progress

Author: trevor-e

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 CodeSignatureInfo, DyldInfo, LinkEditInfo
 
 from ...utils.logging import get_logger
 from .binary_utils import parse_null_terminated_strings
@@ -308,3 +308,48 @@ def extract_dyld_info(self) -> 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,
         )
+
+    @sentry_sdk.trace
+    def extract_code_signature_info(self) -> CodeSignatureInfo | None:
+        """Extract code signature information from LC_CODE_SIGNATURE load command."""
+        if not self.binary.has_code_signature:
+            return None
+
+        cs = self.binary.code_signature
+        return CodeSignatureInfo(
+            size=cs.data_size,
+            offset=cs.data_offset,
+        )
+
+    @sentry_sdk.trace
+    def extract_linkedit_info(self) -> LinkEditInfo:
+        """Extract __LINKEDIT segment component sizes from load commands."""
+        symbol_table_size = 0
+        string_table_size = 0
+        function_starts_size = 0
+        segment_size = 0
+
+        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
+
+        for segment in self.binary.segments:
+            if segment.name == "__LINKEDIT":
+                segment_size = segment.file_size
+                break
+
+        return LinkEditInfo(
+            symbol_table_size=symbol_table_size,
+            string_table_size=string_table_size,
+            function_starts_size=function_starts_size,
+            segment_size=segment_size,
+        )

src/launchpad/size/analyzers/apple.py

@@ -440,6 +440,8 @@ def _analyze_binary(
         segments = self._extract_segments_info(parser.binary)
         load_commands = self._extract_load_commands_info(parser.binary)
         dyld_info = parser.extract_dyld_info()
+        code_signature_info = parser.extract_code_signature_info()
+        linkedit_info = parser.extract_linkedit_info()
 
         symbol_info = None
         dwarf_relocations = None
@@ -499,6 +501,8 @@ def _analyze_binary(
             dyld_info=dyld_info,
             dwarf_relocations=dwarf_relocations,
             strippable_symbols_size=strippable_symbols_size,
+            code_signature_info=code_signature_info,
+            linkedit_info=linkedit_info,
         )
 
     @sentry_sdk.trace

src/launchpad/size/models/apple.py

@@ -134,6 +134,24 @@ class DyldInfo:
     export_trie_size: int = 0
 
 
+@dataclass
+class CodeSignatureInfo:
+    """Code signature information extracted from LC_CODE_SIGNATURE load command."""
+
+    size: int = 0
+    offset: int = 0
+
+
+@dataclass
+class LinkEditInfo:
+    """Link edit segment components extracted from various load commands in __LINKEDIT."""
+
+    symbol_table_size: int = 0
+    string_table_size: int = 0
+    function_starts_size: int = 0
+    segment_size: int = 0
+
+
 @dataclass
 class MachOBinaryAnalysis:
     """Mach-O binary analysis results."""
@@ -153,6 +171,8 @@ class MachOBinaryAnalysis:
     dyld_info: DyldInfo | None = None
     dwarf_relocations: DwarfRelocationsData | None = None
     strippable_symbols_size: int = 0
+    code_signature_info: CodeSignatureInfo | None = None
+    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 +503,94 @@ 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] = []
+    def _build_linkedit_children(self, binary_analysis: MachOBinaryAnalysis) -> List[TreemapElement]:
+        """Build child elements for the __LINKEDIT segment.
+
+        Includes symbol table, string table, function starts, DYLD info, and code signature.
+        """
+        linkedit_children: List[TreemapElement] = []
+
+        # Add link edit components (symbol table, string table, function starts)
+        le = binary_analysis.linkedit_info
+        if le is not None:
+            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 children
         di = binary_analysis.dyld_info
-        if di is None:
-            return dyld_children
+        if di is not None:
+            if di.chained_fixups_size > 0:
+                linkedit_children.append(
+                    TreemapElement(
+                        name="Chained Fixups",
+                        size=di.chained_fixups_size,
+                        type=TreemapType.DYLD,
+                        path=None,
+                        is_dir=False,
+                        children=[],
+                    )
+                )
 
-        if di.chained_fixups_size > 0:
-            dyld_children.append(
-                TreemapElement(
-                    name="Chained Fixups",
-                    size=di.chained_fixups_size,
-                    type=TreemapType.DYLD,
-                    path=None,
-                    is_dir=False,
-                    children=[],
+            if di.export_trie_size > 0:
+                linkedit_children.append(
+                    TreemapElement(
+                        name="Export Trie",
+                        size=di.export_trie_size,
+                        type=TreemapType.DYLD,
+                        path=None,
+                        is_dir=False,
+                        children=[],
+                    )
                 )
-            )
 
-        if di.export_trie_size > 0:
-            dyld_children.append(
+        # Add code signature
+        cs = binary_analysis.code_signature_info
+        if cs is not None and cs.size > 0:
+            linkedit_children.append(
                 TreemapElement(
-                    name="Export Trie",
-                    size=di.export_trie_size,
-                    type=TreemapType.DYLD,
+                    name="Code Signature",
+                    size=cs.size,
+                    type=TreemapType.CODE_SIGNATURE,
                     path=None,
                     is_dir=False,
                     children=[],
                 )
             )
 
-        return dyld_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)