zip_meta.py

import hashlib
from io import BytesIO
from zipfile import ZipFile
from operator import itemgetter


def read_zip(contentBytes):
    zip_metadata = dict()
    file_list = None
    errors = None
    try:
        zip_bytes = BytesIO(contentBytes)
        zip_object = ZipFile(zip_bytes)
        file_list = [read_zip_file(z, zip_object) for z in zip_object.filelist]
        if len(file_list) > 30:
            file_list = sorted(file_list, key=itemgetter("size"), reverse=True)[0:30]
            file_list.append("(List Truncated)")
        if file_list is not None:
            zip_metadata["contents"] = file_list

    except Exception as e:
        if str(e).strip() != "":
            errors = str(e).strip()[0:100]
    finally:
        if bool(zip_metadata) is True:
            return zip_metadata
        elif errors is not None:
            return {"error": errors}
        else:
            return None


def read_zip_file(zip_file, zip_object):
    zip_info = dict()
    try:
        zip_info["name"] = zip_file.filename[0:100]
        if zip_file.is_dir() is False:
            zip_info["size"] = zip_file.file_size
            if bool(zip_file.flag_bits & 0x1) is True:
                zip_info["protected"] = True
            else:
                try:
                    zip_file_bytes = zip_object.read(zip_file.filename)
                    zip_info["md5"] = hashlib.md5(zip_file_bytes).hexdigest()
                    zip_info["sha1"] = hashlib.sha1(zip_file_bytes).hexdigest()
                except Exception as e:
                    zip_info["error"] = str(e).strip()[0:100]
                    pass
    finally:
        return zip_info