Guess Tag From Filename

gtff.py

#!/usr/bin/env python

# Guess Tags From Filename, using python and mutagen.

import mutagen
from mutagen.easyid3 import EasyID3
import re
import os.path
import sys

def tag(filename, regex, constant_tags):
    basename = os.path.basename(filename)
    basename = basename[:basename.rindex('.')]  # strip extension
    m = regex.search(basename)
    if not m:
        raise Exception("Regex {} doesn't match {}".format(regex.pattern, filename))
    f = mutagen.File(filename, easy=True)
    for name, i in regex.groupindex.items():
        f[name] = m.group(i)
    for k, v in constant_tags.items():
        f[k] = v
    f.save()

def pattern_to_regex(pattern):
    # Escape everything, but then unescape angle brackets to make conversion
    # into a regex easier.
    # TODO: Imbalanced angle brackets in the pattern behave poorly.
    pattern = re.escape(pattern)
    pattern = re.sub(r'\\([<>])', r'\1', pattern)
    pattern = re.sub('(<[^>]*>)', r'(?P\1.*?)', pattern)
    pattern += '$'
    regex = re.compile(pattern)
    tags = regex.groupindex.keys()
    invalid_tags = [t for t in tags if t not in EasyID3.valid_keys]
    if invalid_tags:
        raise Exception('Invalid tag names: {}'.format(invalid_tags))
    return regex

if __name__ == '__main__':
    import argparse
    parser = argparse.ArgumentParser()
    parser.add_argument('-p', '--pattern', default='<artist> - <title>', help='filename pattern, eg: <artist> - <title>. An empty pattern can be specified so only constant tags are applied.')
    parser.add_argument('-t', '--tag', metavar='TAGNAME=VALUE', action='append', default=[], help='specify a constant tag')
    parser.add_argument('--valid-tags', action='store_true', help="print valid tag names for mutagen's EasyMP3 interface")
    parser.add_argument('files', nargs='*')
    args = parser.parse_args()

    if args.valid_tags:
        print '\n'.join(sorted(EasyID3.valid_keys.keys()))
        sys.exit(0)

    constant_tags = {}
    invalid_tags = []
    for t in args.tag:
        k, v = t.split('=', 1)
        if k not in EasyID3.valid_keys:
            invalid_tags.append(k)
        constant_tags[k] = v
    if invalid_tags:
        raise Exception('Invalid tag names: {}'.format(invalid_tags))

    regex = pattern_to_regex(args.pattern)
    for f in args.files:
        tag(f, regex, constant_tags)