| 1 | #if 0 |
|---|
| 2 | # $Id: mpeginfo.py,v 1.33 2005/02/15 18:52:51 dischi Exp $ |
|---|
| 3 | # $Log: mpeginfo.py,v $ |
|---|
| 4 | # Revision 1.33 2005/02/15 18:52:51 dischi |
|---|
| 5 | # some strange bugfix (what is this doing?) |
|---|
| 6 | # |
|---|
| 7 | # Revision 1.32 2005/01/21 16:37:02 dischi |
|---|
| 8 | # try to find bad timestamps |
|---|
| 9 | # |
|---|
| 10 | # Revision 1.31 2005/01/08 12:06:45 dischi |
|---|
| 11 | # make sure the buffer is big enough |
|---|
| 12 | # |
|---|
| 13 | # Revision 1.30 2005/01/02 14:57:27 dischi |
|---|
| 14 | # detect ac3 in normal mpeg2 |
|---|
| 15 | # |
|---|
| 16 | # Revision 1.29 2004/11/27 14:42:12 dischi |
|---|
| 17 | # remove future warning |
|---|
| 18 | # |
|---|
| 19 | # Revision 1.28 2004/11/15 21:43:36 dischi |
|---|
| 20 | # remove bad debugging stuff |
|---|
| 21 | # |
|---|
| 22 | # Revision 1.27 2004/11/12 18:10:45 dischi |
|---|
| 23 | # add ac3 support in mpeg streams |
|---|
| 24 | # |
|---|
| 25 | # Revision 1.26 2004/10/04 18:06:54 dischi |
|---|
| 26 | # test length of remaining buffer |
|---|
| 27 | # |
|---|
| 28 | # Revision 1.25 2004/07/11 19:37:25 dischi |
|---|
| 29 | # o read more bytes on ts scan |
|---|
| 30 | # o support for AC3 in private streams |
|---|
| 31 | # |
|---|
| 32 | # Revision 1.24 2004/07/03 09:01:32 dischi |
|---|
| 33 | # o fix PES start detection inside TS |
|---|
| 34 | # o try to find out if the stream is progressive or interlaced |
|---|
| 35 | # |
|---|
| 36 | # Revision 1.23 2004/06/23 19:44:10 dischi |
|---|
| 37 | # better length detection, big cleanup |
|---|
| 38 | # |
|---|
| 39 | # Revision 1.22 2004/06/22 21:37:34 dischi |
|---|
| 40 | # o PES support |
|---|
| 41 | # o basic length detection for TS and PES |
|---|
| 42 | # |
|---|
| 43 | # Revision 1.21 2004/06/21 20:37:34 dischi |
|---|
| 44 | # basic support for mpeg-ts |
|---|
| 45 | # |
|---|
| 46 | # Revision 1.20 2004/03/13 23:41:59 dischi |
|---|
| 47 | # add AudioInfo to mpeg for all streams |
|---|
| 48 | # |
|---|
| 49 | # Revision 1.19 2004/02/11 20:11:54 dischi |
|---|
| 50 | # Updated length calculation for mpeg files. This may not work for all files. |
|---|
| 51 | # |
|---|
| 52 | # |
|---|
| 53 | # MMPython - Media Metadata for Python |
|---|
| 54 | # Copyright (C) 2003 Thomas Schueppel |
|---|
| 55 | # |
|---|
| 56 | # This program is free software; you can redistribute it and/or modify |
|---|
| 57 | # it under the terms of the GNU General Public License as published by |
|---|
| 58 | # the Free Software Foundation; either version 2 of the License, or |
|---|
| 59 | # (at your option) any later version. |
|---|
| 60 | # |
|---|
| 61 | # This program is distributed in the hope that it will be useful, but |
|---|
| 62 | # WITHOUT ANY WARRANTY; without even the implied warranty of MER- |
|---|
| 63 | # CHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General |
|---|
| 64 | # Public License for more details. |
|---|
| 65 | # |
|---|
| 66 | # You should have received a copy of the GNU General Public License along |
|---|
| 67 | # with this program; if not, write to the Free Software Foundation, Inc., |
|---|
| 68 | # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
|---|
| 69 | # |
|---|
| 70 | # ----------------------------------------------------------------------- |
|---|
| 71 | #endif |
|---|
| 72 | |
|---|
| 73 | import re |
|---|
| 74 | import os |
|---|
| 75 | import struct |
|---|
| 76 | import string |
|---|
| 77 | import fourcc |
|---|
| 78 | |
|---|
| 79 | from mmpython import mediainfo |
|---|
| 80 | import mmpython |
|---|
| 81 | import stat |
|---|
| 82 | |
|---|
| 83 | ##------------------------------------------------------------------------ |
|---|
| 84 | ## START_CODE |
|---|
| 85 | ## |
|---|
| 86 | ## Start Codes, with 'slice' occupying 0x01..0xAF |
|---|
| 87 | ##------------------------------------------------------------------------ |
