Lib/wave.py

   1 # Stuff to parse WAVE files.
   2 #
   3 # Usage.
   4 #
   5 # Reading WAVE files:
   6 #       f = wave.open(file, 'r')
   7 # where file is either the name of a file or an open file pointer.
   8 # The open file pointer must have methods read(), seek(), and close().
   9 # When the setpos() and rewind() methods are not used, the seek()
  10 # method is not  necessary.
  11 #
  12 # This returns an instance of a class with the following public methods:
  13 #       getnchannels()  -- returns number of audio channels (1 for
  14 #                          mono, 2 for stereo)
  15 #       getsampwidth()  -- returns sample width in bytes
  16 #       getframerate()  -- returns sampling frequency
  17 #       getnframes()    -- returns number of audio frames
  18 #       getcomptype()   -- returns compression type ('NONE' for AIFF files)
  19 #       getcompname()   -- returns human-readable version of
  20 #                          compression type ('not compressed' for AIFF files)
  21 #       getparams()     -- returns a tuple consisting of all of the
  22 #                          above in the above order
  23 #       getmarkers()    -- returns None (for compatibility with the
  24 #                          aifc module)
  25 #       getmark(id)     -- raises an error since the mark does not
  26 #                          exist (for compatibility with the aifc module)
  27 #       readframes(n)   -- returns at most n frames of audio
  28 #       rewind()        -- rewind to the beginning of the audio stream
  29 #       setpos(pos)     -- seek to the specified position
  30 #       tell()          -- return the current position
  31 #       close()         -- close the instance (make it unusable)
  32 # The position returned by tell() and the position given to setpos()
  33 # are compatible and have nothing to do with the actual postion in the
  34 # file.
  35 # The close() method is called automatically when the class instance
  36 # is destroyed.
  37 #
  38 # Writing WAVE files:
  39 #       f = wave.open(file, 'w')
  40 # where file is either the name of a file or an open file pointer.
  41 # The open file pointer must have methods write(), tell(), seek(), and
  42 # close().
  43 #
  44 # This returns an instance of a class with the following public methods:
  45 #       setnchannels(n) -- set the number of channels
  46 #       setsampwidth(n) -- set the sample width
  47 #       setframerate(n) -- set the frame rate
  48 #       setnframes(n)   -- set the number of frames
  49 #       setcomptype(type, name)
  50 #                       -- set the compression type and the
  51 #                          human-readable compression type
  52 #       setparams(nchannels, sampwidth, framerate, nframes, comptype, compname)
  53 #                       -- set all parameters at once
  54 #       tell()          -- return current position in output file
  55 #       writeframesraw(data)
  56 #                       -- write audio frames without pathing up the
  57 #                          file header
  58 #       writeframes(data)
  59 #                       -- write audio frames and patch up the file header
  60 #       close()         -- patch up the file header and close the
  61 #                          output file
  62 # You should set the parameters before the first writeframesraw or
  63 # writeframes.  The total number of frames does not need to be set,
  64 # but when it is set to the correct value, the header does not have to
  65 # be patched up.
  66 # It is best to first set all parameters, perhaps possibly the
  67 # compression type, and then write audio frames using writeframesraw.
  68 # When all frames have been written, either call writeframes('') or
  69 # close() to patch up the sizes in the header.
  70 # The close() method is called automatically when the class instance
  71 # is destroyed.
