get_crx_manifests.py

#!/usr/bin/env python
import json
import requests
import StringIO
import sys
import zipfile


def download_crx(identifier):
    sys.stderr.write("Downloading {0}\n".format(identifier))
    try:
        response = requests.get('https://clients2.google.com/service/update2/crx?response=redirect&os=mac&arch=x64&os_arch=x86_64&nacl_arch=x86-64&prod=chromecrx&prodchannel=canary&prodversion=74.0.3686.0&lang=en-US&acceptformat=crx2,crx3&x=id%3D{identifier}%26installsource%3Dondemand%26uc'.format(identifier=identifier))
        sio = StringIO.StringIO(response.content)
        zf = zipfile.ZipFile(sio)
        return zf
    except Exception as e:
        sys.stderr.write("Failed downloading {0}: {1}\n".format(identifier, e))
        return None

def extract_manifest(crx_file):
    try:
        manifest = crx_file.read('manifest.json')
        manifest = json.loads(manifest)
        return manifest
    except Exception:
        sys.stderr.write("Failed extracting manifest\n")
        return {}




if __name__ == '__main__':
    import sys

    chrome_identifiers = open(sys.argv[1], 'rb').read().splitlines()

    manifests = {}

    for chrome_identifier in chrome_identifiers:
        crx = download_crx(chrome_identifier)
        if crx:
            manifests[chrome_identifier] = extract_manifest(crx)

    sys.stdout.write(json.dumps(manifests))