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                     'DisplayUnit, 54b2, uint',
  99                     'FrameRate, 2383e3, float',
 100                 ),
 101                 'Audio, e1, sub', (
 102                     'SamplingFrequency, b5, float',
 103                     'OutputSamplingFrequency, 78b5, float',
 104                     'Channels, 9f, uint',
 105                     'BitDepth, 6264, uint',
 106                 ),
 107                 'ContentEncodings, 6d80, sub', (
 108                     'ContentEncoding*, 6240, sub', (
 109                         'ContentEncodingOrder, 5031, uint',
 110                         'ContentEncodingScope, 5032, uint',
 111                         'ContentEncodingType, 5033, uint',
 112                         'ContentCompression, 5034, sub', (
 113                             'ContentCompAlgo, 4254, uint',
 114                             'ContentCompSettings, 4255, binary',
 115                         ),
 116                     ),
 117                 ),
 118             ),
 119         ),
 120
 121         'Cues, 1c53bb6b, sub', (
 122             'CuePoint*, bb, sub', (
 123                 'CueTime, b3, uint',
 124                 'CueTrackPositions*, b7, sub', (
 125                     'CueTrack, f7, uint',
 126                     'CueClusterPosition, f1, uint',
 127                 ),
 128             ),
 129         ),
 130
 131         'Attachments, 1941a469, sub', (
 132             'AttachedFile*, 61a7, sub', (
 133                 'FileDescription, 467e, str',
 134                 'FileName, 466e, str',
 135                 'FileMimeType, 4660, str',
 136                 'FileData, 465c, binary',
 137                 'FileUID, 46ae, uint',
 138             ),
 139         ),
 140
 141         'Chapters, 1043a770, sub', (
 142             'EditionEntry*, 45b9, sub', (
 143                 'EditionUID, 45bc, uint',
 144                 'EditionFlagHidden, 45bd, uint',
 145                 'EditionFlagDefault, 45db, uint',
 146                 'EditionFlagOrdered, 45dd, uint',
 147                 'ChapterAtom*, b6, sub', (
 148                     'ChapterUID, 73c4, uint',
 149                     'ChapterTimeStart, 91, uint',
 150                     'ChapterTimeEnd, 92, uint',
 151                     'ChapterFlagHidden, 98, uint',
 152                     'ChapterFlagEnabled, 4598, uint',
 153                     'ChapterSegmentUID, 6e67, binary',
 154                     'ChapterSegmentEditionUID, 6ebc, uint',
 155                     'ChapterDisplay*, 80, sub', (
 156                         'ChapString, 85, str',
 157                         'ChapLanguage*, 437c, str',
 158                         'ChapCountry*, 437e, str',
 159                     ),
 160                 ),
 161             ),
 162         ),
 163         'Tags*, 1254c367, sub', (
 164             'Tag*, 7373, sub', (
 165                 'Targets, 63c0, sub', (
 166                     'TargetTypeValue, 68ca, uint',
 167                     'TargetTrackUID, 63c5, uint',
 168                     'TargetEditionUID, 63c9, uint',
 169                     'TargetChapterUID, 63c4, uint',
 170                     'TargetAttachmentUID, 63c6, uint',
 171                  ),
 172                 'SimpleTag*, 67c8, sub', (
 173                     'TagName, 45a3, str',
 174                     'TagLanguage, 447a, str',
 175                     'TagString, 4487, str'
 176                 ),
 177             ),
 178         ),
 179     ),
 180 )
 181
 182
 183 import sys
 184 from math import ldexp
 185 from binascii import hexlify
 186
 187 def byte2num(s):
 188     return int(hexlify(s), 16)
 189
 190 class EOF(Exception): pass
 191
 192 def camelcase_to_words(name):
 193     parts = []
 194     start = 0
 195     for i in range(1, len(name)):
 196         if name[i].isupper() and (name[i-1].islower() or
 197                                   name[i+1:i+2].islower()):
 198             parts.append(name[start:i])
 199             start = i
 200     parts.append(name[start:])
