vcs.maemo.org Git - mussorgsky/blob - src/parsefilename.py

   1 # This file came originally from
   2 # Picard, the next-generation MusicBrainz tagger
   3 #   This specific file didn't contain a Copyright notice
   4 #   but the whole project is under GPL 2 or later
   5 #
   6 # Original project URL:
   7 # https://code.launchpad.net/picard
   8 #
   9 # Original file URL:
  10 # http://bazaar.launchpad.net/~garyvdm/picard/keep_copy/annotate/head%3A/picard/parsefilename.py
  11 #
  12 # Modifications by Ivan Frade <ivan.frade@gmail.com>
  13 #   Added new regular expresion
  14 #   Added function to clean a little bit the filename before processing
  15 #
  16
  17 # ParseFilename - Infer metadata from filepath
  18 # Copyright (C) 2008?, Picard
  19 #
  20 # This program is free software; you can redistribute it and/or modify
  21 # it under the terms of the GNU General Public License as published by
  22 # the Free Software Foundation; either version 2 of the License, or
  23 # (at your option) any later version.
  24 #
  25 # This program is distributed in the hope that it will be useful,
  26 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  27 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  28 # GNU General Public License for more details.
  29 #
  30 # You should have received a copy of the GNU General Public License
  31 # along with this program; if not, write to the Free Software Foundation,
  32 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
  33 #
  34 import re
  35
  36 _patterns = [
  37     # AlbumArtist/1999 - Album/01-TrackTitle.ext
  38     re.compile(r"(?:.*(/|\\))?(?P<artist>.*)(/|\\)((?P<year>\d{4}) - )(?P<album>.*)(/|\\)(?P<tracknum>\d{2})-(?P<title>.*)\.(?:\w{2,5})$"),
  39     # AlbumArtist - Album/01 - TrackTitle.ext
  40     re.compile(r"(?:.*(/|\\))?(?P<artist>.*) - (?P<album>.*)(/|\\)(?P<tracknum>\d{2}) - (?P<title>.*)\.(?:\w{2,5})$"),
  41     # AlbumArtist - Album/01-TrackTitle.ext
  42     re.compile(r"(?:.*(/|\\))?(?P<artist>.*) - (?P<album>.*)(/|\\)(?P<tracknum>\d{2})-(?P<title>.*)\.(?:\w{2,5})$"),
  43     # AlbumArtist - Album/01. TrackTitle.ext
  44     re.compile(r"(?:.*(/|\\))?(?P<artist>.*) - (?P<album>.*)(/|\\)(?P<tracknum>\d{2})\. (?P<title>.*)\.(?:\w{2,5})$"),
  45     # AlbumArtist - Album/01 TrackTitle.ext
  46     re.compile(r"(?:.*(/|\\))?(?P<artist>.*) - (?P<album>.*)(/|\\)(?P<tracknum>\d{2}) (?P<title>.*)\.(?:\w{2,5})$"),
  47     # AlbumArtist - Album/01_Artist_-_TrackTitle.ext
  48     re.compile(r"(?:.*(/|\\))?(?P<albumartist>.*) - (?P<album>.*)(/|\\)(?P<tracknum>\d{2})_(?P<artist>.*)_-_(?P<title>.*)\.(?:\w{2,5})$"),
  49     # Album/Artist - Album - 01 - TrackTitle.ext
  50     re.compile(r"(?:.*(/|\\))?(?P<artist>.*)(/|\\)(?P=artist) - (?P<album>.*) - (?P<tracknum>\d{2}) - (?P<title>.*)\.(?:\w{2,5})$"),
  51     # AlbumArtist/Album/Artist - 01 - TrackTitle.ext
  52     re.compile(r"(?:.*(/|\\))?(?P<albumartist>.*)(/|\\)(?P<album>.*)(/|\\)(?P<artist>.*) - (?P<tracknum>\d{2}) - (?P<title>.*)\.(?:\w{2,5})$"),
  53     # AlbumArtist/Album/01. Artist - TrackTitle.ext
  54     re.compile(r"(?:.*(/|\\))?(?P<albumartist>.*)(/|\\)(?P<album>.*)(/|\\)(?P<tracknum>\d{2})\. (?P<artist>.*) - (?P<title>.*)\.(?:\w{2,5})$"),
  55     # AlbumArtist/Album/01 - Artist - TrackTitle.ext
  56     re.compile(r"(?:.*(/|\\))?(?P<albumartist>.*)(/|\\)(?P<album>.*)(/|\\)(?P<tracknum>\d{2}) - (?P<artist>.*) - (?P<title>.*)\.(?:\w{2,5})$"),
