TOOLS/matroska.py

   1 #!/usr/bin/python3
   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                 'MaxCache, 6df8, uint',
  84                 'DefaultDuration, 23e383, uint',
  85                 'TrackTimecodeScale, 23314f, float',
  86                 'MaxBlockAdditionID, 55ee, uint',
  87                 'Name, 536e, str',
  88                 'Language, 22b59c, str',
  89                 'CodecID, 86, str',
  90                 'CodecPrivate, 63a2, binary',
  91                 'CodecDecodeAll, aa, uint',
  92                 'Video, e0, sub', (
  93                     'FlagInterlaced, 9a, uint',
  94                     'PixelWidth, b0, uint',
  95                     'PixelHeight, ba, uint',
  96                     'DisplayWidth, 54b0, uint',
  97                     'DisplayHeight, 54ba, uint',
  98                     'FrameRate, 2383e3, float',
  99                 ),
 100                 'Audio, e1, sub', (
 101                     'SamplingFrequency, b5, float',
 102                     'Channels, 9f, uint',
 103                     'BitDepth, 6264, uint',
 104                 ),
 105                 'ContentEncodings, 6d80, sub', (
 106                     'ContentEncoding*, 6240, sub', (
 107                         'ContentEncodingOrder, 5031, uint',
 108                         'ContentEncodingScope, 5032, uint',
 109                         'ContentEncodingType, 5033, uint',
 110                         'ContentCompression, 5034, sub', (
 111                             'ContentCompAlgo, 4254, uint',
 112                             'ContentCompSettings, 4255, binary',
 113                         ),
 114                     ),
 115                 ),
 116             ),
 117         ),
 118
 119         'Cues, 1c53bb6b, sub', (
 120             'CuePoint*, bb, sub', (
 121                 'CueTime, b3, uint',
 122                 'CueTrackPositions*, b7, sub', (
 123                     'CueTrack, f7, uint',
 124                     'CueClusterPosition, f1, uint',
 125                 ),
 126             ),
 127         ),
 128
 129         'Attachments, 1941a469, sub', (
 130             'AttachedFile*, 61a7, sub', (
 131                 'FileName, 466e, str',
 132                 'FileMimeType, 4660, str',
 133                 'FileData, 465c, binary',
 134                 'FileUID, 46ae, uint',
 135             ),
 136         ),
 137
 138         'Chapters, 1043a770, sub', (
 139             'EditionEntry*, 45b9, sub', (
 140                 'EditionUID, 45bc, uint',
 141                 'EditionFlagHidden, 45bd, uint',
 142                 'EditionFlagDefault, 45db, uint',
 143                 'EditionFlagOrdered, 45dd, uint',
 144                 'ChapterAtom*, b6, sub', (
 145                     'ChapterUID, 73c4, uint',
 146                     'ChapterTimeStart, 91, uint',
 147                     'ChapterTimeEnd, 92, uint',
 148                     'ChapterFlagHidden, 98, uint',
 149                     'ChapterFlagEnabled, 4598, uint',
 150                     'ChapterSegmentUID, 6e67, binary',
 151                     'ChapterSegmentEditionUID, 6ebc, uint',
 152                     'ChapterDisplay*, 80, sub', (
 153                         'ChapString, 85, str',
 154                         'ChapLanguage*, 437c, str',
 155                         'ChapCountry*, 437e, str',
 156                     ),
 157                 ),
 158             ),
 159         ),
 160         'Tags*, 1254c367, sub', (
 161             'Tag*, 7373, sub', (
 162                 'Targets, 63c0, sub', (
 163                     'TargetTypeValue, 68ca, uint',
 164                     'TargetTrackUID, 63c5, uint',
 165                     'TargetEditionUID, 63c9, uint',
 166                     'TargetChapterUID, 63c4, uint',
 167                     'TargetAttachmentUID, 63c6, uint',
 168                  ),
 169                 'SimpleTag*, 67c8, sub', (
 170                     'TagName, 45a3, str',
 171                     'TagLanguage, 447a, str',
 172                     'TagString, 4487, str'
 173                 ),
 174             ),
 175         ),
 176     ),
 177 )
 178
 179
 180 import sys
 181 from math import ldexp
 182 from binascii import hexlify
 183
 184 def byte2num(s):