  72
  73 import __builtin__
  74
  75 Error = 'wave.Error'
  76
  77 WAVE_FORMAT_PCM = 0x0001
  78
  79 _array_fmts = None, 'b', 'h', None, 'l'
  80
  81 def _read_long(file):
  82         x = 0L
  83         for i in range(4):
  84                 byte = file.read(1)
  85                 if byte == '':
  86                         raise EOFError
  87                 x = x + (ord(byte) << (8 * i))
  88         if x >= 0x80000000L:
  89                 x = x - 0x100000000L
  90         return int(x)
  91
  92 def _read_ulong(file):
  93         x = 0L
  94         for i in range(4):
  95                 byte = file.read(1)
  96                 if byte == '':
  97                         raise EOFError
  98                 x = x + (ord(byte) << (8 * i))
  99         return x
 100
 101 def _read_short(file):
 102         x = 0
 103         for i in range(2):
 104                 byte = file.read(1)
 105                 if byte == '':
 106                         raise EOFError
 107                 x = x + (ord(byte) << (8 * i))
 108         if x >= 0x8000:
 109                 x = x - 0x10000
 110         return x
 111
 112 def _write_short(f, x):
 113         d, m = divmod(x, 256)
 114         f.write(chr(m))
 115         f.write(chr(d))
 116
 117 def _write_long(f, x):
 118         if x < 0:
 119                 x = x + 0x100000000L
 120         for i in range(4):
 121                 d, m = divmod(x, 256)
 122                 f.write(chr(int(m)))
 123                 x = d
 124
 125 class Chunk:
 126         def __init__(self, file):
 127                 self.file = file
 128                 self.chunkname = self.file.read(4)
 129                 if len(self.chunkname) < 4:
 130                         raise EOFError
 131                 self.chunksize = _read_long(self.file)
 132                 self.size_read = 0
 133                 self.offset = self.file.tell()
 134
 135         def rewind(self):
 136                 self.file.seek(self.offset, 0)
 137                 self.size_read = 0
 138
 139         def setpos(self, pos):
 140                 if pos < 0 or pos > self.chunksize:
 141                         raise RuntimeError
 142                 self.file.seek(self.offset + pos, 0)
 143                 self.size_read = pos
 144
 145         def read(self, length):
 146                 if self.size_read >= self.chunksize:
 147                         return ''
 148                 if length > self.chunksize - self.size_read:
 149                         length = self.chunksize - self.size_read
 150                 data = self.file.read(length)
 151                 self.size_read = self.size_read + len(data)
 152                 return data
 153
 154         def skip(self):
 155                 try:
 156                         self.file.seek(self.chunksize - self.size_read, 1)
 157                 except RuntimeError:
 158                         while self.size_read < self.chunksize:
 159                                 dummy = self.read(8192)
 160                                 if not dummy:
 161                                         raise EOFError
 162
 163 class Wave_read:
 164         # Variables used in this class:
 165         #
 166         # These variables are available to the user though appropriate
 167         # methods of this class:
 168         # _file -- the open file with methods read(), close(), and seek()
 169         #               set through the __init__() method
 170         # _nchannels -- the number of audio channels
 171         #               available through the getnchannels() method
 172         # _nframes -- the number of audio frames
 173         #               available through the getnframes() method
 174         # _sampwidth -- the number of bytes per audio sample
 175         #               available through the getsampwidth() method
 176         # _framerate -- the sampling frequency
 177         #               available through the getframerate() method
 178         # _comptype -- the AIFF-C compression type ('NONE' if AIFF)
 179         #               available through the getcomptype() method
 180         # _compname -- the human-readable AIFF-C compression type
 181         #               available through the getcomptype() method
 182         # _soundpos -- the position in the audio stream
 183         #               available through the tell() method, set through the
 184         #               setpos() method
 185         #
 186         # These variables are used internally only:
 187         # _fmt_chunk_read -- 1 iff the FMT chunk has been read
 188         # _data_seek_needed -- 1 iff positioned correctly in audio
 189         #               file for readframes()
 190         # _data_chunk -- instantiation of a chunk class for the DATA chunk
 191         # _framesize -- size of one frame in the file
 192
 193         access _file, _nchannels, _nframes, _sampwidth, _framerate, \
 194                   _comptype, _compname, _soundpos, \
 195                   _fmt_chunk_read, _data_seek_needed, \
 196                   _data_chunk, _framesize: private
 197
 198         def initfp(self, file):
 199                 self._file = file
 200                 self._convert = None
 201                 self._soundpos = 0
 202                 form = self._file.read(4)
 203                 if form != 'RIFF':
 204                         raise Error, 'file does not start with RIFF id'
 205                 formlength = _read_long(self._file)
 206                 if formlength <= 0:
 207                         raise Error, 'invalid FORM chunk data size'
 208                 formdata = self._file.read(4)
 209                 formlength = formlength - 4
 210                 if formdata != 'WAVE':
 211                         raise Error, 'not a WAVE file'
 212                 self._fmt_chunk_read = 0
 213                 while formlength > 0:
 214                         self._data_seek_needed = 1
 215                         chunk = Chunk(self._file)
 216                         if chunk.chunkname == 'fmt ':
 217                                 self._read_fmt_chunk(chunk)
 218                                 self._fmt_chunk_read = 1
 219                         elif chunk.chunkname == 'data':
 220                                 if not self._fmt_chunk_read:
 221                                         raise Error, 'data chunk before fmt chunk'
 222                                 self._data_chunk = chunk
 223                                 self._nframes = chunk.chunksize / self._framesize
 224                                 self._data_seek_needed = 0
 225                         formlength = formlength - 8 - chunk.chunksize
 226                         if formlength > 0:
 227                                 chunk.skip()
 228                 if not self._fmt_chunk_read or not self._data_chunk:
 229                         raise Error, 'fmt chunk and/or data chunk missing'
 230
 231         def __init__(self, f):
 232                 if type(f) == type(''):
 233                         f = __builtin__.open(f, 'r')
 234                 # else, assume it is an open file object already
 235                 self.initfp(f)
 236
 237         def __del__(self):
 238                 if self._file:
 239                         self.close()
 240
 241         #
 242         # User visible methods.
