Merge pull request #30 from KritantaDev/master

Backport to python 3.8.0

Author: arandomdev

bin/dyldex

@@ -8,13 +8,7 @@ import mmap
 import os
 import sys
 from io import BufferedReader
-
-try:
-	assert sys.version_info >= (3, 9, 5)
-except AssertionError:
-	print("Python 3.9.5 or greater is required", file=sys.stderr)
-	exit(1)
-
+from typing import List
 try:
 	progressbar.streams
 except AttributeError:
@@ -123,7 +117,7 @@ def _extractImage(
 	statusBar.update(unit="Extractor", status="Done")
 
 
-def _filterImages(imagePaths: list[str], filterTerm: str):
+def _filterImages(imagePaths: List[str], filterTerm: str):
 	filteredPaths = []
 	filterTerm = filterTerm.lower()
 

setup.py

@@ -5,7 +5,7 @@
       description='Extract Binaries from Apple\'s Dyld Shared Cache',
       long_description='file: README.md',
       long_description_content_type='text/markdown',
-      python_requires='>=3.9.5',
+      python_requires='>=3.8',
       author='arandomdev',
       url='https://github.com/arandomdev/dyldextractor',
       install_requires=['progressbar2', 'capstone==4.0.2'],

src/DyldExtractor/converter/objc_fixer.py

@@ -1,5 +1,6 @@
 import struct
 import ctypes
+from typing import List, Set, Dict, Tuple
 import capstone as cp
 
 from DyldExtractor.extraction_context import ExtractionContext
@@ -289,7 +290,7 @@ def _findAddInstr(
 
 		return None
 
-	def _getOpcodes(self, opStr: str) -> list[str]:
+	def _getOpcodes(self, opStr: str) -> List[str]:
 		return [
 			opcode.strip()
 			for opcode
@@ -416,7 +417,7 @@ def run(self) -> None:
 			pass
 		pass
 
-	def _disasmText(self) -> tuple[int, str, tuple[str]]:
+	def _disasmText(self) -> Tuple[int, str, Tuple[str]]:
 		"""Disassemble and save the __text section."""
 
 		self._statusBar.update(status="Disassembling Text (will appear frozen)")
@@ -478,8 +479,8 @@ def _findAddInstructions(
 		self,
 		startIdx: int,
 		adrpReg: str,
-		_processed: set[int] = None
-	) -> set[int]:
+		_processed: Set[int] = None
+	) -> Set[int]:
 		"""Find ADD instructions given an ADRP register.
 
 		This will recursively follow branches and stop
@@ -626,28 +627,28 @@ def run(self):
 
 		# caches that map the original definition address
 		# to its new processed address.
-		self._categoryCache: dict[int, int] = {}
-		self._classCache: dict[int, int] = {}
-		self._classDataCache: dict[int, int] = {}
-		self._ivarListCache: dict[int, int] = {}
-		self._protocolListCache: dict[int, int] = {}
-		self._protocolCache: dict[int, int] = {}
-		self._propertyListCache: dict[int, int] = {}
-		self._methodListCache: dict[int, int] = {}
-		self._stringCache: dict[int, int] = {}
-		self._intCache: dict[int, int] = {}
+		self._categoryCache: Dict[int, int] = {}
+		self._classCache: Dict[int, int] = {}
+		self._classDataCache: Dict[int, int] = {}
+		self._ivarListCache: Dict[int, int] = {}
+		self._protocolListCache: Dict[int, int] = {}
+		self._protocolCache: Dict[int, int] = {}
+		self._propertyListCache: Dict[int, int] = {}
+		self._methodListCache: Dict[int, int] = {}
+		self._stringCache: Dict[int, int] = {}
+		self._intCache: Dict[int, int] = {}
 
 		# connects a selrefs old target to its pointer address
-		self._selRefCache: dict[int, int] = {}
+		self._selRefCache: Dict[int, int] = {}
 