|---|
| 88 | START_CODE = { |
|---|
| 89 | 0x00 : 'picture_start_code', |
|---|
| 90 | 0xB0 : 'reserved', |
|---|
| 91 | 0xB1 : 'reserved', |
|---|
| 92 | 0xB2 : 'user_data_start_code', |
|---|
| 93 | 0xB3 : 'sequence_header_code', |
|---|
| 94 | 0xB4 : 'sequence_error_code', |
|---|
| 95 | 0xB5 : 'extension_start_code', |
|---|
| 96 | 0xB6 : 'reserved', |
|---|
| 97 | 0xB7 : 'sequence end', |
|---|
| 98 | 0xB8 : 'group of pictures', |
|---|
| 99 | } |
|---|
| 100 | for i in range(0x01,0xAF): |
|---|
| 101 | START_CODE[i] = 'slice_start_code' |
|---|
| 102 | |
|---|
| 103 | ##------------------------------------------------------------------------ |
|---|
| 104 | ## START CODES |
|---|
| 105 | ##------------------------------------------------------------------------ |
|---|
| 106 | PICTURE = 0x00 |
|---|
| 107 | USERDATA = 0xB2 |
|---|
| 108 | SEQ_HEAD = 0xB3 |
|---|
| 109 | SEQ_ERR = 0xB4 |
|---|
| 110 | EXT_START = 0xB5 |
|---|
| 111 | SEQ_END = 0xB7 |
|---|
| 112 | GOP = 0xB8 |
|---|
| 113 | |
|---|
| 114 | SEQ_START_CODE = 0xB3 |
|---|
| 115 | PACK_PKT = 0xBA |
|---|
| 116 | SYS_PKT = 0xBB |
|---|
| 117 | PADDING_PKT = 0xBE |
|---|
| 118 | AUDIO_PKT = 0xC0 |
|---|
| 119 | VIDEO_PKT = 0xE0 |
|---|
| 120 | PRIVATE_STREAM1 = 0xBD |
|---|
| 121 | PRIVATE_STREAM2 = 0xBf |
|---|
| 122 | |
|---|
| 123 | TS_PACKET_LENGTH = 188 |
|---|
| 124 | TS_SYNC = 0x47 |
|---|
| 125 | |
|---|
| 126 | ##------------------------------------------------------------------------ |
|---|
| 127 | ## FRAME_RATE |
|---|
| 128 | ## |
|---|
| 129 | ## A lookup table of all the standard frame rates. Some rates adhere to |
|---|
| 130 | ## a particular profile that ensures compatibility with VLSI capabilities |
|---|
| 131 | ## of the early to mid 1990s. |
|---|
| 132 | ## |
|---|
| 133 | ## CPB |
|---|
| 134 | ## Constrained Parameters Bitstreams, an MPEG-1 set of sampling and |
|---|
| 135 | ## bitstream parameters designed to normalize decoder computational |
|---|
| 136 | ## complexity, buffer size, and memory bandwidth while still addressing |
|---|
| 137 | ## the widest possible range of applications. |
|---|
| 138 | ## |
|---|
| 139 | ## Main Level |
|---|
| 140 | ## MPEG-2 Video Main Profile and Main Level is analogous to MPEG-1's |
|---|
| 141 | ## CPB, with sampling limits at CCIR 601 parameters (720x480x30 Hz or |
|---|
| 142 | ## 720x576x24 Hz). |
|---|
| 143 | ## |
|---|
| 144 | ##------------------------------------------------------------------------ |
|---|
| 145 | FRAME_RATE = [ |
|---|
| 146 | 0, |
|---|
| 147 | round(24000.0/1001*100)/100, ## 3-2 pulldown NTSC (CPB/Main Level) |
|---|
| 148 | 24, ## Film (CPB/Main Level) |
|---|
| 149 | 25, ## PAL/SECAM or 625/60 video |
|---|
| 150 | round(30000.0/1001*100)/100, ## NTSC (CPB/Main Level) |
|---|
| 151 | 30, ## drop-frame NTSC or component 525/60 (CPB/Main Level) |
|---|
| 152 | 50, ## double-rate PAL |
|---|
| 153 | round(60000.0/1001*100)/100, ## double-rate NTSC |
|---|
| 154 | 60, ## double-rate, drop-frame NTSC/component 525/60 video |
|---|
| 155 | ] |
|---|
| 156 | |
|---|
| 157 | ##------------------------------------------------------------------------ |
|---|
| 158 | ## ASPECT_RATIO -- INCOMPLETE? |
|---|
| 159 | ## |
|---|
| 160 | ## This lookup table maps the header aspect ratio index to a common name. |
|---|
| 161 | ## These are just the defined ratios for CPB I believe. As I understand |
|---|
| 162 | ## it, a stream that doesn't adhere to one of these aspect ratios is |
|---|
| 163 | ## technically considered non-compliant. |
|---|
| 164 | ##------------------------------------------------------------------------ |
|---|
| 165 | ASPECT_RATIO = [ 'Forbidden', |
|---|
| 166 | '1/1 (VGA)', |
|---|
| 167 | '4/3 (TV)', |
|---|
| 168 | '16/9 (Large TV)', |
|---|
| 169 | '2.21/1 (Cinema)', |
|---|
| 170 | ] |
|---|
| 171 | |
|---|
| 172 | |
|---|
| 173 | class MpegInfo(mediainfo.AVInfo): |
