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/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,
+        )

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

@@ -2,6 +2,7 @@
 from launchpad.size.insights.insight import Insight, InsightsInput
 from launchpad.size.models.apple import SmallFilesInsightResult
 from launchpad.size.models.common import FileInfo
+from launchpad.size.models.insights import FileSavingsResult
 
 
 class SmallFilesInsight(Insight[SmallFilesInsightResult]):
@@ -26,13 +27,19 @@ def generate(self, input: InsightsInput) -> SmallFilesInsightResult | None:
         for file_info in input.file_analysis.files:
             if file_info.size < APPLE_FILESYSTEM_BLOCK_SIZE:
                 small_files.append(file_info)
-                # Calculate wasted space due to block size alignment
-                wasted_space = APPLE_FILESYSTEM_BLOCK_SIZE - file_info.size
-                total_savings += wasted_space
+                total_savings += APPLE_FILESYSTEM_BLOCK_SIZE - file_info.size
 
         if len(small_files) > 0:
+            file_savings = [
+                FileSavingsResult(
+                    file_path=file.path,
+                    total_savings=APPLE_FILESYSTEM_BLOCK_SIZE - file.size,
+                )
+                for file in small_files
+            ]
+
             return SmallFilesInsightResult(
-                files=small_files,
+                files=file_savings,
                 file_count=len(small_files),
                 total_savings=total_savings,
             )

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

@@ -4,7 +4,7 @@
 
 from launchpad.size.insights.insight import Insight, InsightsInput
 from launchpad.size.models.common import FileInfo
-from launchpad.size.models.insights import UnnecessaryFilesInsightResult
+from launchpad.size.models.insights import FileSavingsResult, UnnecessaryFilesInsightResult
 
 
 class UnnecessaryFilesInsight(Insight[UnnecessaryFilesInsightResult]):
@@ -48,9 +48,10 @@ def generate(self, input: InsightsInput) -> UnnecessaryFilesInsightResult | None
 
         if unnecessary_files:
             unnecessary_files.sort(key=lambda f: f.size, reverse=True)
+            files = [FileSavingsResult(file_path=file.path, total_savings=file.size) for file in unnecessary_files]
 
             return UnnecessaryFilesInsightResult(
-                files=unnecessary_files,
+                files=files,
                 total_savings=total_size,
             )
 

src/launchpad/size/insights/common/duplicate_files.py

@@ -1,45 +1,126 @@
+import hashlib
 import os
 
 from collections import defaultdict
+from pathlib import Path
 from typing import Dict, List
 
 from launchpad.size.insights.insight import Insight, InsightsInput
-from launchpad.size.models.common import FileInfo
-from launchpad.size.models.insights import DuplicateFileGroup, DuplicateFilesInsightResult
+from launchpad.size.models.common import FileInfo, TreemapType
+from launchpad.size.models.insights import (
+    DuplicateFileGroup,
+    DuplicateFilesInsightResult,
+)
 
 
 class DuplicateFilesInsight(Insight[DuplicateFilesInsightResult]):
-    def generate(self, input: InsightsInput) -> DuplicateFilesInsightResult:
-        files_by_hash: Dict[str, List[FileInfo]] = defaultdict(list)
-        for file in input.file_analysis.files:
-            if file.hash_md5:
-                files_by_hash[file.hash_md5].append(file)
+    EXTENSION_ALLOWLIST = [".xcprivacy"]
+
+    # Make sure to group all duplicates in a directory that has one of these extensions
+    DIRECTORY_EXTENSIONS = [".bundle"]
 
+    def generate(self, input: InsightsInput) -> DuplicateFilesInsightResult | None:
         groups: List[DuplicateFileGroup] = []
         total_savings = 0
 
