mirror of
https://github.com/QuasarApp/LIEF.git
synced 2025-04-28 21:34:32 +00:00
235 lines
8.1 KiB
Python
235 lines
8.1 KiB
Python
|
#!/usr/bin/env python2
|
|||
|
|
#-*- coding: utf-8 -*-
|
|||
|
|
|
|||
|
|
# This file is used to generate config file for LIEF test
|
|||
|
|
# Basically it parse it parse the output of readelf
|
|||
|
|
|
|||
|
|
import re
|
|||
|
|
import yaml
|
|||
|
|
import subprocess
|
|||
|
|
import sys
|
|||
|
|
import hashlib
|
|||
|
|
|
|||
|
|
p = subprocess.Popen(["readelf", "-a", sys.argv[1]], stdout=subprocess.PIPE)
|
|||
|
|
(output, err) = p.communicate()
|
|||
|
|
|
|||
|
|
data = output
|
|||
|
|
binary = dict()
|
|||
|
|
header = dict()
|
|||
|
|
|
|||
|
|
#ELF64
|
|||
|
|
elf64_regex = {
|
|||
|
|
"entrypoint" : "Adresse du point d'entrée:\s+0x([0-9a-f]+)",
|
|||
|
|
"sectionoffset" : "Début des en-têtes de section\s*:\s+([0-9]+)",
|
|||
|
|
"offsetToPhdr" : "Début des en-têtes de programme :\s+([0-9]+)",
|
|||
|
|
"nbShdr" : "Nombre d'en-têtes de section\s*:\s+([0-9]+)",
|
|||
|
|
"nbPhdr" : "Nombre d'en-tête du programme\s*:\s+([0-9]+)",
|
|||
|
|
|
|||
|
|
"sections" : ur'\[\s*(\d+)\]\s?(\S+|\s*)\s+\S+\s+([0-9A-Fa-f]+)\s+([0-9A-Fa-f]+)\n\s+([0-9A-Fa-f]+)\s+([0-9A-Fa-f]+)',
|
|||
|
|
"segments" : ur'\s+(\w+)\s+0x([0-9A-Fa-f]+)\s+0x([0-9A-Fa-f]+)\s+0x([0-9A-Fa-f]+)\n\s+0x([0-9A-Fa-f]+)\s+0x([0-9A-Fa-f]+)',
|
|||
|
|
|
|||
|
|
"dynamicSymGlob": ur'^T(?:.*)\.dynsym.*\n((?:\s{2,}(?:.*)\n)+)',
|
|||
|
|
"dynamicSymLoc" : ur'([0-9]+):\s+[0-9a-fA-F]+\s+[0-9]+\s+\S+\s+\S+\s+\S+\s+\S+\s?([^@\n]*)',
|
|||
|
|
|
|||
|
|
"staticSymGlob" : ur'^T(?:.*)\.symtab.*\n((?:.*\n)+)$\s',
|
|||
|
|
"staticSymLoc" : ur'([0-9]+):\s+[0-9a-fA-F]+\s+[0-9]+\s+\S+\s+\S+\s+\S+\s+\S+ (\S*)[@\n]+',
|
|||
|
|
|
|||
|
|
"dynRelocaGlob" : ur'^S(?:.*)\.rel[a]*\.dyn.*\n((?:.{2,}(?:.*)\n)+)',
|
|||
|
|
"dynRelocaLoc" : ur'^([0-9a-fA-F]+)\s+([0-9a-fA-F]+)\s+\S+\s+([0-9a-fA-F]+)\s*(?:\n|(?:\s+(\S+)))',
|
|||
|
|
|
|||
|
|
"pltRelocaGlob" : ur'^S(?:.*)\.rel[a]*\.plt.*\n((?:.{2,}(?:.*)\n)+)',
|
|||
|
|
"pltRelocaLoc" : ur'^([0-9a-fA-F]+)\s+([0-9a-fA-F]+)\s+\S+\s+([0-9a-fA-F]+)\s*(?:\n|(?:\s+(\S+)))'
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
|
|||
|
|
# ELF32
|
|||
|
|
elf64_regex = {
|
|||
|
|
"entrypoint" : "Adresse du point d'entrée:\s+0x([0-9a-f]+)",
|
|||
|
|
"sectionoffset" : "Début des en-têtes de section\s*:\s+([0-9]+)",
|
|||
|
|
"offsetToPhdr" : "Début des en-têtes de programme :\s+([0-9]+)",
|
|||
|
|
"nbShdr" : "Nombre d'en-têtes de section\s*:\s+([0-9]+)",
|
|||
|
|
"nbPhdr" : "Nombre d'en-tête du programme\s*:\s+([0-9]+)",
|
|||
|
|
|
|||
|
|
"sections" : ur'\s+\[\s*([0-9]+)\]\s?(\s|\S+)\s+\S+\s+([0-9A-Fa-f]+)\s+([0-9A-Fa-f]+)\s+([0-9A-Fa-f]+)\s+([0-9A-Fa-f]+)',
|
|||
|
|
"segments" : ur'\s+(\w+)\s+0x([0-9A-Fa-f]+)\s+0x([0-9A-Fa-f]+)\s0x([0-9A-Fa-f]+)\s0x([0-9A-Fa-f]+)\s0x([0-9A-Fa-f]+)',
|
|||
|
|
|
|||
|
|
"dynamicSymGlob": ur'^T(?:.*)\.dynsym.*\n((?:\s{2,}(?:.*)\n)+)',
|
|||
|
|
"dynamicSymLoc" : ur'([0-9]+):\s+[0-9a-fA-F]+\s+[0-9]+\s+\S+\s+\S+\s+\S+\s+\S+\s?([^@\n]*)',
|
|||
|
|
|
|||
|
|
"staticSymGlob" : ur'^T(?:.*)\.symtab.*\n((?:.*\n)+)$\s',
|
|||
|
|
"staticSymLoc" : ur'([0-9]+):\s+[0-9a-fA-F]+\s+[0-9]+\s+\S+\s+\S+\s+\S+\s+\S+ (\S*)[@\n]+',
|
|||
|
|
|
|||
|
|
"dynRelocaGlob" : ur'^S(?:.*)\.rel[a]*\.dyn.*\n((?:.{2,}(?:.*)\n)+)',
|
|||
|
|
"dynRelocaLoc" : ur'^([0-9a-fA-F]+)\s+([0-9a-fA-F]+)\s+\S+\s+([0-9a-fA-F]+)\s*(?:\n|(?:\s+([^@\n]+)))',
|
|||
|
|
|
|||
|
|
"pltRelocaGlob" : ur'^S(?:.*)\.rel[a]*\.plt.*\n((?:.{2,}(?:.*)\n)+)',
|
|||
|
|
"pltRelocaLoc" : ur'^([0-9a-fA-F]+)\s+([0-9a-fA-F]+)\s+\S+\s+([0-9a-fA-F]+)\s*(?:\n|(?:\s+([^@\n]+)))'
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
#
|
|||
|
|
# File info
|
|||
|
|
#
|
|||
|
|
binary["filename"] = str(sys.argv[1]).split("/")[-1]
|
|||
|
|
binary["hash"] = hashlib.md5(sys.argv[1]).hexdigest()
|
|||
|
|
#path: "@CMAKE_CURRENT_SOURCE_DIR@/samples/ELF/x86-64/binaries/ls"
|
|||
|
|
|
|||
|
|
#
|
|||
|
|
# header
|
|||
|
|
#
|
|||
|
|
entrypoint = re.search(elf64_regex["entrypoint"], data).groups()[0]
|
|||
|
|
sectionoffset = re.search(elf64_regex["sectionoffset"], data).groups()[0]
|
|||
|
|
offsetToPhdr = re.search(elf64_regex["offsetToPhdr"], data).groups()[0]
|
|||
|
|
nbShdr = re.search(elf64_regex["nbShdr"], data).groups()[0]
|
|||
|
|
nbPhdr = re.search(elf64_regex["nbPhdr"], data).groups()[0]
|
|||
|
|
header['entryPoint'] = int(entrypoint, 16)
|
|||
|
|
header['offsetToShdr'] = int(sectionoffset)
|
|||
|
|
header['offsetToPhdr'] = int(offsetToPhdr)
|
|||
|
|
header['nbShdr'] = int(nbShdr)
|
|||
|
|
header['nbPhdr'] = int(nbPhdr)
|
|||
|
|
|
|||
|
|
#
|
|||
|
|
# Sections
|
|||
|
|
#
|
|||
|
|
section_regexp = re.compile(elf64_regex["sections"], re.MULTILINE)
|
|||
|
|
sections_yaml = []
|
|||
|
|
sections = re.findall(section_regexp, data)
|
|||
|
|
for section in sections:
|
|||
|
|
section_yaml = {
|
|||
|
|
'nb' : int(section[0]),
|
|||
|
|
'name' : '%s' % (section[1].strip()),
|
|||
|
|
'address': int(section[2],16),
|
|||
|
|
'offset' : int(section[3],16),
|
|||
|
|
'size' : int(section[4],16)
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
sections_yaml.append(section_yaml)
|
|||
|
|
|
|||
|
|
#
|
|||
|
|
# Segments
|
|||
|
|
#
|
|||
|
|
segment_regexp = re.compile(elf64_regex["segments"], re.MULTILINE)
|
|||
|
|
segments = re.findall(segment_regexp, data)
|
|||
|
|
segments_yaml = []
|
|||
|
|
for segment in segments:
|
|||
|
|
segment_yaml = {
|
|||
|
|
'offset' : int(segment[1], 16),
|
|||
|
|
'vAddress': int(segment[2], 16),
|
|||
|
|
'pAddress': int(segment[3], 16),
|
|||
|
|
'fSize' : int(segment[4], 16),
|
|||
|
|
'vSize' : int(segment[5], 16)
|
|||
|
|
}
|
|||
|
|
segments_yaml.append(segment_yaml)
|
|||
|
|
|
|||
|
|
#
|
|||
|
|
# Relocations
|
|||
|
|
#
|
|||
|
|
#relocations_regexp = re.compile(ur'^([0-9a-fA-F]+)\s+([0-9a-fA-F]+)\s+(.\S+)\s+([0-9a-fA-F]+)\s+(.\S+)', re.MULTILINE)
|
|||
|
|
#relocations = re.findall(relocations_regexp, data)
|
|||
|
|
#relocations_yaml = []
|
|||
|
|
#for relocation in relocations:
|
|||
|
|
# relocation_yaml = {
|
|||
|
|
# 'offset': int(relocation[0], 16),
|
|||
|
|
# 'info' : int(relocation[1], 16),
|
|||
|
|
# 'name' : relocation[4]
|
|||
|
|
# }
|
|||
|
|
# if relocation_yaml not in relocations_yaml:
|
|||
|
|
# relocations_yaml.append(relocation_yaml);
|
|||
|
|
|
|||
|
|
#
|
|||
|
|
# Dynamic symboles
|
|||
|
|
#
|
|||
|
|
extract_regexp = re.compile(elf64_regex["dynamicSymGlob"], re.MULTILINE)
|
|||
|
|
dynsyms_yaml = []
|
|||
|
|
if len(re.findall(extract_regexp, data)) > 0:
|
|||
|
|
extracted = re.findall(extract_regexp, data)[0]
|
|||
|
|
dynsyms_regexp = re.compile(elf64_regex["dynamicSymLoc"], re.MULTILINE)
|
|||
|
|
dynsyms = re.findall(dynsyms_regexp, extracted)
|
|||
|
|
dynsyms_yaml = []
|
|||
|
|
for dynsym in dynsyms:
|
|||
|
|
dynsym_yaml = {
|
|||
|
|
'num' : int(dynsym[0]),
|
|||
|
|
'name': dynsym[1]
|
|||
|
|
}
|
|||
|
|
dynsyms_yaml.append(dynsym_yaml)
|
|||
|
|
|
|||
|
|
#
|
|||
|
|
# Static symbols
|
|||
|
|
#
|
|||
|
|
extract_regexp = re.compile(elf64_regex["staticSymGlob"], re.MULTILINE)
|
|||
|
|
staticsyms = []
|
|||
|
|
extracted = ""
|
|||
|
|
|
|||
|
|
if len(re.findall(extract_regexp, data)) > 0:
|
|||
|
|
extracted = re.findall(extract_regexp, data)[0]
|
|||
|
|
staticsyms_regexp = re.compile(elf64_regex["staticSymLoc"], re.MULTILINE)
