Add code signature, string table, and symbol table sizes

src/launchpad/parsers/apple/macho_parser.py

@@ -11,7 +11,7 @@
 import lief
 import sentry_sdk
 
-from launchpad.size.models.apple import DyldInfo
+from launchpad.size.models.apple import LinkEditInfo
 
 from ...utils.logging import get_logger
 from .binary_utils import parse_null_terminated_strings
@@ -300,11 +300,53 @@ def find_symbol(addr: int) -> lief.Symbol | None:
         return symbols
 
     @sentry_sdk.trace
-    def extract_dyld_info(self) -> DyldInfo:
-        """Extract DYLD information from LC_DYLD_INFO load commands."""
+    def extract_linkedit_info(self) -> LinkEditInfo:
+        """Extract all __LINKEDIT segment component sizes from load commands."""
+        symbol_table_size = 0
+        string_table_size = 0
+        function_starts_size = 0
+        segment_size = 0
+        chained_fixups_size = 0
+        export_trie_size = 0
+        code_signature_size = 0
+        code_signature_offset = 0
+
+        # Determine if binary is 64-bit (nlist_64 is 16 bytes, nlist is 12 bytes)
+        is_64bit = self.binary.header.magic in [
+            lief.MachO.MACHO_TYPES.MAGIC_64,
+            lief.MachO.MACHO_TYPES.CIGAM_64,
+        ]
+        entry_size = 16 if is_64bit else 12
+
+        for cmd in self.binary.commands:
+            if isinstance(cmd, lief.MachO.SymbolCommand):
+                symbol_table_size = cmd.numberof_symbols * entry_size
+                string_table_size = cmd.strings_size
+            elif isinstance(cmd, lief.MachO.FunctionStarts):
+                function_starts_size = cmd.data_size
+
         dyld_chained_fixups = self.binary.dyld_chained_fixups
         dyld_exports_trie = self.binary.dyld_exports_trie
-        return DyldInfo(
-            chained_fixups_size=(dyld_chained_fixups.data_size if dyld_chained_fixups else 0),
-            export_trie_size=dyld_exports_trie.data_size if dyld_exports_trie else 0,
+        chained_fixups_size = dyld_chained_fixups.data_size if dyld_chained_fixups else 0
+        export_trie_size = dyld_exports_trie.data_size if dyld_exports_trie else 0
+
+        if self.binary.has_code_signature and self.binary.code_signature is not None:
+            cs = self.binary.code_signature
+            code_signature_size = cs.data_size
+            code_signature_offset = cs.data_offset
+
+        for segment in self.binary.segments:
+            if segment.name == "__LINKEDIT":
+                segment_size = segment.file_size
+                break
+
+        return LinkEditInfo(
+            segment_size=segment_size,
+            symbol_table_size=symbol_table_size,
+            string_table_size=string_table_size,
+            function_starts_size=function_starts_size,
+            chained_fixups_size=chained_fixups_size,
+            export_trie_size=export_trie_size,
+            code_signature_size=code_signature_size,
+            code_signature_offset=code_signature_offset,
         )

src/launchpad/size/analyzers/apple.py

@@ -439,7 +439,7 @@ def _analyze_binary(
         objc_method_names = parser.parse_objc_method_names()
         segments = self._extract_segments_info(parser.binary)
         load_commands = self._extract_load_commands_info(parser.binary)
-        dyld_info = parser.extract_dyld_info()
+        linkedit_info = parser.extract_linkedit_info()
 
         symbol_info = None
         dwarf_relocations = None
@@ -496,9 +496,9 @@ def _analyze_binary(
             segments=segments,
             load_commands=load_commands,
             header_size=parser.get_header_size(),
-            dyld_info=dyld_info,
             dwarf_relocations=dwarf_relocations,
             strippable_symbols_size=strippable_symbols_size,
+            linkedit_info=linkedit_info,
         )
 
     @sentry_sdk.trace

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

@@ -17,11 +17,11 @@ def generate(self, input: InsightsInput) -> MainBinaryExportMetadataResult | Non
             if not analysis.is_main_binary:
                 continue
 
