@@ -357,6 +357,10 @@ - [X] track2xmcd - [X] tracksplit - [X] tracktag +* DONE Check for FLAC metadata chunk overflow + Although APEv2 and ID3 tags support very large objects (hundreds of MB), + FLAC metadata chunks have a maxmimum of about 16MB per chunk, + which may be hit accidentally. * TODO Ensure non-image cuesheets are handled properly There are 4 different types of cuesheets, 1 single file type and 3 different multiple file variants. @@ -484,3 +488,4 @@ * TODO Fix or replace Python's built-in aifc module The current implementation suffers from bugs. ** TODO Document AIFF better + TODO @@ -229,25 +229,19 @@ class FlacMetaData(MetaData): #STREAMINFO must always be first and is always a fixed size built_blocks.append(blocks.next().build_block()) - #VORBISCOMMENT must always be next - vorbis = blocks.next() - try: - built_blocks.append(vorbis.build_block()) - except FlacMetaDataBlockTooLarge: - #if VORBISCOMMENT is too large, substitute a blank one - #(this only happens when one pushes over 16MB(!) of text - # into a comment, which simply isn't going to happen - # accidentcally) - built_blocks.append(FlacVorbisComment( - vorbis_data={}, - vendor_string=vorbis.vendor_string).build_block()) - #then come the rest of the blocks in any order for block in blocks: try: built_blocks.append(block.build_block()) except FlacMetaDataBlockTooLarge: - pass + if (isinstance(block,VorbisComment)): + #if VORBISCOMMENT is too large, substitute a blank one + #(this only happens when one pushes over 16MB(!) of text + # into a comment, which simply isn't going to happen + # accidentcally) + built_blocks.append(FlacVorbisComment( + vorbis_data={}, + vendor_string=block.vendor_string).build_block()) #finally, append a fresh PADDING block built_blocks.append( @@ -583,7 +577,15 @@ class FlacAudio(AudioFile): @classmethod def is_type(cls, file): if (file.read(4) == 'fLaC'): - return True + block_ids = list(cls.__block_ids__(file)) + if ((len(block_ids) == 0) or (0 not in block_ids)): + messenger = Messenger("audiotools",None) + messenger.error(_(u"STREAMINFO block not found")) + elif (block_ids[0] != 0): + messenger = Messenger("audiotools",None) + messenger.error(_(u"STREAMINFO not first metadata block. Please fix with tracklint(1)")) + else: + return True else: #I've seen FLAC files tagged with ID3v2 comments. #Though the official flac binaries grudgingly accept these, @@ -594,7 +596,7 @@ class FlacAudio(AudioFile): ID3v2Comment.skip(file) if (file.read(4) == 'fLaC'): messenger = Messenger("audiotools",None) - messenger.error(_(u"ID3v2 tag found at start of FLAC file. Please remove.")) + messenger.error(_(u"ID3v2 tag found at start of FLAC file. Please remove with tracklint(1)")) return False def lossless(self): @@ -740,6 +742,17 @@ class FlacAudio(AudioFile): p = FlacAudio.METADATA_BLOCK_HEADER.parse(flacfile.read(4)) return (p.last_block, p.block_type, p.block_length) + @classmethod + def __block_ids__(cls, flacfile): + p = Con.Container(last_block=False, + block_type=None, + block_length=0) + + while (not p.last_block): + p = FlacAudio.METADATA_BLOCK_HEADER.parse_stream(flacfile) + yield p.block_type + flacfile.seek(p.block_length,1) + def set_cuesheet(self,cuesheet): if (cuesheet is None): return audiotools/__flac__.py @@ -97,7 +97,7 @@ MUSICBRAINZ_PORT = config.getint_default("MusicBrainz","port",80) THUMBNAIL_FORMAT = config.get_default("Thumbnail","format","jpeg") THUMBNAIL_SIZE = config.getint_default("Thumbnail","size",150) -VERSION = "2.13alpha3" +VERSION = "2.13beta1" FILENAME_FORMAT = config.get_default( "Filenames","format", audiotools/__init__.py @@ -65,6 +65,8 @@ For example, a track name value of \[lq]\fCSome Title \fR\[rq] becomes \[bu] Empty fields