 		# A list of class pointers that are being processed.
-		self._classesProcessing: list[int] = []
+		self._classesProcessing: List[int] = []
 
 		# A list of pointers that need to be updated at the end
 		# The first int is the address to the pointer that needs
 		# to be changed. The second int is the address of the
 		# target class
-		self._futureClasses: list[tuple[int, int]] = []
+		self._futureClasses: List[Tuple[int, int]] = []
 
 		self._processSections()
 		self._finalizeFutureClasses()
@@ -858,7 +859,7 @@ def _processCategory(self, categoryAddr: int) -> int:
 		self._categoryCache[categoryAddr] = newCategoryAddr
 		return newCategoryAddr
 
-	def _processClass(self, classAddr: int) -> tuple[int, bool]:
+	def _processClass(self, classAddr: int) -> Tuple[int, bool]:
 		"""Process a class definition.
 
 		Args:

src/DyldExtractor/converter/slide_info.py

@@ -3,6 +3,8 @@
 	Type,
 	TypeVar,
 	Union,
+	Tuple,
+	List
 )
 
 from DyldExtractor.extraction_context import ExtractionContext
@@ -65,7 +67,7 @@ def run(self) -> None:
 
 	def _rebaseSegment(
 		self,
-		pageStarts: tuple[int],
+		pageStarts: Tuple[int],
 		segment: segment_command_64
 	) -> None:
 		"""Process all slide info for a segment"""
@@ -170,7 +172,7 @@ def run(self) -> None:
 
 	def _rebaseSegment(
 		self,
-		pageStarts: tuple[int],
+		pageStarts: Tuple[int],
 		segment: segment_command_64
 	) -> None:
 		# check if the segment is included in the mapping
@@ -230,7 +232,7 @@ def _rebasePage(self, pageOffset, delta) -> None:
 
 def _getMappingSlidePairs(
 	extractionCtx: ExtractionContext
-) -> list[tuple[Union[dyld_cache_mapping_info, dyld_cache_slide_info2]]]:
+) -> List[Tuple[Union[dyld_cache_mapping_info, dyld_cache_slide_info2]]]:
 	dyldCtx = extractionCtx.dyldCtx
 	logger = extractionCtx.logger
 

src/DyldExtractor/converter/stub_fixer.py

@@ -1,7 +1,7 @@
 import dataclasses
 import enum
 import struct
-from typing import Iterator
+from typing import Iterator, List, Tuple, Dict
 
 from DyldExtractor.extraction_context import ExtractionContext
 from DyldExtractor.file_context import FileContext
@@ -61,7 +61,7 @@ def __init__(self, extractionCtx: ExtractionContext) -> None:
 		self._enumerateSymbols()
 		pass
 
-	def symbolizeAddr(self, addr: int) -> list[bytes]:
+	def symbolizeAddr(self, addr: int) -> List[bytes]:
 		"""Get the name of a function at the address.
 
 		Args:
@@ -79,8 +79,8 @@ def symbolizeAddr(self, addr: int) -> list[bytes]:
 	def _enumerateExports(self) -> None:
 		# process the dependencies iteratively,
 		# skipping ones already processed
-		depsQueue: list[_DependencyInfo] = []
-		depsProcessed: list[bytes] = []
+		depsQueue: List[_DependencyInfo] = []
+		depsProcessed: List[bytes] = []
 
 		# load commands for all dependencies
 		DEP_LCS = (
@@ -179,7 +179,7 @@ def _getDepInfo(self, dylib: dylib_command) -> _DependencyInfo:
 	def _readDepExports(
 		self,
 		depInfo: _DependencyInfo
-	) -> list[dyld_trie.ExportInfo]:
+	) -> List[dyld_trie.ExportInfo]:
 		exportOff = None
 		exportSize = None
 