 201     return '_'.join(parts).lower()
 202
 203 class MatroskaElement(object):
 204
 205     def __init__(self, name, elid, valtype, namespace):
 206         self.name = name
 207         self.definename = '{}_ID_{}'.format(namespace, name.upper())
 208         self.fieldname = camelcase_to_words(name)
 209         self.structname = 'ebml_' + self.fieldname
 210         self.elid = elid
 211         self.valtype = valtype
 212         if valtype == 'sub':
 213             self.ebmltype = 'EBML_TYPE_SUBELEMENTS'
 214             self.valname = 'struct ' + self.structname
 215         else:
 216             self.ebmltype = 'EBML_TYPE_' + valtype.upper()
 217             try:
 218                 self.valname = {'uint': 'uint64_t', 'str': 'struct bstr',
 219                                 'binary': 'struct bstr', 'ebml_id': 'uint32_t',
 220                                 'float': 'double', 'sint': 'int64_t',
 221                                 }[valtype]
 222             except KeyError:
 223                 raise SyntaxError('Unrecognized value type ' + valtype)
 224         self.subelements = ()
 225
 226     def add_subelements(self, subelements):
 227         self.subelements = subelements
 228         self.subids = set(x[0].elid for x in subelements)
 229
 230 elementd = {}
 231 elementlist = []
 232 def parse_elems(l, namespace):
 233     subelements = []
 234     for el in l:
 235         if isinstance(el, str):
 236             name, hexid, eltype = [x.strip() for x in el.split(',')]
 237             multiple = name.endswith('*')
 238             name = name.strip('*')
 239             new = MatroskaElement(name, hexid, eltype, namespace)
 240             elementd[hexid] = new
 241             elementlist.append(new)
 242             subelements.append((new, multiple))
 243         else:
 244             new.add_subelements(parse_elems(el, namespace))
 245     return subelements
 246
 247 parse_elems(elements_ebml, 'EBML')
 248 parse_elems(elements_matroska, 'MATROSKA')
 249
 250 def generate_C_header():
 251     print('// Generated by TOOLS/matroska.py, do not edit manually')
 252     print()
 253
 254     for el in elementlist:
 255         print('#define {0.definename:40} 0x{0.elid}'.format(el))
 256
 257     print()
 258
 259     for el in reversed(elementlist):
 260         if not el.subelements:
 261             continue
 262         print()
 263         print('struct {0.structname} {{'.format(el))
 264         l = max(len(subel.valname) for subel, multiple in el.subelements)+1
 265         for subel, multiple in el.subelements:
 266             print('    {e.valname:{l}} {star}{e.fieldname};'.format(
 267                     e=subel, l=l, star=' *'[multiple]))
 268         print()
 269         for subel, multiple in el.subelements:
 270             print('    int  n_{0.fieldname};'.format(subel))
 271         print('};')
 272
 273     for el in elementlist:
 274         if not el.subelements:
 275             continue
 276         print('extern const struct ebml_elem_desc {0.structname}_desc;'.format(
 277                 el))
 278
 279     print()
 280     print('#define MAX_EBML_SUBELEMENTS', max(len(el.subelements)
 281                                               for el in elementlist))
 282
 283
 284
 285 def generate_C_definitions():
 286     print('// Generated by TOOLS/matroska.py, do not edit manually')
 287     print()
 288     for el in reversed(elementlist):
 289         print()
 290         if el.subelements:
 291             print('#define N', el.fieldname)
 292             print('E_S("{}", {})'.format(el.name, len(el.subelements)))
 293             for subel, multiple in el.subelements:
 294                 print('F({0.definename}, {0.fieldname}, {1})'.format(
 295                         subel, int(multiple)))
 296             print('}};')
 297             print('#undef N')
 298         else:
 299             print('E("{0.name}", {0.fieldname}, {0.ebmltype})'.format(el))
 300
 301 def read(s, length):
 302     t = s.read(length)
 303     if len(t) != length:
 304         raise EOF