|---|
| 174 | def __init__(self,file): |
|---|
| 175 | mediainfo.AVInfo.__init__(self) |
|---|
| 176 | self.context = 'video' |
|---|
| 177 | self.sequence_header_offset = 0 |
|---|
| 178 | |
|---|
| 179 | # detect TS (fast scan) |
|---|
| 180 | self.valid = self.isTS(file) |
|---|
| 181 | |
|---|
| 182 | if not self.valid: |
|---|
| 183 | # detect system mpeg (many infos) |
|---|
| 184 | self.valid = self.isMPEG(file) |
|---|
| 185 | |
|---|
| 186 | if not self.valid: |
|---|
| 187 | # detect PES |
|---|
| 188 | self.valid = self.isPES(file) |
|---|
| 189 | |
|---|
| 190 | if self.valid: |
|---|
| 191 | self.mime = 'video/mpeg' |
|---|
| 192 | if not self.video: |
|---|
| 193 | self.video.append(mediainfo.VideoInfo()) |
|---|
| 194 | |
|---|
| 195 | if self.sequence_header_offset <= 0: |
|---|
| 196 | return |
|---|
| 197 | |
|---|
| 198 | self.progressive(file) |
|---|
| 199 | |
|---|
| 200 | for vi in self.video: |
|---|
| 201 | vi.width, vi.height = self.dxy(file) |
|---|
| 202 | vi.fps, vi.aspect = self.framerate_aspect(file) |
|---|
| 203 | vi.bitrate = self.bitrate(file) |
|---|
| 204 | if self.length: |
|---|
| 205 | vi.length = self.length |
|---|
| 206 | |
|---|
| 207 | if not self.type: |
|---|
| 208 | if self.video[0].width == 480: |
|---|
| 209 | self.type = 'MPEG2 video' # SVCD spec |
|---|
| 210 | elif self.video[0].width == 352: |
|---|
| 211 | self.type = 'MPEG1 video' # VCD spec |
|---|
| 212 | else: |
|---|
| 213 | self.type = 'MPEG video' |
|---|
| 214 | |
|---|
| 215 | if mediainfo.DEBUG > 2: |
|---|
| 216 | self.__scan__() |
|---|
| 217 | |
|---|
| 218 | |
|---|
| 219 | def dxy(self,file): |
|---|
| 220 | """ |
|---|
| 221 | get width and height of the video |
|---|
| 222 | """ |
|---|
| 223 | file.seek(self.sequence_header_offset+4,0) |
|---|
| 224 | v = file.read(4) |
|---|
| 225 | x = struct.unpack('>H',v[:2])[0] >> 4 |
|---|
| 226 | y = struct.unpack('>H',v[1:3])[0] & 0x0FFF |
|---|
| 227 | return (x,y) |
|---|
| 228 | |
|---|
| 229 | |
|---|
| 230 | def framerate_aspect(self,file): |
|---|
| 231 | """ |
|---|
| 232 | read framerate and aspect ratio |
|---|
| 233 | """ |
|---|
| 234 | file.seek(self.sequence_header_offset+7,0) |
|---|
| 235 | v = struct.unpack( '>B', file.read(1) )[0] |
|---|
| 236 | try: |
|---|
| 237 | fps = FRAME_RATE[v&0xf] |
|---|
| 238 | except IndexError: |
|---|
| 239 | fps = None |
|---|
| 240 | try: |
|---|
| 241 | aspect = ASPECT_RATIO[v>>4] |
|---|
| 242 | except IndexError: |
|---|
| 243 | if mediainfo.DEBUG: |
|---|
| 244 | print 'Index error: %s' % (v>>4) |
|---|
| 245 | aspect = None |
|---|
| 246 | return (fps, aspect) |
|---|
| 247 | |
|---|
| 248 | |
|---|
| 249 | def progressive(self, file): |
|---|
| 250 | """ |
|---|
| 251 | Try to find out with brute force if the mpeg is interlaced or not. |
|---|
| 252 | Search for the Sequence_Extension in the extension header (01B5) |
|---|
| 253 | """ |
|---|
| 254 | file.seek(0) |
|---|
| 255 | buffer = '' |
|---|
| 256 | count = 0 |
|---|
| 257 | while 1: |
|---|
| 258 | if len(buffer) < 1000: |
|---|
| 259 | count += 1 |
|---|
| 260 | if count > 1000: |
|---|
| 261 | break |
|---|
| 262 | buffer += file.read(1024) |
|---|
| 263 | if len(buffer) < 1000: |
|---|
| 264 | break |
|---|
| 265 | pos = buffer.find('\x00\x00\x01\xb5') |
|---|
| 266 | if pos == -1 or len(buffer) - pos < 5: |
|---|
| 267 | buffer = buffer[-10:] |
|---|
| 268 | continue |
|---|
| 269 | ext = (ord(buffer[pos+4]) >> 4) |
|---|
| 270 | if ext == 8: |
|---|
| 271 | pass |
|---|
| 272 | elif ext == 1: |
|---|
| 273 | if (ord(buffer[pos+5]) >> 3) & 1: |
|---|
| 274 | self.keys.append('progressive') |
|---|
| 275 | self.progressive = 1 |
|---|
| 276 | else: |
|---|
| 277 | self.keys.append('interlaced') |
|---|
| 278 | self.interlaced = 1 |
|---|
| 279 | return True |
|---|
| 280 | else: |
|---|
| 281 | print 'ext', ext |
|---|
| 282 | buffer = buffer[pos+4:] |
|---|