@@ -215,7 +215,7 @@ def _readDepExports(
 	def _cacheDepExports(
 		self,
 		depInfo: _DependencyInfo,
-		exports: list[dyld_trie.ExportInfo]
+		exports: List[dyld_trie.ExportInfo]
 	) -> None:
 		for export in exports:
 			if not export.address:
@@ -331,7 +331,7 @@ def getResolverTarget(address):
 		)
 
 		# A cache of resolved stub chains
-		self._resolveCache: dict[int, int] = {}
+		self._resolveCache: Dict[int, int] = {}
 		pass
 
 	def generateStubNormal(self, stubAddress: int, ldrAddress: int) -> bytes:
@@ -421,7 +421,7 @@ def resolveStubChain(self, address: int) -> int:
 		self._resolveCache[address] = target
 		return target
 
-	def resolveStub(self, address: int) -> tuple[int, _StubFormat]:
+	def resolveStub(self, address: int) -> Tuple[int, _StubFormat]:
 		"""Get the stub and its format.
 
 		Args:
@@ -464,7 +464,7 @@ def getStubHelperData(self, address: int) -> int:
 
 		return data
 
-	def getResolverData(self, address: int) -> tuple[int, int]:
+	def getResolverData(self, address: int) -> Tuple[int, int]:
 		"""Get the data of a resolver.
 
 		This is a stub helper that branches to a function
@@ -1015,17 +1015,17 @@ def run(self):
 		self._fixIndirectSymbols(symbolPtrs, stubMap)
 		pass
 
-	def _enumerateSymbolPointers(self) -> dict[bytes, tuple[int]]:
+	def _enumerateSymbolPointers(self) -> Dict[bytes, Tuple[int]]:
 		"""Generate a mapping between a pointer's symbol and its address.
 		"""
 
 		# read all the bind records as they're a source of symbolic info
-		bindRecords: dict[int, _BindRecord] = {}
+		bindRecords: Dict[int, _BindRecord] = {}
 		dyldInfo: dyld_info_command = self._machoCtx.getLoadCommand(
 			(LoadCommands.LC_DYLD_INFO, LoadCommands.LC_DYLD_INFO_ONLY)
 		)
 		if dyldInfo:
-			records: list[_BindRecord] = []
+			records: List[_BindRecord] = []
 			try:
 				if dyldInfo.weak_bind_size:
 					# usually contains records for c++ symbols like "new"
@@ -1067,7 +1067,7 @@ def _enumerateSymbolPointers(self) -> dict[bytes, tuple[int]]:
 			pass
 
 		# enumerate all symbol pointers
-		symbolPtrs: dict[bytes, list[int]] = {}
+		symbolPtrs: Dict[bytes, List[int]] = {}
 
 		def _addToMap(ptrSymbol: bytes, ptrAddr: int, section: section_64):
 			if ptrSymbol in symbolPtrs:
@@ -1210,12 +1210,12 @@ def _fixStubHelpers(self) -> None:
 
 	def _fixStubs(
 		self,
-		symbolPtrs: dict[bytes, tuple[int]]
-	) -> dict[bytes, tuple[int]]:
+		symbolPtrs: Dict[bytes, Tuple[int]]
+	) -> Dict[bytes, Tuple[int]]:
 		"""Relink stubs to their symbol pointers
 		"""
 
-		stubMap: dict[bytes, list[int]] = {}
+		stubMap: Dict[bytes, List[int]] = {}
 
 		def _addToMap(stubName: bytes, stubAddr: int):
 			if stubName in stubMap:
@@ -1360,7 +1360,7 @@ def _addToMap(stubName: bytes, stubAddr: int):
 
 		return stubMap
 
-	def _fixCallsites(self, stubMap: dict[bytes, tuple[int]]) -> None:
+	def _fixCallsites(self, stubMap: Dict[bytes, Tuple[int]]) -> None:
 		if (
 			b"__TEXT" not in self._machoCtx.segments
 			or b"__text" not in self._machoCtx.segments[b"__TEXT"].sects
@@ -1443,8 +1443,8 @@ def _fixCallsites(self, stubMap: dict[bytes, tuple[int]]) -> None:
 
 	def _fixIndirectSymbols(
 		self,
-		symbolPtrs: dict[bytes, tuple[int]],
-		stubMap: dict[bytes, tuple[int]]
+		symbolPtrs: Dict[bytes, Tuple[int]],
+		stubMap: Dict[bytes, Tuple[int]]
 	) -> None:
 		"""Fix indirect symbols.
 