-            dyld_info = analysis.dyld_info
-            if dyld_info is None:
+            linkedit_info = analysis.linkedit_info
+            if linkedit_info is None:
                 continue
 
-            export_trie_size = dyld_info.export_trie_size
+            export_trie_size = linkedit_info.export_trie_size
             if export_trie_size >= self.MIN_EXPORTS_THRESHOLD:
                 export_files.append(
                     FileSavingsResult(

src/launchpad/size/models/apple.py

@@ -127,12 +127,30 @@ class LoadCommandInfo:
 
 
 @dataclass
-class DyldInfo:
-    """DYLD-specific information extracted from related DYLD load commands."""
+class LinkEditInfo:
+    """Link edit segment components extracted from various load commands in __LINKEDIT.
 
+    Consolidates all __LINKEDIT segment data including symbol tables, DYLD info, and code signature.
+    """
+
+    # Segment
+    segment_size: int = 0
+
+    # Symbol tables (from LC_SYMTAB)
+    symbol_table_size: int = 0
+    string_table_size: int = 0
+
+    # Debugging (from LC_FUNCTION_STARTS)
+    function_starts_size: int = 0
+
+    # DYLD (from LC_DYLD_CHAINED_FIXUPS, LC_DYLD_EXPORTS_TRIE)
     chained_fixups_size: int = 0
     export_trie_size: int = 0
 
+    # Code signing (from LC_CODE_SIGNATURE)
+    code_signature_size: int = 0
+    code_signature_offset: int = 0
+
 
 @dataclass
 class MachOBinaryAnalysis:
@@ -150,9 +168,9 @@ class MachOBinaryAnalysis:
     swift_metadata: SwiftMetadata | None = None
     symbol_info: SymbolInfo | None = None
     header_size: int = 0
-    dyld_info: DyldInfo | None = None
     dwarf_relocations: DwarfRelocationsData | None = None
     strippable_symbols_size: int = 0
+    linkedit_info: LinkEditInfo | None = None
 
 
 @dataclass

src/launchpad/size/treemap/macho_element_builder.py

@@ -435,11 +435,11 @@ def _add_segments(
                         )
                     )
 
-            dyld_children_size = 0
+            linkedit_children_size = 0
             if segment_name == "__LINKEDIT":
-                dyld_children = self._build_dyld_load_command_children(binary_analysis)
-                segment_children.extend(dyld_children)
-                dyld_children_size = sum(c.size for c in dyld_children)
+                linkedit_children = self._build_linkedit_children(binary_analysis)
+                segment_children.extend(linkedit_children)
+                linkedit_children_size = sum(c.size for c in linkedit_children)
 
             displayed_section_size = sum(c.size for c in segment_children)
 
@@ -448,7 +448,7 @@ def _add_segments(
                 seg_total_size = segment.size
 
             total_section_declared = sum(s.size for s in segment.sections) if segment.sections else 0
-            segment_overhead = seg_total_size - total_section_declared - dyld_children_size
+            segment_overhead = seg_total_size - total_section_declared - linkedit_children_size
             actual_segment_size = displayed_section_size + max(0, segment_overhead)
 
             if actual_segment_size > 0:
@@ -503,37 +503,93 @@ def _build_metadata_components(self, binary_analysis: MachOBinaryAnalysis) -> Li
 
         return metadata_children
 
-    def _build_dyld_load_command_children(self, binary_analysis: MachOBinaryAnalysis) -> List[TreemapElement]:
-        dyld_children: List[TreemapElement] = []
-        di = binary_analysis.dyld_info
-        if di is None:
-            return dyld_children
+    def _build_linkedit_children(self, binary_analysis: MachOBinaryAnalysis) -> List[TreemapElement]:
+        """Build child elements for the __LINKEDIT segment.
 