| 283 | return False |
|---|
| 284 | |
|---|
| 285 | |
|---|
| 286 | ##------------------------------------------------------------------------ |
|---|
| 287 | ## bitrate() |
|---|
| 288 | ## |
|---|
| 289 | ## From the MPEG-2.2 spec: |
|---|
| 290 | ## |
|---|
| 291 | ## bit_rate -- This is a 30-bit integer. The lower 18 bits of the |
|---|
| 292 | ## integer are in bit_rate_value and the upper 12 bits are in |
|---|
| 293 | ## bit_rate_extension. The 30-bit integer specifies the bitrate of the |
|---|
| 294 | ## bitstream measured in units of 400 bits/second, rounded upwards. |
|---|
| 295 | ## The value zero is forbidden. |
|---|
| 296 | ## |
|---|
| 297 | ## So ignoring all the variable bitrate stuff for now, this 30 bit integer |
|---|
| 298 | ## multiplied times 400 bits/sec should give the rate in bits/sec. |
|---|
| 299 | ## |
|---|
| 300 | ## TODO: Variable bitrates? I need one that implements this. |
|---|
| 301 | ## |
|---|
| 302 | ## Continued from the MPEG-2.2 spec: |
|---|
| 303 | ## |
|---|
| 304 | ## If the bitstream is a constant bitrate stream, the bitrate specified |
|---|
| 305 | ## is the actual rate of operation of the VBV specified in annex C. If |
|---|
| 306 | ## the bitstream is a variable bitrate stream, the STD specifications in |
|---|
| 307 | ## ISO/IEC 13818-1 supersede the VBV, and the bitrate specified here is |
|---|
| 308 | ## used to dimension the transport stream STD (2.4.2 in ITU-T Rec. xxx | |
|---|
| 309 | ## ISO/IEC 13818-1), or the program stream STD (2.4.5 in ITU-T Rec. xxx | |
|---|
| 310 | ## ISO/IEC 13818-1). |
|---|
| 311 | ## |
|---|
| 312 | ## If the bitstream is not a constant rate bitstream the vbv_delay |
|---|
| 313 | ## field shall have the value FFFF in hexadecimal. |
|---|
| 314 | ## |
|---|
| 315 | ## Given the value encoded in the bitrate field, the bitstream shall be |
|---|
| 316 | ## generated so that the video encoding and the worst case multiplex |
|---|
| 317 | ## jitter do not cause STD buffer overflow or underflow. |
|---|
| 318 | ## |
|---|
| 319 | ## |
|---|
| 320 | ##------------------------------------------------------------------------ |
|---|
| 321 | |
|---|
| 322 | |
|---|
| 323 | ## Some parts in the code are based on mpgtx (mpgtx.sf.net) |
|---|
| 324 | |
|---|
| 325 | def bitrate(self,file): |
|---|
| 326 | """ |
|---|
| 327 | read the bitrate (most of the time broken) |
|---|
| 328 | """ |
|---|
| 329 | file.seek(self.sequence_header_offset+8,0) |
|---|
| 330 | t,b = struct.unpack( '>HB', file.read(3) ) |
|---|
| 331 | vrate = t << 2 | b >> 6 |
|---|
| 332 | return vrate * 400 |
|---|
| 333 | |
|---|
| 334 | |
|---|
| 335 | def ReadSCRMpeg2(self, buffer): |
|---|
| 336 | """ |
|---|
| 337 | read SCR (timestamp) for MPEG2 at the buffer beginning (6 Bytes) |
|---|
| 338 | """ |
|---|
| 339 | highbit = (ord(buffer[0])&0x20)>>5 |
|---|
| 340 | |
|---|
| 341 | low4Bytes= ((long(ord(buffer[0])) & 0x18) >> 3) << 30 |
|---|
| 342 | low4Bytes |= (ord(buffer[0]) & 0x03) << 28 |
|---|
| 343 | low4Bytes |= ord(buffer[1]) << 20 |
|---|
| 344 | low4Bytes |= (ord(buffer[2]) & 0xF8) << 12 |
|---|
| 345 | low4Bytes |= (ord(buffer[2]) & 0x03) << 13 |
|---|
| 346 | low4Bytes |= ord(buffer[3]) << 5 |
|---|
| 347 | low4Bytes |= (ord(buffer[4])) >> 3 |
|---|
| 348 | |
|---|
| 349 | sys_clock_ref=(ord(buffer[4]) & 0x3) << 7 |
|---|
| 350 | sys_clock_ref|=(ord(buffer[5]) >> 1) |
|---|
| 351 | |
|---|
| 352 | return (long(highbit * (1<<16) * (1<<16)) + low4Bytes) / 90000 |
|---|
| 353 | |
|---|
| 354 | |
|---|
| 355 | def ReadSCRMpeg1(self, buffer): |
|---|
| 356 | """ |
|---|
| 357 | read SCR (timestamp) for MPEG1 at the buffer beginning (5 Bytes) |
|---|
| 358 | """ |
|---|
| 359 | highbit = (ord(buffer[0]) >> 3) & 0x01 |
|---|
| 360 | |
|---|
| 361 | low4Bytes = ((long(ord(buffer[0])) >> 1) & 0x03) << 30 |
|---|
| 362 | low4Bytes |= ord(buffer[1]) << 22; |
|---|
| 363 | low4Bytes |= (ord(buffer[2]) >> 1) << 15; |
