id3reader.py

""" Read ID3 tags from a file.
    Ned Batchelder, http://nedbatchelder.com/code/modules/id3reader.html
"""

__version__ = '1.53.20070415'    # History at the end of the file.

# ID3 specs: http://www.id3.org/develop.html

import struct, sys, zlib

# These are the text encodings, indexed by the first byte of a text value.
_encodings = ['iso8859-1', 'utf-16', 'utf-16be', 'utf-8']

# Simple pseudo-id's, mapped to their various representations.
# Use these ids with getValue, and you don't need to know what
# version of ID3 the file contains.
_simpleDataMapping = {
    'album':        ('TALB', 'TAL', 'v1album', 'TOAL'),
    'artist':    ('TPE1', 'TP1', 'v1performer', 'TOPE'),
    'title':        ('TIT2', 'TT2', 'v1title'),
    'track':        ('TRCK', 'TRK', 'v1track'),
    'year':         ('TYER', 'TYE', 'v1year'),
    'genre':        ('TCON', 'TCO', 'v1genre'),
    'comment':      ('COMM', 'COM', 'v1comment'),
    'composer':     ('TCM'),
    'disc':         ('TPA'),
    'encoder':      ('TENC', 'TEN'),
}

# Provide booleans for older Pythons.
try:
    True, False
except NameError:
    True, False = 1==1, 1==0

# Tracing
_t = False
def _trace(msg):
    print msg

# Coverage
_c = False
_features = {}
def _coverage(feat):
    #if _t: _trace('feature '+feat)
    _features[feat] = _features.setdefault(feat, 0)+1

def _safestr(s):
    """ Get a good string for printing, that won't throw exceptions,
        no matter what's in it.
    """
    try:
        return unicode(s).encode(sys.getdefaultencoding())
    except UnicodeError:
        return '?: '+repr(s)

# Can I just say that I think the whole concept of genres is bogus,
# since they are so subjective?  And the idea of letting someone else pick
# one of these things and then have it affect the categorization of my music
# is extra bogus.  And the list itself is absurd. Polsk Punk?
genres = [
    # 0-19
    'Blues', 'Classic Rock', 'Country', 'Dance', 'Disco', 'Funk', 'Grunge', 'Hip - Hop', 'Jazz', 'Metal',
    'New Age', 'Oldies', 'Other', 'Pop', 'R&B', 'Rap', 'Reggae', 'Rock', 'Techno', 'Industrial',
    # 20-39
    'Alternative', 'Ska', 'Death Metal', 'Pranks', 'Soundtrack', 'Euro - Techno', 'Ambient', 'Trip - Hop', 'Vocal', 'Jazz + Funk',
    'Fusion', 'Trance', 'Classical', 'Instrumental', 'Acid', 'House', 'Game', 'Sound Clip', 'Gospel', 'Noise',
    # 40-59
    'Alt Rock', 'Bass', 'Soul', 'Punk', 'Space', 'Meditative', 'Instrumental Pop', 'Instrumental Rock', 'Ethnic', 'Gothic',
    'Darkwave', 'Techno - Industrial', 'Electronic', 'Pop - Folk', 'Eurodance', 'Dream', 'Southern Rock', 'Comedy', 'Cult', 'Gangsta Rap',
    # 60-79
    'Top 40', 'Christian Rap', 'Pop / Funk', 'Jungle', 'Native American', 'Cabaret', 'New Wave', 'Psychedelic', 'Rave', 'Showtunes',
    'Trailer', 'Lo - Fi', 'Tribal', 'Acid Punk', 'Acid Jazz', 'Polka', 'Retro', 'Musical', 'Rock & Roll', 'Hard Rock',
    # 80-99
    'Folk', 'Folk / Rock', 'National Folk', 'Swing', 'Fast - Fusion', 'Bebob', 'Latin', 'Revival', 'Celtic', 'Bluegrass',
    'Avantgarde', 'Gothic Rock', 'Progressive Rock', 'Psychedelic Rock', 'Symphonic Rock', 'Slow Rock', 'Big Band', 'Chorus', 'Easy Listening', 'Acoustic',
    # 100-119
    'Humour', 'Speech', 'Chanson', 'Opera', 'Chamber Music', 'Sonata', 'Symphony', 'Booty Bass', 'Primus', 'Porn Groove',
    'Satire', 'Slow Jam', 'Club', 'Tango', 'Samba', 'Folklore', 'Ballad', 'Power Ballad', 'Rhythmic Soul', 'Freestyle',
    # 120-139
    'Duet', 'Punk Rock', 'Drum Solo', 'A Cappella', 'Euro - House', 'Dance Hall', 'Goa', 'Drum & Bass', 'Club - House', 'Hardcore',
    'Terror', 'Indie', 'BritPop', 'Negerpunk', 'Polsk Punk', 'Beat', 'Christian Gangsta Rap', 'Heavy Metal', 'Black Metal', 'Crossover',
    # 140-147
    'Contemporary Christian', 'Christian Rock', 'Merengue', 'Salsa', 'Thrash Metal', 'Anime', 'JPop', 'Synthpop'
    ]

class Id3Error(Exception):
    """ An exception caused by id3reader properly handling a bad ID3 tag.
    """
    pass

class _Header:
    """ Represent the ID3 header in a tag.
    """
    def __init__(self):
        self.majorVersion = 0
        self.revision = 0
        self.flags = 0
        self.size = 0
        self.bUnsynchronized = False
        self.bExperimental = False
        self.bFooter = False

    def __str__(self):
        return str(self.__dict__)

class _Frame:
    """ Represent an ID3 frame in a tag.
    """
    def __init__(self):
        self.id = ''
        self.size = 0
        self.flags = 0
        self.rawData = ''
        self.bTagAlterPreserve = False
        self.bFileAlterPreserve = False
        self.bReadOnly = False
        self.bCompressed = False
        self.bEncrypted = False
        self.bInGroup = False

    def __str__(self):
        return str(self.__dict__)

    def __repr__(self):
        return str(self.__dict__)

    def _interpret(self):
        """ Examine self.rawData and create a self.value from it.
        """
        if len(self.rawData) == 0:
            # This is counter to the spec, but seems harmless enough.
            #if _c: _coverage('zero data')
            return

        if self.bCompressed:
            # Decompress the compressed data.
            self.rawData = zlib.decompress(self.rawData)

        if self.id[0] == 'T':
            # Text fields start with T
            encoding = ord(self.rawData[0])
            if 0 <= encoding < len(_encodings):
                #if _c: _coverage('encoding%d' % encoding)
                value = self.rawData[1:].decode(_encodings[encoding])
            else:
                #if _c: _coverage('bad encoding')
                value = self.rawData[1:]
            # Don't let trailing zero bytes fool you.
            if value:
                value = value.strip('\0')
            # The value can actually be a list.
            if '\0' in value:
                value = value.split('\0')
                #if _c: _coverage('textlist')
            self.value = value
        elif self.id[0] == 'W':
            # URL fields start with W
            self.value = self.rawData.strip('\0')
            if self.id == 'WXXX':
                self.value = self.value.split('\0')
        elif self.id == 'CDM':
            # ID3v2.2.1 Compressed Data Metaframe
            if self.rawData[0] == 'z':
                self.rawData = zlib.decompress(self.rawData[5:])
            else:
                #if _c: _coverage('badcdm!')
                raise Id3Error, 'Unknown CDM compression: %02x' % self.rawData[0]
            #@TODO: re-interpret the decompressed frame.

        elif self.id in _simpleDataMapping['comment']:
            # comment field

            # In limited testing a typical comment looks like
            # '\x00XXXID3v1 Comment\x00comment test' so in this
            # case we need to find the second \x00 to know where
            # where we start for a comment.  In case we only find
            # one \x00, lets just start at the beginning for the
            # value
            s = str(self.rawData)

            pos = 0
            count = 0
            while pos < len(s) and count < 2:
                if ord(s[pos]) == 0:
                    count = count + 1
                pos = pos + 1
            if count < 2:
                pos = 1

            if pos > 0 and pos < len(s):
                s = s[pos:]
                if ord(s[-1]) == 0:
                    s = s[:-1]

            self.value = s

class Reader:
    """ An ID3 reader.
        Create one on a file object, and then use getValue('TIT2') (for example)
        to pull values.
    """
    def __init__(self, file):
        """ Create a reader from a file or filename. """
        self.file = file
        self.header = None
        self.frames = {}
        self.allFrames = []
        self.bytesLeft = 0
        self.padbytes = ''

        bCloseFile = False
        # If self.file is a string of some sort, then open it to get a file.
        if isinstance(self.file, (type(''), type(u''))):
            self.file = open(self.file, 'rb')
            bCloseFile = True

        self._readId3()

        if bCloseFile:
            self.file.close()

    def _readBytes(self, num, desc=''):
        """ Read some bytes from the file.
            This method implements the "unsynchronization" scheme,
            where 0xFF bytes may have had 0x00 bytes stuffed after
            them.  These zero bytes have to be removed transparently.
        """
        #if _t: _trace("ask %d (%s)" % (num,desc))
        if num > self.bytesLeft:
            #if _c: _coverage('long!')
            raise Id3Error, 'Long read (%s): (%d > %d)' % (desc, num, self.bytesLeft)
        bytes = self.file.read(num)
        self.bytesLeft -= num

        if len(bytes) < num:
            #if _t: _trace("short read with %d left, %d total" % (self.bytesLeft, self.header.size))
            #if _c: _coverage('short!')
            raise Id3Error, 'Short read (%s): (%d < %d)' % (desc, len(bytes), num)

        if self.header.bUnsynchronized:
            nUnsync = 0
            i = 0
            while True:
                i = bytes.find('\xFF\x00', i)
                if i == -1:
                    break
                #if _t: _trace("unsync at %d" % (i+1))
                #if _c: _coverage('unsyncbyte')
                nUnsync += 1
                # This is a stuffed byte to remove
                bytes = bytes[:i+1] + bytes[i+2:]
                # Have to read one more byte from the file to adjust
                bytes += self.file.read(1)
                self.bytesLeft -= 1
                i += 1
            #if _t: _trace("unsync'ed %d" % (nUnsync))

        return bytes

    def _unreadBytes(self, num):
        self.file.seek(-num, 1)
        self.bytesLeft += num

    def _getSyncSafeInt(self, bytes):
        assert len(bytes) == 4
        if type(bytes) == type(''):
            bytes = [ ord(c) for c in bytes ]
        return (bytes[0] << 21) + (bytes[1] << 14) + (bytes[2] << 7) + bytes[3]

    def _getInteger(self, bytes):
        i = 0;
        if type(bytes) == type(''):
            bytes = [ ord(c) for c in bytes ]
        for b in bytes:
            i = i*256+b
        return i

    def _addV1Frame(self, id, rawData):
        if id == 'v1genre':
            assert len(rawData) == 1
            nGenre = ord(rawData)
            try:
                value = genres[nGenre]
            except IndexError:
                if nGenre == 255:
                    value = None
                else:
                    value = "(%d)" % nGenre
        else:
            value = rawData.strip(' \t\r\n').split('\0')[0]
        if value:
            frame = _Frame()
            frame.id = id
            frame.rawData = rawData
            frame.value = value
            self.frames[id] = frame
            self.allFrames.append(frame)

    def _pass(self):
        """ Do nothing, for when we need to plug in a no-op function.
        """
        pass

    def _readId3(self):
        header = self.file.read(10)
        if len(header) < 10:
            return
        hstuff = struct.unpack('!3sBBBBBBB', header)
        if hstuff[0] != "ID3":
            # Doesn't look like an ID3v2 tag,
            # Try reading an ID3v1 tag.
            self._readId3v1()
            return

        self.header = _Header()
        self.header.majorVersion = hstuff[1]
        self.header.revision = hstuff[2]
        self.header.flags = hstuff[3]
        self.header.size = self._getSyncSafeInt(hstuff[4:8])

        self.bytesLeft = self.header.size

        self._readExtHeader = self._pass

        if self.header.majorVersion == 2:
            #if _c: _coverage('id3v2.2.%d' % self.header.revision)
            self._readFrame = self._readFrame_rev2
        elif self.header.majorVersion == 3:
            #if _c: _coverage('id3v2.3.%d' % self.header.revision)
            self._readFrame = self._readFrame_rev3
        elif self.header.majorVersion == 4:
            #if _c: _coverage('id3v2.4.%d' % self.header.revision)
            self._readFrame = self._readFrame_rev4
        else:
            #if _c: _coverage('badmajor!')
            raise Id3Error, "Unsupported major version: %d" % self.header.majorVersion

        # Interpret the flags
        self._interpretFlags()

        # Read any extended header
        self._readExtHeader()

        # Read the frames
        while self.bytesLeft > 0:
            frame = self._readFrame()
            if frame:
                frame._interpret()
                self.frames[frame.id] = frame
                self.allFrames.append(frame)
            else:
                #if _c: _coverage('padding')
                break

    def _interpretFlags(self):
        """ Interpret ID3v2.x flags.
        """
        if self.header.flags & 0x80:
            self.header.bUnsynchronized = True
            #if _c: _coverage('unsynctag')

        if self.header.majorVersion == 2:
            if self.header.flags & 0x40:
                #if _c: _coverage('compressed')
                # "Since no compression scheme has been decided yet,
                # the ID3 decoder (for now) should just ignore the entire
                # tag if the compression bit is set."
                self.header.bCompressed = True

        if self.header.majorVersion >= 3:
            if self.header.flags & 0x40:
                #if _c: _coverage('extheader')
                if self.header.majorVersion == 3:
                    self._readExtHeader = self._readExtHeader_rev3
                else:
                    self._readExtHeader = self._readExtHeader_rev4
            if self.header.flags & 0x20:
                #if _c: _coverage('experimental')
                self.header.bExperimental = True

        if self.header.majorVersion >= 4:
            if self.header.flags & 0x10:
                #if _c: _coverage('footer')
                self.header.bFooter = True

    def _readExtHeader_rev3(self):
        """ Read the ID3v2.3 extended header.
        """
        # We don't interpret this yet, just eat the bytes.
        size = self._getInteger(self._readBytes(4, 'rev3ehlen'))
        self._readBytes(size, 'rev3ehdata')

    def _readExtHeader_rev4(self):
        """ Read the ID3v2.4 extended header.
        """
        # We don't interpret this yet, just eat the bytes.
        size = self._getSyncSafeInt(self._readBytes(4, 'rev4ehlen'))
        self._readBytes(size-4, 'rev4ehdata')

    def _readId3v1(self):
        """ Read the ID3v1 tag.
            spec: http://www.id3.org/id3v1.html
        """
        self.file.seek(-128, 2)
        tag = self.file.read(128)
        if len(tag) != 128:
            return
        if tag[0:3] != 'TAG':
            return
        self.header = _Header()
        self.header.majorVersion = 1
        self.header.revision = 0

        self._addV1Frame('v1title', tag[3:33])
        self._addV1Frame('v1performer', tag[33:63])
        self._addV1Frame('v1album', tag[63:93])
        self._addV1Frame('v1year', tag[93:97])
        self._addV1Frame('v1comment', tag[97:127])
        self._addV1Frame('v1genre', tag[127])
        if tag[125] == '\0' and tag[126] != '\0':
            #if _c: _coverage('id3v1.1')
            self.header.revision = 1
            self._addV1Frame('v1track', str(ord(tag[126])))
        else:
            #if _c: _coverage('id3v1.0')
            pass
        return

    _validIdChars = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789'

    def _isValidId(self, id):
        """ Determine if the id bytes make a valid ID3 id.
        """
        for c in id:
            if not c in self._validIdChars:
                #if _c: _coverage('bad id')
                return False
        #if _c: _coverage('id '+id)
        return True

    def _readFrame_rev2(self):
        """ Read a frame for ID3v2.2: three-byte ids and lengths.
            spec: http://www.id3.org/id3v2-00.txt
        """
        if self.bytesLeft < 6:
            return None
        id = self._readBytes(3, 'rev2id')
        if len(id) < 3 or not self._isValidId(id):
            self._unreadBytes(len(id))
            return None
        hstuff = struct.unpack('!BBB', self._readBytes(3, 'rev2len'))
        frame = _Frame()
        frame.id = id
        frame.size = self._getInteger(hstuff[0:3])
        frame.rawData = self._readBytes(frame.size, 'rev2data')
        return frame

    def _readFrame_rev3(self):
        """ Read a frame for ID3v2.3: four-byte ids and lengths.
        """
        if self.bytesLeft < 10:
            return None
        id = self._readBytes(4,'rev3id')
        if len(id) < 4 or not self._isValidId(id):
            self._unreadBytes(len(id))
            return None
        hstuff = struct.unpack('!BBBBh', self._readBytes(6,'rev3head'))
        frame = _Frame()
        frame.id = id
        frame.size = self._getInteger(hstuff[0:4])
        cbData = frame.size
        frame.flags = hstuff[4]
        #if _t: _trace('flags = %x' % frame.flags)
        frame.bTagAlterPreserve = (frame.flags & 0x8000 != 0)
        frame.bFileAlterPreserve = (frame.flags & 0x4000 != 0)
        frame.bReadOnly = (frame.flags & 0x2000 != 0)
        frame.bCompressed = (frame.flags & 0x0080 != 0)
        if frame.bCompressed:
            frame.decompressedSize = self._getInteger(self._readBytes(4, 'decompsize'))
            cbData -= 4
            #if _c: _coverage('compress')
        frame.bEncrypted = (frame.flags & 0x0040 != 0)
        if frame.bEncrypted:
            frame.encryptionMethod = self._readBytes(1, 'encrmethod')
            cbData -= 1
            #if _c: _coverage('encrypt')
        frame.bInGroup = (frame.flags & 0x0020 != 0)
        if frame.bInGroup:
            frame.groupid = self._readBytes(1, 'groupid')
            cbData -= 1
            #if _c: _coverage('groupid')

        frame.rawData = self._readBytes(cbData, 'rev3data')
        return frame

    def _readFrame_rev4(self):
        """ Read a frame for ID3v2.4: four-byte ids and lengths.
        """
        if self.bytesLeft < 10:
            return None
        id = self._readBytes(4,'rev4id')
        if len(id) < 4 or not self._isValidId(id):
            self._unreadBytes(len(id))
            return None
        hstuff = struct.unpack('!BBBBh', self._readBytes(6,'rev4head'))
        frame = _Frame()
        frame.id = id
        frame.size = self._getSyncSafeInt(hstuff[0:4])
        cbData = frame.size
        frame.flags = hstuff[4]
        frame.bTagAlterPreserve = (frame.flags & 0x4000 != 0)
        frame.bFileAlterPreserve = (frame.flags & 0x2000 != 0)
        frame.bReadOnly = (frame.flags & 0x1000 != 0)
        frame.bInGroup = (frame.flags & 0x0040 != 0)
        if frame.bInGroup:
            frame.groupid = self._readBytes(1, 'groupid')
            cbData -= 1
            #if _c: _coverage('groupid')