-        if di.chained_fixups_size > 0:
-            dyld_children.append(
+        Includes symbol table, string table, function starts, DYLD info, and code signature.
+        """
+        linkedit_children: List[TreemapElement] = []
+
+        le = binary_analysis.linkedit_info
+        if le is None:
+            return linkedit_children
+
+        # Add symbol table and string table
+        if le.string_table_size > 0:
+            linkedit_children.append(
+                TreemapElement(
+                    name="String Table",
+                    size=le.string_table_size,
+                    type=TreemapType.EXECUTABLES,
+                    path=None,
+                    is_dir=False,
+                    children=[],
+                )
+            )
+
+        if le.symbol_table_size > 0:
+            linkedit_children.append(
+                TreemapElement(
+                    name="Symbol Table",
+                    size=le.symbol_table_size,
+                    type=TreemapType.EXECUTABLES,
+                    path=None,
+                    is_dir=False,
+                    children=[],
+                )
+            )
+
+        if le.function_starts_size > 0:
+            linkedit_children.append(
+                TreemapElement(
+                    name="Function Starts",
+                    size=le.function_starts_size,
+                    type=TreemapType.EXECUTABLES,
+                    path=None,
+                    is_dir=False,
+                    children=[],
+                )
+            )
+
+        # Add DYLD info
+        if le.chained_fixups_size > 0:
+            linkedit_children.append(
                 TreemapElement(
                     name="Chained Fixups",
-                    size=di.chained_fixups_size,
+                    size=le.chained_fixups_size,
                     type=TreemapType.DYLD,
                     path=None,
                     is_dir=False,
                     children=[],
                 )
             )
 
-        if di.export_trie_size > 0:
-            dyld_children.append(
+        if le.export_trie_size > 0:
+            linkedit_children.append(
                 TreemapElement(
                     name="Export Trie",
-                    size=di.export_trie_size,
+                    size=le.export_trie_size,
                     type=TreemapType.DYLD,
                     path=None,
                     is_dir=False,
                     children=[],
                 )
             )
 
-        return dyld_children
+        # Add code signature
+        if le.code_signature_size > 0:
+            linkedit_children.append(
+                TreemapElement(
+                    name="Code Signature",
+                    size=le.code_signature_size,
+                    type=TreemapType.CODE_SIGNATURE,
+                    path=None,
+                    is_dir=False,
+                    children=[],
+                )
+            )
+
+        return linkedit_children
 
     def _add_unmapped_region(self, binary_analysis: MachOBinaryAnalysis, binary_children: List[TreemapElement]) -> None:
         total_accounted = sum(c.size for c in binary_children)

tests/integration/size/test_treemap_generation.py

@@ -361,9 +361,27 @@ def find_node_by_path(root: TreemapElement, path: str) -> TreemapElement | None:
         # Verify __LINKEDIT segment
         has_linkedit = "__LINKEDIT" in main_exe_sections
         assert has_linkedit
-        linkedit_size = main_exe_sections["__LINKEDIT"].size
+        linkedit = main_exe_sections["__LINKEDIT"]
+        linkedit_size = linkedit.size
         assert linkedit_size == 269360
 
+        # Verify __LINKEDIT has children (symbol table, string table, etc.)
+        linkedit_children = {child.name: child for child in linkedit.children}
+        assert len(linkedit_children) == 6, "__LINKEDIT should have exactly 6 child components"
+
+        # Verify expected __LINKEDIT components exist
+        assert "Symbol Table" in linkedit_children, "Symbol table should be present"
+        assert "String Table" in linkedit_children, "String table should be present"
+        assert "Function Starts" in linkedit_children, "Function starts should be present"
+
+        # Verify exact sizes for __LINKEDIT components
+        assert linkedit_children["Symbol Table"].size == 11072, "Symbol table size mismatch"
+        assert linkedit_children["String Table"].size == 16584, "String table size mismatch"
+        assert linkedit_children["Function Starts"].size == 13584, "Function starts size mismatch"
+        assert linkedit_children["Chained Fixups"].size == 87616, "Chained fixups size mismatch"
+        assert linkedit_children["Export Trie"].size == 85056, "Export trie size mismatch"
+        assert linkedit_children["Code Signature"].size == 43488, "Code signature size mismatch"
+
         # Verify Unmapped section (track size changes)
         has_unmapped = "Unmapped" in main_exe_sections
         assert has_unmapped

tests/unit/test_main_binary_export_metadata_insight.py

@@ -2,7 +2,7 @@
 
 from launchpad.size.insights.apple.main_binary_export_metadata import MainBinaryExportMetadataInsight
 from launchpad.size.insights.insight import InsightsInput
-from launchpad.size.models.apple import DyldInfo, MachOBinaryAnalysis
+from launchpad.size.models.apple import LinkEditInfo, MachOBinaryAnalysis
 from launchpad.size.models.common import BaseAppInfo, FileAnalysis
 from launchpad.size.models.insights import MainBinaryExportMetadataResult
 