|---|
| 364 | low4Bytes |= ord(buffer[3]) << 7; |
|---|
| 365 | low4Bytes |= ord(buffer[4]) >> 1; |
|---|
| 366 | |
|---|
| 367 | return (long(highbit) * (1<<16) * (1<<16) + low4Bytes) / 90000; |
|---|
| 368 | |
|---|
| 369 | |
|---|
| 370 | def ReadPTS(self, buffer): |
|---|
| 371 | """ |
|---|
| 372 | read PTS (PES timestamp) at the buffer beginning (5 Bytes) |
|---|
| 373 | """ |
|---|
| 374 | high = ((ord(buffer[0]) & 0xF) >> 1) |
|---|
| 375 | med = (ord(buffer[1]) << 7) + (ord(buffer[2]) >> 1) |
|---|
| 376 | low = (ord(buffer[3]) << 7) + (ord(buffer[4]) >> 1) |
|---|
| 377 | return ((long(high) << 30 ) + (med << 15) + low) / 90000 |
|---|
| 378 | |
|---|
| 379 | |
|---|
| 380 | def ReadHeader(self, buffer, offset): |
|---|
| 381 | """ |
|---|
| 382 | Handle MPEG header in buffer on position offset |
|---|
| 383 | Return -1 on error, new offset or 0 if the new offset can't be scanned |
|---|
| 384 | """ |
|---|
| 385 | if buffer[offset:offset+3] != '\x00\x00\x01': |
|---|
| 386 | return -1 |
|---|
| 387 | |
|---|
| 388 | id = ord(buffer[offset+3]) |
|---|
| 389 | |
|---|
| 390 | if id == PADDING_PKT: |
|---|
| 391 | return offset + (ord(buffer[offset+4]) << 8) + ord(buffer[offset+5]) + 6 |
|---|
| 392 | |
|---|
| 393 | if id == PACK_PKT: |
|---|
| 394 | if ord(buffer[offset+4]) & 0xF0 == 0x20: |
|---|
| 395 | self.type = 'MPEG1 video' |
|---|
| 396 | self.get_time = self.ReadSCRMpeg1 |
|---|
| 397 | return offset + 12 |
|---|
| 398 | elif (ord(buffer[offset+4]) & 0xC0) == 0x40: |
|---|
| 399 | self.type = 'MPEG2 video' |
|---|
| 400 | self.get_time = self.ReadSCRMpeg2 |
|---|
| 401 | return offset + (ord(buffer[offset+13]) & 0x07) + 14 |
|---|
| 402 | else: |
|---|
| 403 | # WTF? Very strange |
|---|
| 404 | return -1 |
|---|
| 405 | |
|---|
| 406 | if 0xC0 <= id <= 0xDF: |
|---|
| 407 | # code for audio stream |
|---|
| 408 | for a in self.audio: |
|---|
| 409 | if a.id == id: |
|---|
| 410 | break |
|---|
| 411 | else: |
|---|
| 412 | self.audio.append(mediainfo.AudioInfo()) |
|---|
| 413 | self.audio[-1].id = id |
|---|
| 414 | self.audio[-1].keys.append('id') |
|---|
| 415 | return 0 |
|---|
| 416 | |
|---|
| 417 | if 0xE0 <= id <= 0xEF: |
|---|
| 418 | # code for video stream |
|---|
| 419 | for v in self.video: |
|---|
| 420 | if v.id == id: |
|---|
| 421 | break |
|---|
| 422 | else: |
|---|
| 423 | self.video.append(mediainfo.VideoInfo()) |
|---|
| 424 | self.video[-1].id = id |
|---|
| 425 | self.video[-1].keys.append('id') |
|---|
| 426 | return 0 |
|---|
| 427 | |
|---|
| 428 | if id == SEQ_HEAD: |
|---|
| 429 | # sequence header, remember that position for later use |
|---|
| 430 | self.sequence_header_offset = offset |
|---|
| 431 | return 0 |
|---|
| 432 | |
|---|
| 433 | if id in (PRIVATE_STREAM1, PRIVATE_STREAM2): |
|---|
| 434 | # private stream. we don't know, but maybe we can guess later |
|---|
| 435 | add = ord(buffer[offset+8]) |
|---|
| 436 | # if (ord(buffer[offset+6]) & 4) or 1: |
|---|
| 437 | # id = ord(buffer[offset+10+add]) |
|---|
| 438 | if buffer[offset+11+add:offset+15+add].find('\x0b\x77') != -1: |
|---|
| 439 | # AC3 stream |
|---|
| 440 | for a in self.audio: |
|---|
| 441 | if a.id == id: |
|---|
| 442 | break |
|---|
| 443 | else: |
|---|
| 444 | self.audio.append(mediainfo.AudioInfo()) |
|---|
| 445 | self.audio[-1].id = id |
|---|
| 446 | self.audio[-1].codec = 'AC3' |
|---|
| 447 | self.audio[-1].keys.append('id') |
|---|
| 448 | return 0 |
|---|
| 449 | |
|---|
| 450 | if id == SYS_PKT: |
|---|
| 451 | return 0 |
|---|
| 452 | |
|---|
| 453 | if id == EXT_START: |
|---|
| 454 | return 0 |
|---|
| 455 | |
|---|
| 456 | return 0 |
|---|
| 457 | |
|---|
| 458 | |
|---|
| 459 | # Normal MPEG (VCD, SVCD) ======================================== |
|---|
| 460 | |
|---|
| 461 | def isMPEG(self, file): |
|---|
| 462 | """ |
|---|
| 463 | This MPEG starts with a sequence of 0x00 followed by a PACK Header |
