xtract.py

import argparse
import os # for file walks
import sys # for error handling
import traceback

import patoolib # to extract archives and compressed data, depending on their file type
import magic # to determine files type not be extension but by their content. mimetype won't work, cannot recognize DICOM files

RUNS = 0
MAX_RUNS = 10

parser = argparse.ArgumentParser()
parser.add_argument('dir', help="directory to extract all files within")
args = parser.parse_args()

def compressed_data_exists(dir):
    for root, subdirs, files in os.walk(os.path.abspath(dir)):
        for fname in files:
            ftype = magic.from_file(os.path.join(root, fname))
            if 'compressed data' in ftype or 'archive data' in ftype:
                return True
    return False

errors = {}
noarchives = []

while compressed_data_exists(args.dir) and RUNS < MAX_RUNS:
    for root, subdirs, files in os.walk(os.path.abspath(args.dir)):
        for fname in filter(lambda fn: fn not in errors and fn not in noarchives, files):
            abs_fname = os.path.join(root, fname)
            ftype = magic.from_file(abs_fname)
            if 'compressed data' in ftype or 'archive data' in ftype:
                try:
                    h,t = os.path.split(fname)
                    patoolib.extract_archive(abs_fname, outdir=os.path.join(root,h))
                except FileExistsError as e:
                    print(f"Could not extract {abs_fname}, destination already exists", file=sys.stderr)
                except Exception as e:
                    print(f"Failure while extracting: {abs_fname}", file=sys.stderr)
                    print(f"Error: {sys.exc_info()[0]}", file=sys.stderr)
                    print(e)
                    errors[fname]=e
                    continue
                else:
                    os.remove(abs_fname) # remove the compressed data, as it is extracted now
            else:
                noarchives.append(fname)
    RUNS += 1

print(f'{RUNS} runs completed')

if errors:
    print('There were errors with the following files:')
    for k,v in errors.items():
        print(f'file: {k}')
        print(f'cmd: {v}')
    sys.exit(1)