 243         #
 244         def getfp(self):
 245                 return self._file
 246
 247         def rewind(self):
 248                 self._data_seek_needed = 1
 249                 self._soundpos = 0
 250
 251         def close(self):
 252                 self._file = None
 253
 254         def tell(self):
 255                 return self._soundpos
 256
 257         def getnchannels(self):
 258                 return self._nchannels
 259
 260         def getnframes(self):
 261                 return self._nframes
 262
 263         def getsampwidth(self):
 264                 return self._sampwidth
 265
 266         def getframerate(self):
 267                 return self._framerate
 268
 269         def getcomptype(self):
 270                 return self._comptype
 271
 272         def getcompname(self):
 273                 return self._compname
 274
 275         def getparams(self):
 276                 return self.getnchannels(), self.getsampwidth(), \
 277                           self.getframerate(), self.getnframes(), \
 278                           self.getcomptype(), self.getcompname()
 279
 280         def getmarkers(self):
 281                 return None
 282
 283         def getmark(self, id):
 284                 raise Error, 'no marks'
 285
 286         def setpos(self, pos):
 287                 if pos < 0 or pos > self._nframes:
 288                         raise Error, 'position not in range'
 289                 self._soundpos = pos
 290                 self._data_seek_needed = 1
 291
 292         def readframes(self, nframes):
 293                 if self._data_seek_needed:
 294                         self._data_chunk.rewind()
 295                         pos = self._soundpos * self._framesize
 296                         if pos:
 297                                 self._data_chunk.setpos(pos)
 298                         self._data_seek_needed = 0
 299                 if nframes == 0:
 300                         return ''
 301                 if self._sampwidth > 1:
 302                         # unfortunately the fromfile() method does not take
 303                         # something that only looks like a file object, so
 304                         # we have to reach into the innards of the chunk object
 305                         import array
 306                         data = array.array(_array_fmts[self._sampwidth])
 307                         nitems = nframes * self._nchannels
 308                         if nitems * self._sampwidth > self._data_chunk.chunksize - self._data_chunk.size_read:
 309                                 nitems = (self._data_chunk.chunksize - self._data_chunk.size_read) / self._sampwidth
 310                         data.fromfile(self._data_chunk.file, nitems)
 311                         self._data_chunk.size_read = self._data_chunk.size_read + nitems * self._sampwidth
 312                         data.byteswap()
 313                         data = data.tostring()
 314                 else:
 315                         data = self._data_chunk.read(nframes * self._framesize)
 316                 if self._convert and data:
 317                         data = self._convert(data)
 318                 self._soundpos = self._soundpos + len(data) / (self._nchannels * self._sampwidth)
 319                 return data
 320
 321         #
 322         # Internal methods.