|---|
| 464 | http://dvd.sourceforge.net/dvdinfo/packhdr.html |
|---|
| 465 | """ |
|---|
| 466 | file.seek(0,0) |
|---|
| 467 | buffer = file.read(10000) |
|---|
| 468 | offset = 0 |
|---|
| 469 | |
|---|
| 470 | # seek until the 0 byte stop |
|---|
| 471 | while buffer[offset] == '\0': |
|---|
| 472 | offset += 1 |
|---|
| 473 | offset -= 2 |
|---|
| 474 | |
|---|
| 475 | # test for mpeg header 0x00 0x00 0x01 |
|---|
| 476 | if not buffer[offset:offset+4] == '\x00\x00\x01%s' % chr(PACK_PKT): |
|---|
| 477 | return 0 |
|---|
| 478 | |
|---|
| 479 | # scan the 100000 bytes of data |
|---|
| 480 | buffer += file.read(100000) |
|---|
| 481 | |
|---|
| 482 | # scan first header, to get basic info about |
|---|
| 483 | # how to read a timestamp |
|---|
| 484 | self.ReadHeader(buffer, offset) |
|---|
| 485 | |
|---|
| 486 | # store first timestamp |
|---|
| 487 | self.start = self.get_time(buffer[offset+4:]) |
|---|
| 488 | while len(buffer) > offset + 1000 and buffer[offset:offset+3] == '\x00\x00\x01': |
|---|
| 489 | # read the mpeg header |
|---|
| 490 | new_offset = self.ReadHeader(buffer, offset) |
|---|
| 491 | |
|---|
| 492 | # header scanning detected error, this is no mpeg |
|---|
| 493 | if new_offset == -1: |
|---|
| 494 | return 0 |
|---|
| 495 | |
|---|
| 496 | if new_offset: |
|---|
| 497 | # we have a new offset |
|---|
| 498 | offset = new_offset |
|---|
| 499 | |
|---|
| 500 | # skip padding 0 before a new header |
|---|
| 501 | while len(buffer) > offset + 10 and \ |
|---|
| 502 | not ord(buffer[offset+2]): |
|---|
| 503 | offset += 1 |
|---|
| 504 | |
|---|
| 505 | else: |
|---|
| 506 | # seek to new header by brute force |
|---|
| 507 | offset += buffer[offset+4:].find('\x00\x00\x01') + 4 |
|---|
| 508 | |
|---|
| 509 | # fill in values for support functions: |
|---|
| 510 | self.__seek_size__ = 1000000 |
|---|
| 511 | self.__sample_size__ = 10000 |
|---|
| 512 | self.__search__ = self._find_timer_ |
|---|
| 513 | self.filename = file.name |
|---|
| 514 | |
|---|
| 515 | # get length of the file |
|---|
| 516 | self.length = self.get_length() |
|---|
| 517 | return 1 |
|---|
| 518 | |
|---|
| 519 | |
|---|
| 520 | def _find_timer_(self, buffer): |
|---|
| 521 | """ |
|---|
| 522 | Return position of timer in buffer or -1 if not found. |
|---|
| 523 | This function is valid for 'normal' mpeg files |
|---|
| 524 | """ |
|---|
| 525 | pos = buffer.find('\x00\x00\x01%s' % chr(PACK_PKT)) |
|---|
| 526 | if pos == -1: |
|---|
| 527 | return -1 |
|---|
| 528 | return pos + 4 |
|---|
| 529 | |
|---|
| 530 | |
|---|
| 531 | |
|---|
| 532 | # PES ============================================================ |
|---|
| 533 | |
|---|
| 534 | |
|---|
| 535 | def ReadPESHeader(self, offset, buffer, id=0): |
|---|
| 536 | """ |
|---|
| 537 | Parse a PES header. |
|---|
| 538 | Since it starts with 0x00 0x00 0x01 like 'normal' mpegs, this |
|---|
| 539 | function will return (0, -1) when it is no PES header or |
|---|
| 540 | (packet length, timestamp position (maybe -1)) |
|---|
| 541 | |
|---|
| 542 | http://dvd.sourceforge.net/dvdinfo/pes-hdr.html |
|---|
| 543 | """ |
|---|
| 544 | if not buffer[0:3] == '\x00\x00\x01': |
|---|
| 545 | return 0, -1 |
|---|
| 546 | |
|---|
| 547 | packet_length = (ord(buffer[4]) << 8) + ord(buffer[5]) + 6 |
|---|
| 548 | align = ord(buffer[6]) & 4 |
|---|
| 549 | header_length = ord(buffer[8]) |
|---|
| 550 | |
|---|
| 551 | # PES ID (starting with 001) |
|---|
| 552 | if ord(buffer[3]) & 0xE0 == 0xC0: |
|---|
| 553 | id = id or ord(buffer[3]) & 0x1F |
|---|
| 554 | for a in self.audio: |
|---|
| 555 | if a.id == id: |
|---|
| 556 | break |
|---|
| 557 | else: |
|---|
| 558 | self.audio.append(mediainfo.AudioInfo()) |
|---|
| 559 | self.audio[-1].id = id |
|---|
| 560 | self.audio[-1].keys.append('id') |
|---|
| 561 | |
|---|
| 562 | elif ord(buffer[3]) & 0xF0 == 0xE0: |
|---|
