extract_docs_from_zip.py

import os
import shutil
import sys
import zipfile


def copy_docs(zip_path, p):
    try:
        with zipfile.ZipFile(zip_path) as z:
            targets = [fn for fn in z.namelist() if fn.endswith('doc') or fn.endswith('docx')]
            for fn in targets:
                with z.open(fn) as zf:
                    try:
                        newpath = os.path.join(p, fn.split('\\')[-1])
                        if not os.path.exists(newpath):
                            with open(newpath, 'wb') as f:
                                shutil.copyfileobj(zf, f)
                    except IOError:
                        newpath = os.path.join(p, fn.split('\\')[-1][-30:])
                        if not os.path.exists(newpath):
                            with open(newpath, 'wb') as f:
                                shutil.copyfileobj(zf, f)
    except zipfile.BadZipfile:
        print(zip_path)


if __name__=='__main__':
    args = sys.argv
    assert len(args) == 3
    zip_dir = args[1]
    p = args[2]
    for cur, dirs, files in os.walk(zip_dir):
        subdir = os.path.join(p, os.path.basename(cur))
        for fn in files:
            zip_path = os.path.join(cur, fn)
            target_dir = os.path.join(subdir, os.path.basename(zip_path).replace('.zip', ''))
            if not os.path.exists(target_dir):
                os.makedirs(target_dir)

            copy_docs(zip_path, target_dir)