 185     return int(hexlify(s), 16)
 186
 187 class EOF(Exception): pass
 188
 189 def camelcase_to_words(name):
 190     parts = []
 191     start = 0
 192     for i in range(1, len(name)):
 193         if name[i].isupper() and (name[i-1].islower() or
 194                                   name[i+1:i+2].islower()):
 195             parts.append(name[start:i])
 196             start = i
 197     parts.append(name[start:])
 198     return '_'.join(parts).lower()
 199
 200 class MatroskaElement(object):
 201
 202     def __init__(self, name, elid, valtype, namespace):
 203         self.name = name
 204         self.definename = '{}_ID_{}'.format(namespace, name.upper())
 205         self.fieldname = camelcase_to_words(name)
 206         self.structname = 'ebml_' + self.fieldname
 207         self.elid = elid
 208         self.valtype = valtype
 209         if valtype == 'sub':
 210             self.ebmltype = 'EBML_TYPE_SUBELEMENTS'
 211             self.valname = 'struct ' + self.structname
 212         else:
 213             self.ebmltype = 'EBML_TYPE_' + valtype.upper()
 214             try:
 215                 self.valname = {'uint': 'uint64_t', 'str': 'struct bstr',
 216                                 'binary': 'struct bstr', 'ebml_id': 'uint32_t',
 217                                 'float': 'double', 'sint': 'int64_t',
 218                                 }[valtype]
 219             except KeyError:
 220                 raise SyntaxError('Unrecognized value type ' + valtype)
 221         self.subelements = ()
 222
 223     def add_subelements(self, subelements):
 224         self.subelements = subelements
 225         self.subids = set(x[0].elid for x in subelements)
 226
 227 elementd = {}
 228 elementlist = []
 229 def parse_elems(l, namespace):
 230     subelements = []
 231     for el in l:
 232         if isinstance(el, str):
 233             name, hexid, eltype = [x.strip() for x in el.split(',')]
 234             multiple = name.endswith('*')
 235             name = name.strip('*')
 236             new = MatroskaElement(name, hexid, eltype, namespace)
 237             elementd[hexid] = new
 238             elementlist.append(new)
 239             subelements.append((new, multiple))
 240         else:
 241             new.add_subelements(parse_elems(el, namespace))
 242     return subelements
 243
 244 parse_elems(elements_ebml, 'EBML')
 245 parse_elems(elements_matroska, 'MATROSKA')
 246
 247 def generate_C_header():
 248     print('// Generated by TOOLS/matroska.py, do not edit manually')
 249     print()
 250
 251     for el in elementlist:
 252         print('#define {0.definename:40} 0x{0.elid}'.format(el))
 253
 254     print()
 255
 256     for el in reversed(elementlist):
 257         if not el.subelements:
 258             continue
 259         print()
 260         print('struct {0.structname} {{'.format(el))
 261         l = max(len(subel.valname) for subel, multiple in el.subelements)+1
 262         for subel, multiple in el.subelements:
 263             print('    {e.valname:{l}} {star}{e.fieldname};'.format(
 264                     e=subel, l=l, star=' *'[multiple]))
 265         print()
 266         for subel, multiple in el.subelements:
 267             print('    int  n_{0.fieldname};'.format(subel))
 268         print('};')
 269
 270     for el in elementlist:
 271         if not el.subelements:
 272             continue
 273         print('extern const struct ebml_elem_desc {0.structname}_desc;'.format(
 274                 el))
 275
 276     print()
 277     print('#define MAX_EBML_SUBELEMENTS', max(len(el.subelements)
 278                                               for el in elementlist))
 279
 280
 281
 282 def generate_C_definitions():
 283     print('// Generated by TOOLS/matroska.py, do not edit manually')
 284     print()
 285     for el in reversed(elementlist):
 286         print()
 287         if el.subelements:
 288             print('#define N', el.fieldname)
 289             print('E_S("{}", {})'.format(el.name, len(el.subelements)))
 290             for subel, multiple in el.subelements:
 291                 print('F({0.definename}, {0.fieldname}, {1})'.format(
 292                         subel, int(multiple)))
 293             print('}};')
 294             print('#undef N')
 295         else:
 296             print('E("{0.name}", {0.fieldname}, {0.ebmltype})'.format(el))
 297
 298 def read(s, length):
 299     t = s.read(length)
 300     if len(t) != length:
 301         raise EOF
 302     return t
 303
 304 def read_id(s):
 305     t = read(s, 1)
 306     i = 0
 307     mask = 128
 308     if ord(t) == 0:
 309         raise SyntaxError
 310     while not ord(t) & mask:
 311         i += 1
 312         mask >>= 1
 313     t += read(s, i)
 314     return t
 315
 316 def read_vint(s):
 317     t = read(s, 1)
 318     i = 0
 319     mask = 128
 320     if ord(t) == 0:
 321         raise SyntaxError
 322     while not ord(t) & mask:
 323         i += 1
 324         mask >>= 1
 325     t = bytes((ord(t) & (mask - 1),))
 326     t += read(s, i)
 327     return i+1, byte2num(t)
 328
 329 def read_str(s, length):
 330     return read(s, length)
 331
 332 def read_uint(s, length):
 333     t = read(s, length)
 334     return byte2num(t)
 335
 336 def read_sint(s, length):
 337     i = read_uint(s, length)
 338     mask = 1 << (length * 8 - 1)
 339     if i & mask:
 340         i -= 2 * mask
 341     return i
 342
 343 def read_float(s, length):
 344     t = read(s, length)
 345     i = byte2num(t)
 346     if length == 4:
 347         f = ldexp((i & 0x7fffff) + (1 << 23), (i >> 23 & 0xff) - 150)
 348         if i & (1 << 31):
 349             f = -f
 350     elif length == 8:
 351         f = ldexp((i & ((1 << 52) - 1)) + (1 << 52), (i >> 52 & 0x7ff) - 1075)
 352         if i & (1 << 63):
 353             f = -f
 354     else:
 355         raise SyntaxError
 356     return f
 357
 358 def parse_one(s, depth, parent, maxlen):
 359     elid = hexlify(read_id(s)).decode('ascii')
 360     elem = elementd.get(elid)
 361     if parent is not None and elid not in parent.subids and elid not in ('ec', 'bf'):
 362         print('Unexpected:', elid)
 363         if 1:
 364             raise NotImplementedError
 365     size, length = read_vint(s)
 366     this_length = len(elid) / 2 + size + length
 367     if elem is not None:
 368         if elem.valtype != 'skip':
 369             print(depth, elid, elem.name, 'size:', length, 'value:', end=' ')
 370         if elem.valtype == 'sub':
 371             print('subelements:')
 372             while length > 0:
 373                 length -= parse_one(s, depth + 1, elem, length)
 374             if length < 0:
 375                 raise SyntaxError
 376         elif elem.valtype == 'str':
 377             print('string', repr(read_str(s, length).decode('utf8', 'replace')))
 378         elif elem.valtype in ('binary', 'ebml_id'):
 379             t = read_str(s, length)
 380             dec = ''
 381             if elem.valtype == 'ebml_id':
 382                 idelem = elementd.get(hexlify(t).decode('ascii'))
 383                 if idelem is None:
 384                     dec = '(UNKNOWN)'
 385                 else:
 386                     dec = '({0.name})'.format(idelem)
 387             if len(t) < 20:
 388                 t = hexlify(t).decode('ascii')
 389             else:
 390                 t = '<skipped {} bytes>'.format(len(t))
 391             print('binary', t, dec)
 392         elif elem.valtype == 'uint':
 393             print('uint', read_uint(s, length))
 394         elif elem.valtype == 'sint':
 395             print('sint', read_sint(s, length))
 396         elif elem.valtype == 'float':
 397             print('float', read_float(s, length))
 398         elif elem.valtype == 'skip':
 399             read(s, length)
 400         else:
 401             raise NotImplementedError
 402     else:
 403         print(depth, 'Unknown element:', elid, 'size:', length)
 404         read(s, length)
 405     return this_length
 406
 407 def parse_toplevel(s):
 408     parse_one(s, 0, None, 1 << 63)
 409
 410 if sys.argv[1] == '--generate-header':
 411     generate_C_header()
 412 elif sys.argv[1] == '--generate-definitions':
 413     generate_C_definitions()
 414 else:
 415     s = open(sys.argv[1], "rb")
 416     while 1:
 417         start = s.tell()
 418         try:
 419             parse_toplevel(s)
 420         except EOF:
 421             if s.tell() != start:
 422                 raise Exception("Unexpected end of file")
 423             break