| 563 | id = id or ord(buffer[3]) & 0xF |
|---|
| 564 | for v in self.video: |
|---|
| 565 | if v.id == id: |
|---|
| 566 | break |
|---|
| 567 | else: |
|---|
| 568 | self.video.append(mediainfo.VideoInfo()) |
|---|
| 569 | self.video[-1].id = id |
|---|
| 570 | self.video[-1].keys.append('id') |
|---|
| 571 | |
|---|
| 572 | # new mpeg starting |
|---|
| 573 | if buffer[header_length+9:header_length+13] == \ |
|---|
| 574 | '\x00\x00\x01\xB3' and not self.sequence_header_offset: |
|---|
| 575 | # yes, remember offset for later use |
|---|
| 576 | self.sequence_header_offset = offset + header_length+9 |
|---|
| 577 | elif ord(buffer[3]) == 189 or ord(buffer[3]) == 191: |
|---|
| 578 | # private stream. we don't know, but maybe we can guess later |
|---|
| 579 | id = id or ord(buffer[3]) & 0xF |
|---|
| 580 | if align and buffer[header_length+9:header_length+11] == '\x0b\x77': |
|---|
| 581 | # AC3 stream |
|---|
| 582 | for a in self.audio: |
|---|
| 583 | if a.id == id: |
|---|
| 584 | break |
|---|
| 585 | else: |
|---|
| 586 | self.audio.append(mediainfo.AudioInfo()) |
|---|
| 587 | self.audio[-1].id = id |
|---|
| 588 | self.audio[-1].codec = 'AC3' |
|---|
| 589 | self.audio[-1].keys.append('id') |
|---|
| 590 | |
|---|
| 591 | else: |
|---|
| 592 | # unknown content |
|---|
| 593 | pass |
|---|
| 594 | |
|---|
| 595 | ptsdts = ord(buffer[7]) >> 6 |
|---|
| 596 | |
|---|
| 597 | if ptsdts and ptsdts == ord(buffer[9]) >> 4: |
|---|
| 598 | if ord(buffer[9]) >> 4 != ptsdts: |
|---|
| 599 | print 'WARNING: bad PTS/DTS, please contact us' |
|---|
| 600 | return packet_length, -1 |
|---|
| 601 | |
|---|
| 602 | # timestamp = self.ReadPTS(buffer[9:14]) |
|---|
| 603 | high = ((ord(buffer[9]) & 0xF) >> 1) |
|---|
| 604 | med = (ord(buffer[10]) << 7) + (ord(buffer[11]) >> 1) |
|---|
| 605 | low = (ord(buffer[12]) << 7) + (ord(buffer[13]) >> 1) |
|---|
| 606 | return packet_length, 9 |
|---|
| 607 | |
|---|
| 608 | return packet_length, -1 |
|---|
| 609 | |
|---|
| 610 | |
|---|
| 611 | |
|---|
| 612 | def isPES(self, file): |
|---|
| 613 | if mediainfo.DEBUG: |
|---|
| 614 | print 'trying mpeg-pes scan' |
|---|
| 615 | file.seek(0,0) |
|---|
| 616 | buffer = file.read(3) |
|---|
| 617 | |
|---|
| 618 | # header (also valid for all mpegs) |
|---|
| 619 | if not buffer == '\x00\x00\x01': |
|---|
| 620 | return 0 |
|---|
| 621 | |
|---|
| 622 | self.sequence_header_offset = 0 |
|---|
| 623 | buffer += file.read(10000) |
|---|
| 624 | |
|---|
| 625 | offset = 0 |
|---|
| 626 | while offset + 1000 < len(buffer): |
|---|
| 627 | pos, timestamp = self.ReadPESHeader(offset, buffer[offset:]) |
|---|
| 628 | if not pos: |
|---|
| 629 | return 0 |
|---|
| 630 | if timestamp != -1 and not hasattr(self, 'start'): |
|---|
| 631 | self.get_time = self.ReadPTS |
|---|
| 632 | self.start = self.get_time(buffer[offset+timestamp:offset+timestamp+5]) |
|---|
| 633 | if self.sequence_header_offset and hasattr(self, 'start'): |
|---|
| 634 | # we have all informations we need |
|---|
| 635 | break |
|---|
| 636 | |
|---|
| 637 | offset += pos |
|---|
| 638 | if offset + 1000 < len(buffer) and len(buffer) < 1000000 or 1: |
|---|
| 639 | # looks like a pes, read more |
|---|
| 640 | buffer += file.read(10000) |
|---|
| 641 | |
|---|
| 642 | if not self.video and not self.audio: |
|---|
| 643 | # no video and no audio? |
|---|
| 644 | return 0 |
|---|
| 645 | |
|---|
| 646 | self.type = 'MPEG-PES' |
|---|
| 647 | |
|---|
| 648 | # fill in values for support functions: |
|---|
| 649 | self.__seek_size__ = 10000000 # 10 MB |
|---|
| 650 | self.__sample_size__ = 500000 # 500 k scanning |
|---|
| 651 | self.__search__ = self._find_timer_PES_ |
|---|
| 652 | self.filename = file.name |
|---|
| 653 | |
|---|
| 654 | # get length of the file |
|---|
| 655 | self.length = self.get_length() |
|---|
| 656 | return 1 |
|---|
| 657 | |
|---|
| 658 | |
|---|
| 659 | def _find_timer_PES_(self, buffer): |