src/DyldExtractor/dyld/dyld_context.py

@@ -1,4 +1,5 @@
 from mmap import mmap
+from typing import List
 
 from DyldExtractor.file_context import FileContext
 from DyldExtractor.dyld.dyld_structs import (
@@ -11,8 +12,8 @@
 class DyldContext(FileContext):
 
 	header: dyld_cache_header
-	mappings: list[dyld_cache_mapping_info]
-	images: list[dyld_cache_image_info]
+	mappings: List[dyld_cache_mapping_info]
+	images: List[dyld_cache_image_info]
 
 	def __init__(self, file: mmap) -> None:
 		"""A wrapper around a dyld file.

src/DyldExtractor/dyld/dyld_trie.py

@@ -4,10 +4,12 @@
 
 from DyldExtractor import leb128
 
+from typing import Tuple, List
+
 from DyldExtractor.macho.macho_constants import *
 
 
-def _readString(file: mmap, readHead: int) -> tuple[bytes, int]:
+def _readString(file: mmap, readHead: int) -> Tuple[bytes, int]:
 	"""Read a null terminated string.
 
 	Returns:
@@ -55,7 +57,7 @@ class ExportReaderError(Exception):
 
 class _ExportTrieReader(object):
 
-	exports: list[ExportInfo]
+	exports: List[ExportInfo]
 
 	def __init__(
 		self,
@@ -121,7 +123,7 @@ def ReadExports(
 	file: mmap,
 	exportOff: int,
 	exportSize: int
-) -> list[ExportInfo]:
+) -> List[ExportInfo]:
 	"""Read an export trie.
 
 	Args:

src/DyldExtractor/leb128.py

@@ -1,6 +1,6 @@
+from typing import Tuple
 
-
-def decodeUleb128(buffer: bytes, readHead: int) -> tuple[int, int]:
+def decodeUleb128(buffer: bytes, readHead: int) -> Tuple[int, int]:
 	"""Read a Uleb128 value.
 
 	Args:
@@ -58,7 +58,7 @@ def encodeUleb128(value: int) -> bytes:
 	return bytes(data)
 
 
-def decodeSleb128(buffer: bytes, readHead: int) -> tuple[int, int]:
+def decodeSleb128(buffer: bytes, readHead: int) -> Tuple[int, int]:
 	"""Read a 64bit Sleb128 value.
 
 	Args:

src/DyldExtractor/macho/macho_context.py

@@ -1,7 +1,7 @@
 import struct
 from mmap import mmap
 
-from typing import Union
+from typing import Union, List, Dict, Tuple
 
 from DyldExtractor.file_context import FileContext
 
@@ -18,10 +18,10 @@
 
 class MachOContext(FileContext):
 
-	loadCommands: list[load_command]
+	loadCommands: List[load_command]
 
-	segments: dict[bytes, SegmentContext]
-	segmentsI: list[SegmentContext]
+	segments: Dict[bytes, SegmentContext]
+	segmentsI: List[SegmentContext]
 
 	def __init__(
 		self,
@@ -51,9 +51,9 @@ def __init__(
 
 	def getLoadCommand(
 		self,
-		cmdFilter: tuple[LoadCommands],
+		cmdFilter: Tuple[LoadCommands],
 		multiple: bool = False
-	) -> Union[load_command, tuple[load_command]]:
+	) -> Union[load_command, Tuple[load_command]]:
 		"""Retreive a load command with its command ID
 
 		Args: