TOOLS/matroska.py

   1 #!/usr/bin/python
   2 """
   3 Generate C definitions for parsing Matroska files.
   4 Can also be used to directly parse Matroska files and display their contents.
   5 """
   6
   7 #
   8 # This file is part of MPlayer.
   9 #
  10 # MPlayer is free software; you can redistribute it and/or modify
  11 # it under the terms of the GNU General Public License as published by
  12 # the Free Software Foundation; either version 2 of the License, or
  13 # (at your option) any later version.
  14 #
  15 # MPlayer is distributed in the hope that it will be useful,
  16 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  18 # GNU General Public License for more details.
  19 #
  20 # You should have received a copy of the GNU General Public License along
  21 # with MPlayer; if not, write to the Free Software Foundation, Inc.,
  22 # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  23 #
  24
  25
  26 elements_ebml = (
  27     'EBML, 1a45dfa3, sub', (
  28         'EBMLVersion, 4286, uint',
  29         'EBMLReadVersion, 42f7, uint',
  30         'EBMLMaxIDLength, 42f2, uint',
  31         'EBMLMaxSizeLength, 42f3, uint',
  32         'DocType, 4282, str',
  33         'DocTypeVersion, 4287, uint',
  34         'DocTypeReadVersion, 4285, uint',
  35     ),
  36
  37     'CRC32, bf, binary',
  38     'Void, ec, binary',
  39 )
  40
  41 elements_matroska = (
  42     'Segment, 18538067, sub', (
  43
  44         'SeekHead*, 114d9b74, sub', (
  45             'Seek*, 4dbb, sub', (
  46                 'SeekID, 53ab, ebml_id',
  47                 'SeekPosition, 53ac, uint',
  48             ),
  49         ),
  50
  51         'Info*, 1549a966, sub', (
  52             'SegmentUID, 73a4, binary',
  53             'PrevUID, 3cb923, binary',
  54             'NextUID, 3eb923, binary',
  55             'TimecodeScale, 2ad7b1, uint',
  56             'DateUTC, 4461, sint',
  57             'Title, 7ba9, str',
  58             'MuxingApp, 4d80, str',
  59             'WritingApp, 5741, str',
  60             'Duration, 4489, float',
  61         ),
  62
  63         'Cluster*, 1f43b675, sub', (
  64             'Timecode, e7, uint',
  65             'BlockGroup*, a0, sub', (
  66                 'Block, a1, binary',
  67                 'BlockDuration, 9b, uint',
  68                 'ReferenceBlock*, fb, sint',
  69             ),
  70             'SimpleBlock*, a3, binary',
  71         ),
  72
  73         'Tracks*, 1654ae6b, sub', (
  74             'TrackEntry*, ae, sub', (
  75                 'TrackNumber, d7, uint',
  76                 'TrackUID, 73c5, uint',
  77                 'TrackType, 83, uint',
  78                 'FlagEnabled, b9, uint',
  79                 'FlagDefault, 88, uint',
  80                 'FlagForced, 55aa, uint',
  81                 'FlagLacing, 9c, uint',
  82                 'MinCache, 6de7, uint',
  83                 'DefaultDuration, 23e383, uint',
  84                 'TrackTimecodeScale, 23314f, float',
  85                 'MaxBlockAdditionID, 55ee, uint',
  86                 'Name, 536e, str',
  87                 'Language, 22b59c, str',
  88                 'CodecID, 86, str',
  89                 'CodecPrivate, 63a2, binary',
  90                 'CodecDecodeAll, aa, uint',
  91                 'Video, e0, sub', (
  92                     'FlagInterlaced, 9a, uint',
  93                     'PixelWidth, b0, uint',
  94                     'PixelHeight, ba, uint',
  95                     'DisplayWidth, 54b0, uint',
  96                     'DisplayHeight, 54ba, uint',
  97                     'FrameRate, 2383e3, float',
  98                 ),
  99                 'Audio, e1, sub', (
 100                     'SamplingFrequency, b5, float',
 101                     'Channels, 9f, uint',
