fast way to get the minSdkVersion from an APK

get-axml.py

#!/usr/bin/python3
# encoding: utf-8

import os
import struct
import sys
# import zipfile
import zlib

from collections import namedtuple
from androguard.core.bytecodes import axml

ZipData = namedtuple("ZipData", ("cd_offset", "eocd_offset", "cd_and_eocd"))


class ZipError(Exception):
    pass


# def get_android_manifest(apkfile):
#     with zipfile.ZipFile(apkfile, "r") as apk:
#         with apk.open("AndroidManifest.xml") as manifest:
#             return manifest.read()


def get_android_manifest(apkfile):
    zdata = zip_data(apkfile)
    with open(apkfile, "rb") as fhi:
        fhi.seek(zdata.cd_offset)
        while True:
            hdr = fhi.read(46)
            if hdr[:4] != b"\x50\x4b\x01\x02":
                break
            n, m, k = struct.unpack("<HHH", hdr[28:34])
            hdr += fhi.read(n + m + k)
            name = hdr[46:46 + n]
            if name == b"AndroidManifest.xml":
                header_offset = int.from_bytes(hdr[42:46], "little")
                compress_size = int.from_bytes(hdr[20:24], "little")
                compress_type = int.from_bytes(hdr[10:12], "little")
                if compress_type != 8:
                    raise ZipError("Expected deflate compression")
                fhi.seek(header_offset)
                lhr = fhi.read(30)
                if lhr[:4] != b"\x50\x4b\x03\x04":
                    raise ZipError("Expected local file header signature")
                ln, lm = struct.unpack("<HH", lhr[26:30])
                lhr += fhi.read(ln + lm)
                return zlib.decompress(fhi.read(compress_size), -15)
    raise RuntimeError("AndroidManifest.xml not found")


def zip_data(apkfile, count=1024):
    with open(apkfile, "rb") as fh:
        fh.seek(-count, os.SEEK_END)
        data = fh.read()
        pos = data.rfind(b"\x50\x4b\x05\x06")
        if pos == -1:
            raise ZipError("Expected end of central directory record (EOCD)")
        fh.seek(pos - len(data), os.SEEK_CUR)
        eocd_offset = fh.tell()
        fh.seek(16, os.SEEK_CUR)
        cd_offset = int.from_bytes(fh.read(4), "little")
        fh.seek(cd_offset)
        cd_and_eocd = fh.read()
    return ZipData(cd_offset, eocd_offset, cd_and_eocd)


def get_minsdkversion_from_binary_android_manifest(data):
    result = 1
    parser = axml.AXMLParser(data)
    while parser.is_valid():
        event = next(parser)
        if event == axml.END_DOCUMENT:
            break
        # FIXME: depth == 2 ?!
        if event == axml.START_TAG and parser.name == "uses-sdk" \
                and parser.namespace == "":
            for i in range(parser.getAttributeCount()):
                if parser.getAttributeName(i) == "minSdkVersion":
                    # FIXME: VALUE_TYPE_STRING -> getMinSdkVersionForCodename()
                    val = parser.getAttributeValueData(i)
                    result = max(result, val)
                    # return result
    return result


if __name__ == "__main__":
    # from androguard.core.bytecodes.apk import APK
    for apkfile in sys.argv[1:]:
        data = get_android_manifest(apkfile)
        print(get_minsdkversion_from_binary_android_manifest(data))
        # apk = APK(apkfile)
        # print(apk.get_min_sdk_version())

# vim: set tw=80 sw=4 sts=4 et fdm=marker :