 305     return t
 306
 307 def read_id(s):
 308     t = read(s, 1)
 309     i = 0
 310     mask = 128
 311     if ord(t) == 0:
 312         raise SyntaxError
 313     while not ord(t) & mask:
 314         i += 1
 315         mask >>= 1
 316     t += read(s, i)
 317     return t
 318
 319 def read_vint(s):
 320     t = read(s, 1)
 321     i = 0
 322     mask = 128
 323     if ord(t) == 0:
 324         raise SyntaxError
 325     while not ord(t) & mask:
 326         i += 1
 327         mask >>= 1
 328     t = bytes((ord(t) & (mask - 1),))
 329     t += read(s, i)
 330     return i+1, byte2num(t)
 331
 332 def read_str(s, length):
 333     return read(s, length)
 334
 335 def read_uint(s, length):
 336     t = read(s, length)
 337     return byte2num(t)
 338
 339 def read_sint(s, length):
 340     i = read_uint(s, length)
 341     mask = 1 << (length * 8 - 1)
 342     if i & mask:
 343         i -= 2 * mask
 344     return i
 345
 346 def read_float(s, length):
 347     t = read(s, length)
 348     i = byte2num(t)
 349     if length == 4:
 350         f = ldexp((i & 0x7fffff) + (1 << 23), (i >> 23 & 0xff) - 150)
 351         if i & (1 << 31):
 352             f = -f
 353     elif length == 8:
 354         f = ldexp((i & ((1 << 52) - 1)) + (1 << 52), (i >> 52 & 0x7ff) - 1075)
 355         if i & (1 << 63):
 356             f = -f
 357     else:
 358         raise SyntaxError
 359     return f
 360
 361 def parse_one(s, depth, parent, maxlen):
 362     elid = hexlify(read_id(s)).decode('ascii')
 363     elem = elementd.get(elid)
 364     if parent is not None and elid not in parent.subids and elid not in ('ec', 'bf'):
 365         print('Unexpected:', elid)
 366         if 1:
 367             raise NotImplementedError
 368     size, length = read_vint(s)
 369     this_length = len(elid) / 2 + size + length
 370     if elem is not None:
 371         if elem.valtype != 'skip':
 372             print(depth, elid, elem.name, 'size:', length, 'value:', end=' ')
 373         if elem.valtype == 'sub':
 374             print('subelements:')
 375             while length > 0:
 376                 length -= parse_one(s, depth + 1, elem, length)
 377             if length < 0:
 378                 raise SyntaxError
 379         elif elem.valtype == 'str':
 380             print('string', repr(read_str(s, length).decode('utf8', 'replace')))
 381         elif elem.valtype in ('binary', 'ebml_id'):
 382             t = read_str(s, length)
 383             dec = ''
 384             if elem.valtype == 'ebml_id':
 385                 idelem = elementd.get(hexlify(t).decode('ascii'))
 386                 if idelem is None:
 387                     dec = '(UNKNOWN)'
 388                 else:
 389                     dec = '({0.name})'.format(idelem)
 390             if len(t) < 20:
 391                 t = hexlify(t).decode('ascii')
 392             else:
 393                 t = '<skipped {} bytes>'.format(len(t))
 394             print('binary', t, dec)
 395         elif elem.valtype == 'uint':
 396             print('uint', read_uint(s, length))
 397         elif elem.valtype == 'sint':
 398             print('sint', read_sint(s, length))
 399         elif elem.valtype == 'float':
 400             print('float', read_float(s, length))
 401         elif elem.valtype == 'skip':
 402             read(s, length)
 403         else:
 404             raise NotImplementedError
 405     else:
 406         print(depth, 'Unknown element:', elid, 'size:', length)
 407         read(s, length)
 408     return this_length
 409
 410 def parse_toplevel(s):
 411     parse_one(s, 0, None, 1 << 63)
 412
 413 if sys.argv[1] == '--generate-header':
 414     generate_C_header()
 415 elif sys.argv[1] == '--generate-definitions':
 416     generate_C_definitions()
 417 else:
 418     s = open(sys.argv[1], "rb")
 419     while 1:
 420         start = s.tell()
 421         try:
 422             parse_toplevel(s)
 423         except EOF:
 424             if s.tell() != start:
 425                 raise Exception("Unexpected end of file")
 426             break