-        for file_list in files_by_hash.values():
-            if len(file_list) > 1:
-                # Calculate savings: total size - size of one copy we keep
-                total_file_size = sum(f.size for f in file_list)
-                savings_for_this_group = total_file_size - file_list[0].size
-
-                if savings_for_this_group > 0:  # Only include if there are actual savings
-                    sorted_files = sorted(file_list, key=lambda f: (-f.size, f.path))
-                    filenames = sorted(set(os.path.basename(f.path) for f in sorted_files))
-                    group_filename = filenames[0]
-
-                    group = DuplicateFileGroup(
-                        filename=group_filename,
-                        files=sorted_files,
-                        total_savings=savings_for_this_group,
-                    )
-                    groups.append(group)
-                    total_savings += savings_for_this_group
-
-        groups = sorted(groups, key=lambda g: (-g.total_savings, g.filename))
-
-        return DuplicateFilesInsightResult(
-            groups=groups,
-            total_savings=total_savings,
-        )
+        covered_containers: set[str] = set()
+        for infos in self._duplicate_directories(input.file_analysis.files).values():
+            if len(infos) < 2:
+                continue
+
+            infos.sort(key=lambda f: (-f.size, f.path))
+            group_size = sum(fi.size for fi in infos)
+            savings = group_size - infos[0].size
+            if savings <= 0:
+                continue
+
+            groups.append(
+                DuplicateFileGroup(
+                    filename=os.path.basename(infos[0].path),
+                    files=infos,
+                    total_savings=savings,
+                )
+            )
+            total_savings += savings
+            for info in infos:
+                covered_containers.add(info.path)
+
+        files_by_hash: Dict[str, List[FileInfo]] = defaultdict(list)
+        for f in input.file_analysis.files:
+            if (
+                f.hash_md5
+                and not self._is_allowed_extension(f.path)
+                and not any(f.path.startswith(c + "/") or f.path == c for c in covered_containers)  # ← NEW GUARD
+            ):
+                files_by_hash[f.hash_md5].append(f)
+
+        for dup_files in files_by_hash.values():
+            if len(dup_files) < 2:
+                continue
+
+            dup_files.sort(key=lambda f: (-f.size, f.path))
+            savings = sum(f.size for f in dup_files) - dup_files[0].size
+            if savings <= 0:
+                continue
+
+            container = self._directory_grouping(dup_files[0].path)
+            name = os.path.basename(container) if container else os.path.basename(dup_files[0].path)
+
+            groups.append(
+                DuplicateFileGroup(
+                    filename=name,
+                    files=dup_files,
+                    total_savings=savings,
+                )
+            )
+            total_savings += savings
+
+        groups.sort(key=lambda g: (-g.total_savings, g.filename))
+
+        if len(groups) > 0:
+            return DuplicateFilesInsightResult(groups=groups, total_savings=total_savings)
+
+        return None
+
+    def _is_allowed_extension(self, file_path: str) -> bool:
+        return any(file_path.endswith(ext) for ext in self.EXTENSION_ALLOWLIST)
+
+    def _directory_grouping(self, file_path: str) -> str | None:
+        p = Path(file_path)
+        for i, part in enumerate(p.parts):
+            if any(part.endswith(ext) for ext in self.DIRECTORY_EXTENSIONS):
+                return str(Path(*p.parts[: i + 1]))
+        return None
+
+    def _duplicate_directories(self, files: List[FileInfo]) -> Dict[str, List[FileInfo]]:
+        dir_to_children: Dict[str, List[FileInfo]] = defaultdict(list)
+        for f in files:
+            if f.hash_md5:
+                root = self._directory_grouping(f.path)
+                if root:
+                    dir_to_children[root].append(f)
+
+        dup_dirs: Dict[str, List[FileInfo]] = defaultdict(list)
+        for root, children in dir_to_children.items():
+            if not children:
+                continue
+
+            md5 = hashlib.md5()
+            for h in sorted(c.hash_md5 for c in children if c.hash_md5):
+                md5.update(h.encode())
+            folder_hash = md5.hexdigest()
+
+            dup_dirs[folder_hash].append(
+                FileInfo(
+                    full_path=(
+                        children[0].full_path.parent / root if children[0].full_path is not None else Path(root)
+                    ),
+                    path=root,
+                    size=sum(c.size for c in children),
+                    file_type="directory",
+                    hash_md5=folder_hash,
+                    treemap_type=TreemapType.FILES,
+                    children=children,
+                )
+            )
+        return dup_dirs

src/launchpad/size/insights/common/hermes_debug_info.py

@@ -5,6 +5,7 @@
 from launchpad.size.insights.insight import Insight, InsightsInput
 from launchpad.size.models.common import FileInfo
 from launchpad.size.models.insights import (
+    FileSavingsResult,
     HermesDebugInfoInsightResult,
 )
 
@@ -47,7 +48,9 @@ def generate(self, input: InsightsInput) -> HermesDebugInfoInsightResult | None:
             reverse=True,
         )
 
+        files = [FileSavingsResult(file_path=file.path, total_savings=file.size) for file in files_with_debug_info]
+
         return HermesDebugInfoInsightResult(
-            files=files_with_debug_info,
+            files=files,
             total_savings=total_savings,
         )