  57     # AlbumArtist/Album/01 - TrackTitle.ext
  58     re.compile(r"(?:.*(/|\\))?(?P<artist>.*)(/|\\)(?P<album>.*)(/|\\)(?P<tracknum>\d{2}) - (?P<title>.*)\.(?:\w{2,5})$"),
  59     # AlbumArtist/Album/01. TrackTitle.ext
  60     re.compile(r"(?:.*(/|\\))?(?P<artist>.*)(/|\\)(?P<album>.*)(/|\\)(?P<tracknum>\d{2})\. (?P<title>.*)\.(?:\w{2,5})$"),
  61     # AlbumArtist/Album/01 TrackTitle.ext
  62     re.compile(r"(?:.*(/|\\))?(?P<artist>.*)(/|\\)(?P<album>.*)(/|\\)(?P<tracknum>\d{2}) (?P<title>.*)\.(?:\w{2,5})$"),
  63     # AlbumArtist/Album/Album-01-TrackTitle.ext
  64     re.compile(r"(?:.*(/|\\))?(?P<albumartist>.*)(/|\\)(?P<album>.*)(/|\\)(?P=album)-(?P<tracknum>\d{2})-(?P<artist>.*)-(?P<title>.*)\.(?:\w{2,5})$"),
  65     # AlbumArtist/Album/Album-01-Artist-TrackTitle.ext
  66     re.compile(r"(?:.*(/|\\))?(?P<artist>.*)(/|\\)(?P<album>.*)(/|\\)(?P=album)-(?P<tracknum>\d{2})-(?P<title>.*)\.(?:\w{2,5})$"),
  67     # AlbumArtist/Album/Artist-01-TrackTitle.ext
  68     re.compile(r"(?:.*(/|\\))?(?P<albumartist>.*)(/|\\)(?P<album>.*)(/|\\)(?P<artist>.*)-(?P<tracknum>\d{2})-(?P<title>.*)\.(?:\w{2,5})$"),
  69     # Artist/Album/TrackTitle.ext
  70     re.compile(r"(?:.*(/|\\))?(?P<artist>.*)(/|\\)(?P<album>.*)(/|\\)(?P<title>.*)\.(?:\w{2,5})$"),
  71 ]
  72
  73 def parseFileName (filename):
  74     for pattern in _patterns:
  75         match = pattern.match(filename)
  76         if match:
  77             metadata = {}
  78             #print "Match with pattern ", counter
  79             metadata["artist"] = match.group("artist")
  80             metadata["title"] = match.group("title")
  81             metadata["album"] = match.group("album")
  82             return metadata
  83     return None
  84
  85 def clean (filename):
  86     """
  87     Some brute force normalization
  88     """
  89     f = re.sub (r"\[.*\d\]", r"", filename) # [Disc 1]
  90     f = re.sub (r"[Dd]isc \d", r"", f)  # Disc 1
  91     f = re.sub (r"-?[Cc][Dd]\ ?\d", r"", f) # CD1, CD 1 <- Maybe useful!
  92     f = re.sub (r"([a-z0-9])([A-Z])([a-z0-9])", r"\1 \2\3", f) # PinkFloyd -> Pink Floyd
  93     return f.replace ("_", " ").replace ("["," ").replace ("]", " ").replace ("-", " ").replace ("  ", " ")
  94
  95
  96 def crawl (path):
  97     """
  98     Test function. Crawl a directory trying to guess the metadata from the filepath
  99     for all music files.
 100     """
 101     import os
 102     print path
 103     fail = 0
 104     ok = 0
 105     for (root, dirs, files) in os.walk (path):
 106         for f in files:
 107             if (f.endswith (".mp3")
 108                 or f.endswith (".m4a")
 109                 or f.endswith (".ogg")):
 110                 filename = os.path.join (root, f)
 111                 print filename
 112                 mdata = parseFileName (clean (filename))
 113                 if (not mdata):
 114                     fail += 1
 115                 else:
 116                     print "  Artist: ", mdata['artist']
 117                     print "  Album :", mdata['album']
 118                     print "  Title :", mdata['title']
 119                     ok += 1
 120     print "aprox %d success out of %d" % (ok, ok+fail)
 121
 122
 123 if __name__ == "__main__":
 124     crawl ("/home/ivan/Desktop")
 125     #crawl ("/media/IPOD/")