6 +22,8 @@ The iOS analysis code ONLY has to work with `.xcarchive.zip` files as input.
 
 # Python rules
 
+ALWAYS USE THE `.venv/bin/python` VERSION WHEN RUNNING COMMANDS.
+
 For the Python code make sure to follow all of Sentry's best practices, as well as modern Python best practices. Try to use types as much as possible. If standard repo setup is not present, feel free to configure it and add it to the repo since this is currently a bare setup.
 
 For the CLI, make sure to use the `click` library.
@@ -32,4 +34,4 @@ For the Mach-O handling, use the `lief` library and follow best practices for th
 
 Included is a `test/artifacts` directory which contains sample "clean room" apps that can be used for writing integration tests and validating the output of this tool. Always write new tests to validate behavior and functionality. Prefer to write integration tests using the sample apps instead of writing smaller unit tests or using mocks.
 
-Make sure to write tests using `pytest`.
+Make sure to write tests using `pytest`.

src/launchpad/artifacts/apple/zipped_xcarchive.py

@@ -61,7 +61,7 @@ def get_plist(self) -> dict[str, Any]:
                 plist_data = plistlib.load(f)
 
             self._plist = plist_data
-            return self._plist
+            return plist_data
         except Exception as e:
             raise RuntimeError(f"Failed to parse Info.plist: {e}")
 
@@ -223,12 +223,41 @@ def get_all_binary_paths(self) -> List[BinaryInfo]:
                                     extension_name,
                                     extension_binary_path,
                                     extension_dsym_path,
-                                    is_main_binary=False,
+                                    is_main_binary=True,  # App extension main executables are main binaries
                                 )
                             )
                     except Exception as e:
                         logger.warning(f"Failed to read extension Info.plist at {extension_path}: {e}")
 
+        # Find Watch app binaries
+        for watch_path in app_bundle_path.rglob("Watch/*.app"):
+            if watch_path.is_dir():
+                watch_plist_path = watch_path / "Info.plist"
+                if watch_plist_path.exists():
+                    try:
+                        import plistlib
+
+                        with open(watch_plist_path, "rb") as f:
+                            watch_plist = plistlib.load(f)
+                        watch_executable = watch_plist.get("CFBundleExecutable")
+                        if watch_executable:
+                            watch_binary_path = watch_path / watch_executable
+                            watch_name = f"Watch/{watch_path.stem}/{watch_executable}"
+
+                            watch_uuid = self._extract_binary_uuid(watch_binary_path)
+                            watch_dsym_path = dsym_files.get(watch_uuid) if watch_uuid else None
+
+                            binaries.append(
+                                BinaryInfo(
+                                    watch_name,
+                                    watch_binary_path,
+                                    watch_dsym_path,
+                                    is_main_binary=True,  # Watch app main executables are main binaries
+                                )
+                            )
+                    except Exception as e:
+                        logger.warning(f"Failed to read Watch app Info.plist at {watch_path}: {e}")
+
         return binaries
 
     def get_asset_catalog_details(self, relative_path: Path) -> List[AssetCatalogElement]:

src/launchpad/size/analyzers/android.py

@@ -170,7 +170,8 @@ def _get_file_analysis(self, apks: list[APK]) -> FileAnalysis:
                                 size=file_size,
                                 file_type=file_type,
                                 treemap_type=treemap_type,
-                                hash_md5=file_hash,
+                                hash=file_hash,
+                                is_dir=False,
                             )
                             path_to_file_info["Dex"] = merged_dex_info
                             logger.debug("Created merged DEX representation: %s", relative_path)
@@ -190,7 +191,8 @@ def _get_file_analysis(self, apks: list[APK]) -> FileAnalysis:
                                 size=merged_size,
                                 file_type=file_type,
                                 treemap_type=treemap_type,
-                                hash_md5="",
+                                hash="",
+                                is_dir=False,
                             )
                             path_to_file_info["Dex"] = merged_dex_info
                         continue
@@ -216,7 +218,8 @@ def _get_file_analysis(self, apks: list[APK]) -> FileAnalysis:
                             size=merged_size,
                             file_type=file_type,
                             treemap_type=treemap_type,
-                            hash_md5="",
+                            hash="",
+                            is_dir=False,
                         )
                         path_to_file_info[relative_path] = merged_file_info
                     else:
@@ -228,7 +231,8 @@ def _get_file_analysis(self, apks: list[APK]) -> FileAnalysis:
                             size=file_size,
                             file_type=file_type,
                             treemap_type=treemap_type,
-                            hash_md5=file_hash,
+                            hash=file_hash,
+                            is_dir=False,
                         )
                         path_to_file_info[relative_path] = file_info
 