 323         #
 324         access *: private
 325
 326         def _read_fmt_chunk(self, chunk):
 327                 wFormatTag = _read_short(chunk)
 328                 self._nchannels = _read_short(chunk)
 329                 self._framerate = _read_long(chunk)
 330                 dwAvgBytesPerSec = _read_long(chunk)
 331                 wBlockAlign = _read_short(chunk)
 332                 if wFormatTag == WAVE_FORMAT_PCM:
 333                         self._sampwidth = (_read_short(chunk) + 7) / 8
 334                 else:
 335                         raise Error, 'unknown format'
 336                 self._framesize = self._nchannels * self._sampwidth
 337                 self._comptype = 'NONE'
 338                 self._compname = 'not compressed'
 339
 340 class Wave_write:
 341         # Variables used in this class:
 342         #
 343         # These variables are user settable through appropriate methods
 344         # of this class:
 345         # _file -- the open file with methods write(), close(), tell(), seek()
 346         #               set through the __init__() method
 347         # _comptype -- the AIFF-C compression type ('NONE' in AIFF)
 348         #               set through the setcomptype() or setparams() method
 349         # _compname -- the human-readable AIFF-C compression type
 350         #               set through the setcomptype() or setparams() method
 351         # _nchannels -- the number of audio channels
 352         #               set through the setnchannels() or setparams() method
 353         # _sampwidth -- the number of bytes per audio sample
 354         #               set through the setsampwidth() or setparams() method
 355         # _framerate -- the sampling frequency
 356         #               set through the setframerate() or setparams() method
 357         # _nframes -- the number of audio frames written to the header
 358         #               set through the setnframes() or setparams() method
 359         #
 360         # These variables are used internally only:
 361         # _datalength -- the size of the audio samples written to the header
 362         # _nframeswritten -- the number of frames actually written
 363         # _datawritten -- the size of the audio samples actually written
 364
 365         access _file, _comptype, _compname, _nchannels, _sampwidth, \
 366                   _framerate, _nframes, _nframeswritten, \
 367                   _datalength, _datawritten: private
 368
 369         def __init__(self, f):
 370                 if type(f) == type(''):
 371                         f = __builtin__.open(f, 'w')
 372                 self.initfp(f)
 373
 374         def initfp(self, file):
 375                 self._file = file
 376                 self._convert = None
 377                 self._nchannels = 0
 378                 self._sampwidth = 0
 379                 self._framerate = 0
 380                 self._nframes = 0
 381                 self._nframeswritten = 0
 382                 self._datawritten = 0
 383                 self._datalength = 0
 384
 385         def __del__(self):
 386                 if self._file:
 387                         self.close()
 388
 389         #
 390         # User visible methods.
 391         #
 392         def setnchannels(self, nchannels):
 393                 if self._datawritten:
 394                         raise Error, 'cannot change parameters after starting to write'
 395                 if nchannels < 1:
 396                         raise Error, 'bad # of channels'
 397                 self._nchannels = nchannels
 398
 399         def getnchannels(self):
 400                 if not self._nchannels:
 401                         raise Error, 'number of channels not set'
 402                 return self._nchannels
 403
 404         def setsampwidth(self, sampwidth):
 405                 if self._datawritten:
 406                         raise Error, 'cannot change parameters after starting to write'
 407                 if sampwidth < 1 or sampwidth > 4:
 408                         raise Error, 'bad sample width'
 409                 self._sampwidth = sampwidth
 410
 411         def getsampwidth(self):
 412                 if not self._sampwidth:
 413                         raise Error, 'sample width not set'
 414                 return self._sampwidth
 415
 416         def setframerate(self, framerate):
 417                 if self._datawritten:
 418                         raise Error, 'cannot change parameters after starting to write'
 419                 if framerate <= 0:
 420                         raise Error, 'bad frame rate'
 421                 self._framerate = framerate
 422
 423         def getframerate(self):
 424                 if not self._framerate:
 425                         raise Error, 'frame rate not set'
 426                 return self._framerate
 427
 428         def setnframes(self, nframes):
 429                 if self._datawritten:
 430                         raise Error, 'cannot change parameters after starting to write'
 431                 self._nframes = nframes
 432
 433         def getnframes(self):
 434                 return self._nframeswritten
 435
 436         def setcomptype(self, comptype, compname):
 437                 if self._datawritten:
 438                         raise Error, 'cannot change parameters after starting to write'
 439                 if comptype not in ('NONE',):
 440                         raise Error, 'unsupported compression type'
