Analyze ReactOS binaries from a build dir

analyze_binaries.py

import sys
import os
import re
import pefile

CMAKE_PREFIX = '# Install script for directory: '
MANIFEST_DIR = pefile.DIRECTORY_ENTRY['IMAGE_DIRECTORY_ENTRY_RESOURCE']
IMPORT_DIR = pefile.DIRECTORY_ENTRY['IMAGE_DIRECTORY_ENTRY_IMPORT']
MANIFEST_ID = pefile.RESOURCE_TYPE['RT_MANIFEST']
INCLUDE_FILES = ('.exe', '.dll', '.acm', '.ax', '.cpl', '.drv', '.ocx')

def check_manifest(filename, peobj):
    if peobj.is_exe():
        if not hasattr(peobj, 'DIRECTORY_ENTRY_RESOURCE'):
            peobj.parse_data_directories(directories=[MANIFEST_DIR])
        if not hasattr(peobj, 'DIRECTORY_ENTRY_RESOURCE'):
            return
        res_entries = peobj.DIRECTORY_ENTRY_RESOURCE.entries
        manifests = [entry for entry in res_entries if entry.id == MANIFEST_ID]
        for entry in manifests:
            for manifest in entry.directory.entries:
                container = manifest.directory.entries[0].data
                data = peobj.get_data(container.struct.OffsetToData, container.struct.Size)
                if data.lower().find('Microsoft.Windows.Common-Controls'.lower()):
                    if not hasattr(peobj, 'DIRECTORY_ENTRY_IMPORT'):
                        peobj.parse_data_directories(directories=[IMPORT_DIR])
                    for import_entry in peobj.DIRECTORY_ENTRY_IMPORT:
                        if import_entry.dll.lower() == 'comctl32.dll':
                            return
                    print 'Missing import:', filename

def check_baseaddress(filename, peobj):
    if not peobj.is_exe() and peobj.OPTIONAL_HEADER.ImageBase == 0x10000000:
        print 'Missing base address:', filename


def analyze(filename):
    if filename.lower().endswith(('.lnk', '.iso')):
        return
    try:
        peobj = pefile.PE(filename, fast_load=True)
    except pefile.PEFormatError:
        # Not a PE file
        return
    check_manifest(filename, peobj)
    check_baseaddress(filename, peobj)


def run_dir(target):
    target_files = []
    source_dir = None

    for line in open(os.path.join(target, 'cmake_install.cmake'), 'r'):
        if line.startswith(CMAKE_PREFIX):
            source_dir = line[len(CMAKE_PREFIX):].strip()
            break

    if not source_dir:
        print 'Unable to find source dir'
        return

    for lst in ['bootcd.lst', 'hybridcd.lst', 'livecd.lst', 'bootcdregtest.lst']:
        for line in open(os.path.join(target, 'boot', lst), 'r'):
            result = re.match(r'.*=(.*)', line)
            if result:
                target_files.append(os.path.normpath(result.group(1)))

    # read cab contents
    dyn_file = os.path.join(target, 'boot', 'bootdata', 'packages', 'reactos.dff.dyn')
    for line in open(dyn_file, 'r'):
        result = re.match(r'"(.*)" \d+', line)
        if result:
            target_files.append(os.path.normpath(os.path.join(source_dir, result.group(1))))

    target_files = list(set(target_files))

    # for root, dirs, files in os.walk(target, topdown=True):
    #     dirs[:] = [d for d in dirs if d not in ['CMakeFiles', 'host-tools']]
    #     for curfile in files:
    #         lower = curfile.lower()
    #         if lower.endswith(INCLUDE_FILES):
    #             target_files.append(os.path.join(root, curfile))
    for curfile in target_files:
        analyze(curfile)
    print 'Done'

def main(dirs):
    for curdir in dirs:
        run_dir(curdir)

if __name__ == '__main__':
    main(sys.argv[1:])