@@ -242,8 +246,12 @@ def _get_file_analysis(self, apks: list[APK]) -> FileAnalysis:
                 files_by_type[file_info.file_type] = []
             files_by_type[file_info.file_type].append(file_info)
 
+        # Separate directories from files (though APKs typically don't have directory entries)
+        directories = [f for f in file_infos if f.is_dir]
+
         return FileAnalysis(
             files=file_infos,
+            directories=directories,
         )
 
     def _get_class_definitions(self, apks: list[APK]) -> list[ClassDefinition]:

src/launchpad/size/analyzers/apple.py

@@ -2,8 +2,6 @@
 
 from __future__ import annotations
 
-import subprocess
-
 from pathlib import Path
 from typing import Any, Dict, List, Tuple
 
@@ -16,7 +14,6 @@
 from launchpad.parsers.apple.macho_symbol_sizes import MachOSymbolSizes
 from launchpad.parsers.apple.objc_symbol_type_aggregator import ObjCSymbolTypeAggregator
 from launchpad.parsers.apple.swift_symbol_type_aggregator import SwiftSymbolTypeAggregator
-from launchpad.size.constants import APPLE_FILESYSTEM_BLOCK_SIZE
 from launchpad.size.hermes.utils import make_hermes_reports
 from launchpad.size.insights.apple.image_optimization import ImageOptimizationInsight
 from launchpad.size.insights.apple.localized_strings import LocalizedStringsInsight
@@ -33,12 +30,11 @@
 from launchpad.size.insights.common.large_images import LargeImageFileInsight
 from launchpad.size.insights.common.large_videos import LargeVideoFileInsight
 from launchpad.size.insights.insight import InsightsInput
-from launchpad.size.models.common import FileAnalysis, FileInfo
-from launchpad.size.models.treemap import FILE_TYPE_TO_TREEMAP_TYPE, TreemapType
 from launchpad.size.treemap.treemap_builder import TreemapBuilder
 from launchpad.size.utils.apple_bundle_size import calculate_bundle_sizes
+from launchpad.size.utils.file_analysis import analyze_apple_files
 from launchpad.utils.apple.code_signature_validator import CodeSignatureValidator
-from launchpad.utils.file_utils import calculate_file_hash, get_file_size, to_nearest_block_size
+from launchpad.utils.file_utils import get_file_size
 from launchpad.utils.logging import get_logger
 from launchpad.utils.performance import trace, trace_ctx
 
@@ -114,7 +110,7 @@ def analyze(self, artifact: AppleArtifact) -> AppleAnalysisResults:
         app_info = self.app_info
         logger.info(f"Analyzing app: {app_info.name} v{app_info.version}")
 
-        file_analysis = self._analyze_files(artifact)
+        file_analysis = analyze_apple_files(artifact)
         logger.info(f"Found {file_analysis.file_count} files, total size: {file_analysis.total_size} bytes")
 
         app_bundle_path = artifact.get_app_bundle_path()
@@ -249,46 +245,6 @@ def _extract_app_info(self, xcarchive: ZippedXCArchive) -> AppleAppInfo:
             code_signature_errors=code_signature_errors,
         )
 
-    @trace("apple.detect_file_type")
-    def _detect_file_type(self, file_path: Path) -> str:
-        """Detect file type using the file command.
-
-        Args:
-            file_path: Path to the file to analyze
-
-        Returns:
-            File type string from file command output, normalized to common types
-        """
-        try:
-            result = subprocess.run(["file", str(file_path)], capture_output=True, text=True, check=True)
-            # Extract just the file type description after the colon
-            file_type = result.stdout.split(":", 1)[1].strip().lower()
-            logger.debug(f"Detected file type for {file_path}: {file_type}")
-
-            # Normalize common file types
-            if "mach-o" in file_type:
-                return "macho"
-            elif "executable" in file_type:
-                return "executable"
-            elif "text" in file_type:
-                return "text"
-            elif "directory" in file_type:
-                return "directory"
-            elif "symbolic link" in file_type:
-                return "symlink"
-            elif "hermes javascript bytecode" in file_type:
-                return "hermes"
-            elif "empty" in file_type:
-                return "empty"
-
-            return file_type
-        except subprocess.CalledProcessError as e:
-            logger.warning(f"Failed to detect file type for {file_path}: {e}")
-            return "unknown"
-        except Exception as e:
-            logger.warning(f"Unexpected error detecting file type for {file_path}: {e}")
-            return "unknown"
-
     def _get_profile_type(self, profile_data: dict[str, Any]) -> Tuple[str, str]:
         """Determine the type of provisioning profile and its name.
         Args:
@@ -325,86 +281,6 @@ def _get_profile_type(self, profile_data: dict[str, Any]) -> Tuple[str, str]:
         # If no devices are provisioned, it's an app store profile
         return "appstore", profile_name
 
-    @trace("apple.analyze_files")
-    def _analyze_files(self, xcarchive: ZippedXCArchive) -> FileAnalysis:
-        """Analyze all files in the app bundle."""
-        logger.debug("Analyzing files in app bundle")
-
-        files: List[FileInfo] = []
-        app_bundle_path = xcarchive.get_app_bundle_path()
-
-        # Walk through all files in the bundle
-        for file_path in app_bundle_path.rglob("*"):
-            if not file_path.is_file():
-                continue
-
-            relative_path = file_path.relative_to(app_bundle_path)
-            file_size = to_nearest_block_size(get_file_size(file_path), APPLE_FILESYSTEM_BLOCK_SIZE)
-
-            # Get file type from extension first
-            # If no extension or unknown type, use file command
-            file_type = file_path.suffix.lower().lstrip(".")
-            if not file_type or file_type == "unknown":
-                file_type = self._detect_file_type(file_path)
-
-            # Calculate hash for duplicate detection
-            file_hash = calculate_file_hash(file_path, algorithm="md5")
-
-            children: List[FileInfo] = []
-            if file_type == "car":
-                children = self._analyze_asset_catalog(xcarchive, relative_path)
-                children_size = sum([child.size for child in children])
-                children.append(
-                    FileInfo(
-                        full_path=file_path,
-                        path=str(relative_path) + "/Other",
-                        size=file_size - children_size,
-                        file_type="unknown",
-                        hash_md5=file_hash,
-                        treemap_type=TreemapType.ASSETS,
-                        children=[],
-                    )
-                )
-
-            file_info = FileInfo(
-                full_path=file_path,
-                path=str(relative_path),
-                size=file_size,
-                file_type=file_type or "unknown",
-                hash_md5=file_hash,
-                treemap_type=FILE_TYPE_TO_TREEMAP_TYPE.get(file_type, TreemapType.FILES),
-                children=children,
-            )
-
-            files.append(file_info)
-
-        return FileAnalysis(files=files)
-
-    @trace("apple.analyze_asset_catalog")
-    def _analyze_asset_catalog(self, xcarchive: ZippedXCArchive, relative_path: Path) -> List[FileInfo]:
-        """Analyze an asset catalog file."""
-        catalog_details = xcarchive.get_asset_catalog_details(relative_path)
-        result: List[FileInfo] = []
-        for element in catalog_details:
-            if element.full_path and element.full_path.exists() and element.full_path.is_file():
-                file_hash = calculate_file_hash(element.full_path, algorithm="md5")
-            else:
-                # not every element is backed by a file, so use imageId as hash
-                file_hash = element.image_id
-
-            result.append(
-                FileInfo(
-                    full_path=element.full_path,
-                    path=str(relative_path) + "/" + element.name,
-                    size=element.size,
-                    file_type=Path(element.full_path).suffix.lstrip(".") if element.full_path else "other",
-                    hash_md5=file_hash,
-                    treemap_type=TreemapType.ASSETS,
-                    children=[],
-                )
-            )
-        return result
-
     def _generate_insight_with_tracing(
         self, insight_class: type, insights_input: InsightsInput, insight_name: str
     ) -> Any:
@@ -506,6 +382,8 @@ def _test_strip_symbols_removal(self, parser: MachOParser, binary_path: Path) ->
         """Test actual symbol removal using LIEF to get real size savings, similar to what strip does."""
         import tempfile
 
+        return 0
+
         try:
             with trace_ctx("strip_symbols.get_original_size"):
                 original_size = binary_path.stat().st_size

src/launchpad/size/insights/apple/localized_strings.py

@@ -1,6 +1,7 @@
 from launchpad.size.insights.insight import Insight, InsightsInput
 from launchpad.size.models.apple import LocalizedStringInsightResult
 from launchpad.size.models.common import FileInfo
+from launchpad.size.models.insights import FileSavingsResult
 
 
 class LocalizedStringsInsight(Insight[LocalizedStringInsightResult]):
@@ -25,10 +26,11 @@ def generate(self, input: InsightsInput) -> LocalizedStringInsightResult | None:
                 localized_files.append(file_info)
                 total_size += file_info.size
 
-        # Only return insight if total size exceeds threshold
         if total_size > self.THRESHOLD_BYTES:
+            file_savings = [FileSavingsResult(file_path=file.path, total_savings=file.size) for file in localized_files]
+
             return LocalizedStringInsightResult(
-                files=localized_files,
+                files=file_savings,
                 total_savings=total_size,
             )
 

src/launchpad/size/insights/apple/main_binary_export_metadata.py

@@ -1,31 +1,42 @@
 from launchpad.size.insights.insight import Insight, InsightsInput
 from launchpad.size.models.apple import MachOBinaryAnalysis, MainBinaryExportMetadataResult
+from launchpad.size.models.insights import FileSavingsResult
 
 
 class MainBinaryExportMetadataInsight(Insight[MainBinaryExportMetadataResult]):
-    """Insight for analyzing the exported symbols metadata in the main binary."""
+    """Insight for analyzing the exported symbols metadata in all main binaries."""
 
     MIN_EXPORTS_THRESHOLD = 1024
 
     def generate(self, input: InsightsInput) -> MainBinaryExportMetadataResult | None:
-        """Generate insight for main binary exported symbols analysis."""
+        """Generate insight for all main binary exported symbols analysis."""
 
-        main_binary_analysis = None
+        export_files: list[FileSavingsResult] = []
+
+        # Analyze all main binaries (main app, app extensions, watch apps)
         for analysis in input.binary_analysis:
             if isinstance(analysis, MachOBinaryAnalysis) and analysis.is_main_binary:
-                main_binary_analysis = analysis
-                break
-
-        if not main_binary_analysis or not main_binary_analysis.binary_analysis:
+                if not analysis.binary_analysis:
+                    continue
+
+                # Look for dyld_exports_trie component in this main binary
+                for component in analysis.binary_analysis.components:
+                    if component.name == "dyld_exports_trie":
+                        if component.size >= self.MIN_EXPORTS_THRESHOLD:
+                            export_files.append(
+                                FileSavingsResult(
+                                    file_path=analysis.binary_relative_path,
+                                    total_savings=component.size,
+                                )
+                            )
+                        break
+
+        if not export_files:
             return None
 
-        dyld_exports_trie_component = None
-        for component in main_binary_analysis.binary_analysis.components:
-            if component.name == "dyld_exports_trie":
-                dyld_exports_trie_component = component
-                break
-
-        if not dyld_exports_trie_component or dyld_exports_trie_component.size < self.MIN_EXPORTS_THRESHOLD:
-            return None
+        total_savings = sum(file.total_savings for file in export_files)
 
-        return MainBinaryExportMetadataResult(total_savings=dyld_exports_trie_component.size)
+        return MainBinaryExportMetadataResult(
+            total_savings=total_savings,
+            files=export_files,
+        )