 102                     'BitDepth, 6264, uint',
 103                 ),
 104                 'ContentEncodings, 6d80, sub', (
 105                     'ContentEncoding*, 6240, sub', (
 106                         'ContentEncodingOrder, 5031, uint',
 107                         'ContentEncodingScope, 5032, uint',
 108                         'ContentEncodingType, 5033, uint',
 109                         'ContentCompression, 5034, sub', (
 110                             'ContentCompAlgo, 4254, uint',
 111                             'ContentCompSettings, 4255, binary',
 112                         ),
 113                     ),
 114                 ),
 115             ),
 116         ),
 117
 118         'Cues, 1c53bb6b, sub', (
 119             'CuePoint*, bb, sub', (
 120                 'CueTime, b3, uint',
 121                 'CueTrackPositions*, b7, sub', (
 122                     'CueTrack, f7, uint',
 123                     'CueClusterPosition, f1, uint',
 124                 ),
 125             ),
 126         ),
 127
 128         'Attachments, 1941a469, sub', (
 129             'AttachedFile*, 61a7, sub', (
 130                 'FileName, 466e, str',
 131                 'FileMimeType, 4660, str',
 132                 'FileData, 465c, binary',
 133                 'FileUID, 46ae, uint',
 134             ),
 135         ),
 136
 137         'Chapters, 1043a770, sub', (
 138             'EditionEntry*, 45b9, sub', (
 139                 'EditionUID, 45bc, uint',
 140                 'EditionFlagHidden, 45bd, uint',
 141                 'EditionFlagDefault, 45db, uint',
 142                 'EditionFlagOrdered, 45dd, uint',
 143                 'ChapterAtom*, b6, sub', (
 144                     'ChapterUID, 73c4, uint',
 145                     'ChapterTimeStart, 91, uint',
 146                     'ChapterTimeEnd, 92, uint',
 147                     'ChapterFlagHidden, 98, uint',
 148                     'ChapterFlagEnabled, 4598, uint',
 149                     'ChapterSegmentUID, 6e67, binary',
 150                     'ChapterSegmentEditionUID, 6ebc, uint',
 151                     'ChapterDisplay*, 80, sub', (
 152                         'ChapString, 85, str',
 153                         'ChapLanguage*, 437c, str',
 154                     ),
 155                 ),
 156             ),
 157         ),
 158         'Tags*, 1254c367, sub', (
 159             'Tag*, 7373, sub', (
 160                 'Targets, 63c0, sub', (
 161                     'TargetTypeValue, 68ca, uint',
 162                     'TargetTrackUID, 63c5, uint',
 163                     'TargetEditionUID, 63c9, uint',
 164                     'TargetChapterUID, 63c4, uint',
 165                     'TargetAttachmentUID, 63c6, uint',
 166                  ),
 167                 'SimpleTag*, 67c8, sub', (
 168                     'TagName, 45a3, str',
 169                     'TagLanguage, 447a, str',
 170                     'TagString, 4487, str'
 171                 ),
 172             ),
 173         ),
 174     ),
 175 )
 176
 177
 178 import sys
 179 from math import ldexp
 180
 181 def byte2num(s):
 182     return int(s.encode('hex'), 16)
 183
 184 def camelcase_to_words(name):
 185     parts = []
 186     start = 0
 187     for i in range(1, len(name)):
 188         if name[i].isupper() and (name[i-1].islower() or
 189                                   name[i+1:i+2].islower()):
 190             parts.append(name[start:i])
 191             start = i
 192     parts.append(name[start:])
 193     return '_'.join(parts).lower()
 194
 195 class MatroskaElement(object):
 196
 197     def __init__(self, name, elid, valtype, namespace):
 198         self.name = name
 199         self.definename = '%s_ID_%s' % (namespace, name.upper())