@@ -26,7 +26,7 @@ def test_generate_with_main_binary_and_dyld_exports_trie(self):
             swift_metadata=None,
             is_main_binary=True,
             header_size=32,
-            dyld_info=DyldInfo(export_trie_size=5000),
+            linkedit_info=LinkEditInfo(export_trie_size=5000),
         )
 
         insights_input = InsightsInput(
@@ -42,7 +42,7 @@ def test_generate_with_main_binary_and_dyld_exports_trie(self):
         assert result.total_savings == 5000
 
     def test_generate_with_main_binary_without_dyld_exports_trie(self):
-        """Test that no insight is generated when main binary has no dyld_info."""
+        """Test that no insight is generated when main binary has no linkedit_info."""
         main_binary_analysis = MachOBinaryAnalysis(
             binary_absolute_path=Path("MyApp"),
             binary_relative_path=Path("MyApp"),
@@ -56,7 +56,7 @@ def test_generate_with_main_binary_without_dyld_exports_trie(self):
             swift_metadata=None,
             is_main_binary=True,
             header_size=32,
-            dyld_info=None,  # No dyld_info
+            linkedit_info=None,  # No linkedit_info
         )
 
         insights_input = InsightsInput(
@@ -85,7 +85,7 @@ def test_generate_with_no_main_binary(self):
             swift_metadata=None,
             is_main_binary=False,  # Not a main binary
             header_size=32,
-            dyld_info=DyldInfo(export_trie_size=5000),  # Has export trie but not main binary
+            linkedit_info=LinkEditInfo(export_trie_size=5000),  # Has export trie but not main binary
         )
 
         insights_input = InsightsInput(
@@ -114,7 +114,7 @@ def test_generate_with_main_binary_but_empty_export_trie(self):
             swift_metadata=None,
             is_main_binary=True,
             header_size=32,
-            dyld_info=DyldInfo(export_trie_size=0),  # Zero-size export trie
+            linkedit_info=LinkEditInfo(export_trie_size=0),  # Zero-size export trie
         )
 
         insights_input = InsightsInput(
@@ -156,7 +156,7 @@ def test_generate_with_multiple_binaries_one_main(self):
             swift_metadata=None,
             is_main_binary=True,
             header_size=32,
-            dyld_info=DyldInfo(export_trie_size=8000),
+            linkedit_info=LinkEditInfo(export_trie_size=8000),
         )
 
         # Create framework binary (non-main)
@@ -173,7 +173,7 @@ def test_generate_with_multiple_binaries_one_main(self):
             swift_metadata=None,
             is_main_binary=False,
             header_size=32,
-            dyld_info=DyldInfo(export_trie_size=3000),  # Framework also has export trie but won't be included
+            linkedit_info=LinkEditInfo(export_trie_size=3000),  # Framework also has export trie but won't be included
         )
 
         insights_input = InsightsInput(
@@ -203,7 +203,7 @@ def test_generate_with_export_trie_below_threshold(self):
             swift_metadata=None,
             is_main_binary=True,
             header_size=32,
-            dyld_info=DyldInfo(export_trie_size=512),  # Below MIN_EXPORTS_THRESHOLD (1024)
+            linkedit_info=LinkEditInfo(export_trie_size=512),  # Below MIN_EXPORTS_THRESHOLD (1024)
         )
 
         insights_input = InsightsInput(
@@ -232,7 +232,7 @@ def test_generate_with_export_trie_at_threshold(self):
             swift_metadata=None,
             is_main_binary=True,
             header_size=32,
-            dyld_info=DyldInfo(export_trie_size=1024),  # Exactly at MIN_EXPORTS_THRESHOLD
+            linkedit_info=LinkEditInfo(export_trie_size=1024),  # Exactly at MIN_EXPORTS_THRESHOLD
         )
 
         insights_input = InsightsInput(