|
|||
|
|
staticsyms = re.findall(staticsyms_regexp, extracted)
|
|||
|
|
|
|||
|
|
staticsyms_yaml = []
|
|||
|
|
|
|||
|
|
for staticsym in staticsyms:
|
|||
|
|
staticsym_yaml = {
|
|||
|
|
'num' : int(staticsym[0]),
|
|||
|
|
'name': staticsym[1]
|
|||
|
|
}
|
|||
|
|
staticsyms_yaml.append(staticsym_yaml)
|
|||
|
|
|
|||
|
|
|
|||
|
|
#
|
|||
|
|
# Dynamic Relocations
|
|||
|
|
#
|
|||
|
|
extract_regexp = re.compile(elf64_regex["dynRelocaGlob"], re.MULTILINE)
|
|||
|
|
relocations_dyn_yaml = []
|
|||
|
|
if len(re.findall(extract_regexp, data)) > 0:
|
|||
|
|
extracted = re.findall(extract_regexp, data)[0]
|
|||
|
|
regexp = re.compile(elf64_regex["dynRelocaLoc"], re.MULTILINE)
|
|||
|
|
relocations = re.findall(regexp, extracted)
|
|||
|
|
relocations_dyn_yaml = []
|
|||
|
|
for reloc in relocations:
|
|||
|
|
relocation_yaml = {
|
|||
|
|
'offset' : int(reloc[0], 16),
|
|||
|
|
'info' : int(reloc[1], 16),
|
|||
|
|
'value' : int(reloc[2], 16),
|
|||
|
|
'name' : reloc[3]
|
|||
|
|
}
|
|||
|
|
relocations_dyn_yaml.append(relocation_yaml)
|
|||
|
|
|
|||
|
|
#
|
|||
|
|
# .plt.got relocations
|
|||
|
|
#
|
|||
|
|
extract_regexp = re.compile(elf64_regex["pltRelocaGlob"], re.MULTILINE)
|
|||
|
|
extracted = re.findall(extract_regexp, data)[0]
|
|||
|
|
regexp = re.compile(elf64_regex["pltRelocaLoc"], re.MULTILINE)
|
|||
|
|
relocations = re.findall(regexp, extracted)
|
|||
|
|
relocations_plt_yaml = []
|
|||
|
|
for reloc in relocations:
|
|||
|
|
relocation_yaml = {
|
|||
|
|
'offset' : int(reloc[0], 16),
|
|||
|
|
'info' : int(reloc[1], 16),
|
|||
|
|
'value' : int(reloc[2], 16),
|
|||
|
|
'name' : reloc[3]
|
|||
|
|
}
|
|||
|
|
relocations_plt_yaml.append(relocation_yaml)
|
|||
|
|
|
|||
|
|
binary['Header'] = header
|
|||
|
|
binary['Sections'] = sections_yaml
|
|||
|
|
binary['Segments'] = segments_yaml
|
|||
|
|
#binary['Relocations'] = relocations_yaml
|
|||
|
|
if len(relocations_plt_yaml) > 0:
|
|||
|
|
binary['PltGotReloc'] = relocations_plt_yaml
|
|||
|
|
|
|||
|
|
if len(dynsyms_yaml) > 0:
|
|||
|
|
binary['DynamicSymbols'] = dynsyms_yaml
|
|||
|
|
|
|||
|
|
if len(relocations_dyn_yaml) > 0:
|
|||
|
|
binary['DynamicReloc'] = relocations_dyn_yaml
|
|||
|
|
|
|||
|
|
if len(staticsyms_yaml) > 0:
|
|||
|
|
binary['StaticSymbols'] = staticsyms_yaml
|
|||
|
|
|
|||
|
|
output = open(binary["filename"] + ".yaml", "w")
|
|||
|
|
yaml.dump(binary, stream=output)
|
|||
|
|
output.close()
|