 200         self.fieldname = camelcase_to_words(name)
 201         self.structname = 'ebml_' + self.fieldname
 202         self.elid = elid
 203         self.valtype = valtype
 204         if valtype == 'sub':
 205             self.ebmltype = 'EBML_TYPE_SUBELEMENTS'
 206             self.valname = 'struct %s' % self.structname
 207         else:
 208             self.ebmltype = 'EBML_TYPE_' + valtype.upper()
 209             try:
 210                 self.valname = {'uint': 'uint64_t', 'str': 'struct bstr',
 211                                 'binary': 'struct bstr', 'ebml_id': 'uint32_t',
 212                                 'float': 'double', 'sint': 'int64_t',
 213                                 }[valtype]
 214             except KeyError:
 215                 raise SyntaxError('Unrecognized value type ' + valtype)
 216         self.subelements = ()
 217
 218     def add_subelements(self, subelements):
 219         self.subelements = subelements
 220         self.subids = set(x[0].elid for x in subelements)
 221
 222 elementd = {}
 223 elementlist = []
 224 def parse_elems(l, namespace):
 225     subelements = []
 226     for el in l:
 227         if isinstance(el, str):
 228             name, hexid, eltype = [x.strip() for x in el.split(',')]
 229             multiple = name.endswith('*')
 230             name = name.strip('*')
 231             new = MatroskaElement(name, hexid, eltype, namespace)
 232             elementd[hexid] = new
 233             elementlist.append(new)
 234             subelements.append((new, multiple))
 235         else:
 236             new.add_subelements(parse_elems(el, namespace))
 237     return subelements
 238
 239 parse_elems(elements_ebml, 'EBML')
 240 parse_elems(elements_matroska, 'MATROSKA')
 241
 242 def generate_C_header():
 243     print('// Generated by TOOLS/matroska.py, do not edit manually')
 244     print
 245
 246     for el in elementlist:
 247         print('#define %-40s 0x%s' % (el.definename, el.elid))
 248
 249     print
 250
 251     for el in reversed(elementlist):
 252         if not el.subelements:
 253             continue
 254         print
 255         print('struct %s {' % el.structname)
 256         l = max(len(subel.valname) for subel, multiple in el.subelements)+1
 257         for subel, multiple in el.subelements:
 258             print('    %-*s %s%s;' % (l, subel.valname, (' ', '*')[multiple],
 259                                     subel.fieldname))
 260         print
 261         for subel, multiple in el.subelements:
 262             print('    int  n_%s;' % (subel.fieldname))
 263         print('};')
 264
 265     for el in elementlist:
 266         if not el.subelements:
 267             continue
 268         print('extern const struct ebml_elem_desc %s_desc;' % el.structname)
 269
 270     print
 271     print('#define MAX_EBML_SUBELEMENTS %d' % max(len(el.subelements)
 272                                                   for el in elementlist))
 273
 274
 275
 276 def generate_C_definitions():
 277     print('// Generated by TOOLS/matroska.py, do not edit manually')
 278     print
 279     for el in reversed(elementlist):
 280         print
 281         if el.subelements:
 282             print('#define N %s' % el.fieldname)
 283             print('E_S("%s", %d)' % (el.name, len(el.subelements)))
 284             for subel, multiple in el.subelements:
 285                 print('F(%s, %s, %d)' % (subel.definename, subel.fieldname,
 286                                          multiple))
 287             print('}};')
 288             print('#undef N')
 289         else:
 290             print('E("%s", %s, %s)' % (el.name, el.fieldname, el.ebmltype))
 291
 292 def read(s, length):
 293     t = s.read(length)
 294     if len(t) != length:
 295         raise IOError
 296     return t
 297
 298 def read_id(s):
 299     t = read(s, 1)
 300     i = 0
 301     mask = 128
 302     if ord(t) == 0:
 303         raise SyntaxError
 304     while not ord(t) & mask:
 305         i += 1
 306         mask >>= 1
 307     t += read(s, i)
 308     return t
 309
 310 def read_vint(s):
 311     t = read(s, 1)
 312     i = 0
 313     mask = 128
 314     if ord(t) == 0:
 315         raise SyntaxError
 316     while not ord(t) & mask:
 317         i += 1
 318         mask >>= 1
 319     t = chr(ord(t) & (mask - 1))
 320     t += read(s, i)
 321     return i+1, byte2num(t)
 322
 323 def read_str(s, length):
 324     return read(s, length)
 325
 326 def read_uint(s, length):
 327     t = read(s, length)
 328     return byte2num(t)
 329
 330 def read_sint(s, length):
 331     i = read_uint(s, length)
 332     mask = 1 << (length * 8 - 1)
 333     if i & mask:
 334         i -= 2 * mask
 335     return i
 336
 337 def read_float(s, length):
 338     t = read(s, length)
 339     i = byte2num(t)
 340     if length == 4:
 341         f = ldexp((i & 0x7fffff) + (1 << 23), (i >> 23 & 0xff) - 150)
 342         if i & (1 << 31):
 343             f = -f
 344     elif length == 8:
 345         f = ldexp((i & ((1 << 52) - 1)) + (1 << 52), (i >> 52 & 0x7ff) - 1075)
 346         if i & (1 << 63):
 347             f = -f
 348     else:
 349         raise SyntaxError
 350     return f
 351
 352 def parse_one(s, depth, parent, maxlen):
 353     elid = read_id(s).encode('hex')
 354     elem = elementd.get(elid)
 355     if parent is not None and elid not in parent.subids and elid not in ('ec', 'bf'):
 356         print('Unexpected:', elid)
 357         if 1:
 358             raise NotImplementedError
 359     size, length = read_vint(s)
 360     this_length = len(elid) / 2 + size + length
 361     if elem is not None:
 362         if elem.valtype != 'skip':
 363             print depth, elid, elem.name, 'size:', length, 'value:',
 364         if elem.valtype == 'sub':
 365             print('subelements:')
 366             while length > 0:
 367                 length -= parse_one(s, depth + 1, elem, length)
 368             if length < 0:
 369                 raise SyntaxError
 370         elif elem.valtype == 'str':
 371             print 'string', repr(read_str(s, length))
 372         elif elem.valtype in ('binary', 'ebml_id'):
 373             t = read_str(s, length)
 374             dec = ''
 375             if elem.valtype == 'ebml_id':
 376                 idelem = elementd.get(t.encode('hex'))
 377                 if idelem is None:
 378                     dec = '(UNKNOWN)'
 379                 else:
 380                     dec = '(%s)' % idelem.name
 381             if len(t) < 20:
 382                 t = t.encode('hex')
 383             else:
 384                 t = '<skipped %d bytes>' % len(t)
 385             print 'binary', t, dec
 386         elif elem.valtype == 'uint':
 387             print 'uint', read_uint(s, length)
 388         elif elem.valtype == 'sint':
 389             print 'sint', read_sint(s, length)
 390         elif elem.valtype == 'float':
 391             print 'float', read_float(s, length)
 392         elif elem.valtype == 'skip':
 393             read(s, length)
 394         else:
 395             raise NotImplementedError
 396     else:
 397         print(depth, 'Unknown element:', elid, 'size:', length)
 398         read(s, length)
 399     return this_length
 400
 401 def parse_toplevel(s):
 402     parse_one(s, 0, None, 1 << 63)
 403
 404 if sys.argv[1] == '--generate-header':
 405     generate_C_header()
 406 elif sys.argv[1] == '--generate-definitions':
 407     generate_C_definitions()
 408 else:
 409     s = open(sys.argv[1])
 410     while 1:
 411         parse_toplevel(s)