        frame.bCompressed = (frame.flags & 0x0008 != 0)
        if frame.bCompressed:
            #if _c: _coverage('compress')
            pass
        frame.bEncrypted = (frame.flags & 0x0004 != 0)
        if frame.bEncrypted:
            frame.encryptionMethod = self._readBytes(1, 'encrmethod')
            cbData -= 1
            #if _c: _coverage('encrypt')
        frame.bUnsynchronized = (frame.flags & 0x0002 != 0)
        if frame.bUnsynchronized:
            #if _c: _coverage('unsyncframe')
            pass
        if frame.flags & 0x0001:
            frame.datalen = self._getSyncSafeInt(self._readBytes(4, 'datalen'))
            cbData -= 4
            #if _c: _coverage('datalenindic')

        frame.rawData = self._readBytes(cbData, 'rev3data')

        return frame

    def getValue(self, id):
        """ Return the value for an ID3 tag id, or for a
            convenience label ('title', 'artist', ...),
            or return None if there is no such value.
        """
        if self.frames.has_key(id):
            if hasattr(self.frames[id], 'value'):
                return self.frames[id].value
        if _simpleDataMapping.has_key(id):
            for id2 in _simpleDataMapping[id]:
                v = self.getValue(id2)
                if v:
                    return v
        return None

    def getRawData(self, id):
        if self.frames.has_key(id):
            return self.frames[id].rawData
        return None

    def dump(self):
        import pprint
        print "Header:"
        print self.header
        print "Frames:"
        for fr in self.allFrames:
            if len(fr.rawData) > 30:
                fr.rawData = fr.rawData[:30]
        pprint.pprint(self.allFrames)
        for fr in self.allFrames:
            if hasattr(fr, 'value'):
                print '%s: %s' % (fr.id, _safestr(fr.value))
            else:
                print '%s= %s' % (fr.id, _safestr(fr.rawData))
        for label in _simpleDataMapping.keys():
            v = self.getValue(label)
            if v:
                print 'Label %s: %s' % (label, _safestr(v))

    def dumpCoverage(self):
        feats = _features.keys()
        feats.sort()
        for feat in feats:
            print "Feature %-12s: %d" % (feat, _features[feat])

if __name__ == '__main__':
    if len(sys.argv) < 2 or '-?' in sys.argv:
        print "Give me a filename"
    else:
        id3 = Reader(sys.argv[1])
        id3.dump()
        #if _c: id3.dumpCoverage()

# History:
# 20040104: Created.
# 20040105: Two bugs: didn't read v1 properly, and didn't like empty strings in values.
#
# 20040109: Properly reads v2.3 properly (4-byte lens, but not synchsafe)
#           Handles unsynchronized tags properly.
#
# 20040110: Total length was wrong for unsynchronized tags.
#           Treat input filename better so path module can be used.
#           Frame ids are more closely scrutinized for validity.
#           Errors are now thrown as our own exception.
#           Pad bytes aren't retained any more.
#           Frame.value is not set if there is no interpretation performed.
#
# 20040111: Tracing and code coverage more formalized.
#           Exceptions are now all Id3Error.
#           Zero-length data in frames is handled pleasantly.
#           Compressed frames are decompressed.
#           Extended headers are read (but uninterpreted).
#           Non-zero pad bytes are handled.
#           Frame flags are read and interpreted.
#           W*** frames are interpreted.
#           Multi-string frames set .value to a list of strings.
#
# 20040113: Strip all trailing zero bytes from text strings.
#           If we opened the file, we should close the file.
#
# 20040205: Do a better job printing strings without throwing.
#           Support genre information, even if it is stupid.
#
# 20040913: When dumping strings, be more robust when trying to print
#               non-character data. Thanks to Duane Harkness for the fix.
#
# 20061230: Fix ommission of self. in a few places.
#
# 20070415: Extended headers in ID3v2.4 weren't skipped properly, throwing
#               everything out of whack.
#           Be more generous about finding album and artist names in the tag.