 441                 self._comptype = comptype
 442                 self._compname = compname
 443
 444         def getcomptype(self):
 445                 return self._comptype
 446
 447         def getcompname(self):
 448                 return self._compname
 449
 450         def setparams(self, (nchannels, sampwidth, framerate, nframes, comptype, compname)):
 451                 if self._datawritten:
 452                         raise Error, 'cannot change parameters after starting to write'
 453                 self.setnchannels(nchannels)
 454                 self.setsampwidth(sampwidth)
 455                 self.setframerate(framerate)
 456                 self.setnframes(nframes)
 457                 self.setcomptype(comptype, compname)
 458
 459         def getparams(self):
 460                 if not self._nchannels or not self._sampwidth or not self._framerate:
 461                         raise Error, 'not all parameters set'
 462                 return self._nchannels, self._sampwidth, self._framerate, \
 463                           self._nframes, self._comptype, self._compname
 464
 465         def setmark(self, id, pos, name):
 466                 raise Error, 'setmark() not supported'
 467
 468         def getmark(self, id):
 469                 raise Error, 'no marks'
 470
 471         def getmarkers(self):
 472                 return None
 473
 474         def tell(self):
 475                 return self._nframeswritten
 476
 477         def writeframesraw(self, data):
 478                 self._ensure_header_written(len(data))
 479                 nframes = len(data) / (self._sampwidth * self._nchannels)
 480                 if self._convert:
 481                         data = self._convert(data)
 482                 if self._sampwidth > 1:
 483                         import array
 484                         data = array.array(_array_fmts[self._sampwidth], data)
 485                         data.byteswap()
 486                         data.tofile(self._file)
 487                         self._datawritten = self._datawritten + len(data) * self._sampwidth
 488                 else:
 489                         self._file.write(data)
 490                         self._datawritten = self._datawritten + len(data)
 491                 self._nframeswritten = self._nframeswritten + nframes
 492
 493         def writeframes(self, data):
 494                 self.writeframesraw(data)
 495                 if self._datalength != self._datawritten:
 496                         self._patchheader()
 497
 498         def close(self):
 499                 self._ensure_header_written(0)
 500                 if self._datalength != self._datawritten:
 501                         self._patchheader()
 502                 self._file.flush()
 503                 self._file = None
 504
 505         #
 506         # Internal methods.
 507         #
 508         access *: private
 509
 510         def _ensure_header_written(self, datasize):
 511                 if not self._datawritten:
 512                         if not self._nchannels:
 513                                 raise Error, '# channels not specified'
 514                         if not self._sampwidth:
 515                                 raise Error, 'sample width not specified'
 516                         if not self._framerate:
 517                                 raise Error, 'sampling rate not specified'
 518                         self._write_header(datasize)
 519
 520         def _write_header(self, initlength):
 521                 self._file.write('RIFF')
 522                 if not self._nframes:
 523                         self._nframes = initlength / (self._nchannels * self._sampwidth)
 524                 self._datalength = self._nframes * self._nchannels * self._sampwidth
 525                 self._form_length_pos = self._file.tell()
 526                 _write_long(self._file, 36 + self._datalength)
 527                 self._file.write('WAVE')
 528                 self._file.write('fmt ')
 529                 _write_long(self._file, 16)
 530                 _write_short(self._file, WAVE_FORMAT_PCM)
 531                 _write_short(self._file, self._nchannels)
 532                 _write_long(self._file, self._framerate)
 533                 _write_long(self._file, self._nchannels * self._framerate * self._sampwidth)
 534                 _write_short(self._file, self._nchannels * self._sampwidth)
 535                 _write_short(self._file, self._sampwidth * 8)
 536                 self._file.write('data')
 537                 self._data_length_pos = self._file.tell()
 538                 _write_long(self._file, self._datalength)
 539
 540         def _patchheader(self):
 541                 if self._datawritten == self._datalength:
 542                         return
 543                 curpos = self._file.tell()
 544                 self._file.seek(self._form_length_pos, 0)
 545                 _write_long(36 + self._datawritten)
 546                 self._file.seek(self._data_length_pos, 0)
 547                 _write_long(self._file, self._datawritten)
 548                 self._file.seek(curpos, 0)
 549                 self._datalength = self._datawritten
 550
 551 def open(f, mode):
 552         if mode == 'r':
 553                 return Wave_read(f)
 554         elif mode == 'w':
 555                 return Wave_write(f)
 556         else:
 557                 raise Error, "mode must be 'r' or 'w'"
 558
 559 openfp = open # B/W compatibility