are removed. .PP \[bu] ID3v2 and ID3v1 tags are stripped from FLAC files. +.PP +\[bu] FLAC files with misordered STREAMINFO blocks are reordered. .SH SEE ALSO .BR cd2track (1), .BR cd2xmcd (1), docs/tracklint.1 @@ -91,7 +91,7 @@ extensions = [cdiomodule, # pass #pkg-config not available setup (name = 'Python Audio Tools', - version = '2.13alpha3', + version = '2.13beta1', description = 'A collection of audio handling utilities', author = 'Brian Langenberger', author_email = 'tuffy@users.sourceforge.net', setup.py @@ -36,7 +36,7 @@ import time gettext.install("audiotools",unicode=True) (METADATA,PCM,EXECUTABLE,CUESHEET,IMAGE,CUSTOM) = range(6) -CASES = set([METADATA,PCM,EXECUTABLE,CUESHEET,IMAGE]) +CASES = set([METADATA]) def nothing(self): pass test/test.py @@ -47,10 +47,7 @@ def checksum(path): YEAR = re.compile(r'^\d{4}$') TEXT_FIELDS = tuple([field for field in audiotools.MetaData.__FIELDS__ - if (field not in ("track_number", - "album_number", - "track_total", - "album_total"))]) + if (field not in audiotools.MetaData.__INTEGER_FIELDS__)]) def display_messages(messenger,track,messages): for message in messages: @@ -71,6 +68,10 @@ def fix_track(track,dry_run=False): if (not dry_run): track = track.fix_id3() log.append(_(u"Removed unnecessary ID3 tags")) + elif (isinstance(track,DisorderedFlacAudio)): + if (not dry_run): + track = track.fix_blocks() + log.append(_(u"Reordered metadata blocks")) return (track,log) @@ -406,17 +407,60 @@ class BrokenFlacAudio(audiotools.FlacAudio): return audiotools.open(self.filename) +class DisorderedFlacAudio(audiotools.FlacAudio): + NAME = "disorderedflac" + + def __read_streaminfo__(self): + f = file(self.filename,"rb") + f.read(4) + + (stop,header_type,length) = (False,None,0) + while (not stop): + (stop,header_type,length) = audiotools.FlacAudio.__read_flac_header__(f) + if (header_type == 0): + p = audiotools.FlacAudio.STREAMINFO.parse(f.read(length)) + + md5sum = "".join(["%.2X" % (x) for x in p.md5]).lower() + + self.__samplerate__ = p.samplerate + self.__channels__ = p.channels + 1 + self.__bitspersample__ = p.bits_per_sample + 1 + self.__total_frames__ = p.total_samples + self.__md5__ = "".join([chr(c) for c in p.md5]) + break + else: + f.seek(length,1) + f.close() + + #rewrites the FLAC track with its metadata blocks in the proper order + def fix_blocks(self): + metadata = self.get_metadata() + + self.set_metadata(metadata) + + return audiotools.open(self.filename) + + @classmethod + def is_type(cls,file): + header = file.read(4) + first_block_type = ord(file.read(1)) + return (header == 'fLaC') and ((first_block_type & 0x7) != 0) + class PureFlacAudio(audiotools.FlacAudio): NAME = "flac" @classmethod def is_type(cls,file): - return (file.read(4) == 'fLaC') + header = file.read(4) + first_block_type = ord(file.read(1)) + return (header == 'fLaC') and ((first_block_type & 0x7) == 0) if (audiotools.FlacAudio in audiotools.AVAILABLE_TYPES): audiotools.AVAILABLE_TYPES = tuple([t for t in audiotools.AVAILABLE_TYPES if t != audiotools.FlacAudio] + \ - [BrokenFlacAudio,PureFlacAudio]) + [BrokenFlacAudio, + DisorderedFlacAudio, + PureFlacAudio]) audiotools.TYPE_MAP = dict([(track_type.NAME,track_type) for track_type in audiotools.AVAILABLE_TYPES tracklint d5b478c10ff7f9dcb4459798e4afe9d2fa172016 Brian Langenberger bjl@usa.net http://github.com/tuffy/python-audio-tools/commit/41e55aca14a06d0375194f73b64903bb02472805 41e55aca14a06d0375194f73b64903bb02472805 2009-09-30T16:45:46-07:00 2009-09-30T16:45:46-07:00 Add a tracklint fix for FLAC files that don't have STREAMINFO as the first block. Nudged version to 2.13beta1, for real this time. def51d3b70aef989dbc9f0505760469fc7b7f4a8 Brian Langenberger bjl@usa.net