|---|
| 660 | """ |
|---|
| 661 | Return position of timer in buffer or -1 if not found. |
|---|
| 662 | This function is valid for PES files |
|---|
| 663 | """ |
|---|
| 664 | pos = buffer.find('\x00\x00\x01') |
|---|
| 665 | offset = 0 |
|---|
| 666 | if pos == -1 or offset + 1000 >= len(buffer): |
|---|
| 667 | return -1 |
|---|
| 668 | |
|---|
| 669 | retpos = -1 |
|---|
| 670 | ackcount = 0 |
|---|
| 671 | while offset + 1000 < len(buffer): |
|---|
| 672 | pos, timestamp = self.ReadPESHeader(offset, buffer[offset:]) |
|---|
| 673 | if timestamp != -1 and retpos == -1: |
|---|
| 674 | retpos = offset + timestamp |
|---|
| 675 | if pos == 0: |
|---|
| 676 | # Oops, that was a mpeg header, no PES header |
|---|
| 677 | offset += buffer[offset:].find('\x00\x00\x01') |
|---|
| 678 | retpos = -1 |
|---|
| 679 | ackcount = 0 |
|---|
| 680 | else: |
|---|
| 681 | offset += pos |
|---|
| 682 | if retpos != -1: |
|---|
| 683 | ackcount += 1 |
|---|
| 684 | if ackcount > 10: |
|---|
| 685 | # looks ok to me |
|---|
| 686 | return retpos |
|---|
| 687 | return -1 |
|---|
| 688 | |
|---|
| 689 | |
|---|
| 690 | # Transport Stream =============================================== |
|---|
| 691 | |
|---|
| 692 | def isTS(self, file): |
|---|
| 693 | file.seek(0,0) |
|---|
| 694 | |
|---|
| 695 | buffer = file.read(TS_PACKET_LENGTH * 2) |
|---|
| 696 | c = 0 |
|---|
| 697 | |
|---|
| 698 | while c + TS_PACKET_LENGTH < len(buffer): |
|---|
| 699 | if ord(buffer[c]) == ord(buffer[c+TS_PACKET_LENGTH]) == TS_SYNC: |
|---|
| 700 | break |
|---|
| 701 | c += 1 |
|---|
| 702 | else: |
|---|
| 703 | return 0 |
|---|
| 704 | |
|---|
| 705 | buffer += file.read(10000) |
|---|
| 706 | self.type = 'MPEG-TS' |
|---|
| 707 | |
|---|
| 708 | while c + TS_PACKET_LENGTH < len(buffer): |
|---|
| 709 | start = ord(buffer[c+1]) & 0x40 |
|---|
| 710 | # maybe load more into the buffer |
|---|
| 711 | if c + 2 * TS_PACKET_LENGTH > len(buffer) and c < 500000: |
|---|
| 712 | buffer += file.read(10000) |
|---|
| 713 | |
|---|
| 714 | # wait until the ts payload contains a payload header |
|---|
| 715 | if not start: |
|---|
| 716 | c += TS_PACKET_LENGTH |
|---|
| 717 | continue |
|---|
| 718 | |
|---|
| 719 | tsid = ((ord(buffer[c+1]) & 0x3F) << 8) + ord(buffer[c+2]) |
|---|
| 720 | adapt = (ord(buffer[c+3]) & 0x30) >> 4 |
|---|
| 721 | |
|---|
| 722 | offset = 4 |
|---|
| 723 | if adapt & 0x02: |
|---|
| 724 | # meta info present, skip it for now |
|---|
| 725 | adapt_len = ord(buffer[c+offset]) |
|---|
| 726 | offset += adapt_len + 1 |
|---|
| 727 | |
|---|
| 728 | if not ord(buffer[c+1]) & 0x40: |
|---|
| 729 | # no new pes or psi in stream payload starting |
|---|
| 730 | pass |
|---|
| 731 | elif adapt & 0x01: |
|---|
| 732 | # PES |
|---|
| 733 | timestamp = self.ReadPESHeader(c+offset, buffer[c+offset:], tsid)[1] |
|---|
| 734 | if timestamp != -1: |
|---|
| 735 | if not hasattr(self, 'start'): |
|---|
| 736 | self.get_time = self.ReadPTS |
|---|
| 737 | timestamp = c + offset + timestamp |
|---|
| 738 | self.start = self.get_time(buffer[timestamp:timestamp+5]) |
|---|
| 739 | elif not hasattr(self, 'audio_ok'): |
|---|
| 740 | timestamp = c + offset + timestamp |
|---|
| 741 | start = self.get_time(buffer[timestamp:timestamp+5]) |
|---|
| 742 | if abs(start - self.start) < 10: |
|---|
| 743 | # looks ok |
|---|
| 744 | self.audio_ok = True |
|---|
| 745 | else: |
|---|
| 746 | # timestamp broken |
|---|
| 747 | del self.start |
|---|
| 748 | if mediainfo.DEBUG: |
|---|
| 749 | print 'Timestamp error, correcting' |
|---|
| 750 | |
|---|
| 751 | if hasattr(self, 'start') and self.start and \ |
|---|
| 752 | self.sequence_header_offset and self.video and self.audio: |
|---|
| 753 | break |
|---|
| 754 | |
|---|
| 755 | c += TS_PACKET_LENGTH |
|---|
| 756 | |
|---|
| 757 | |
|---|
| 758 | if not self.sequence_header_offset: |
|---|
| 759 | return 0 |
|---|
| 760 | |
|---|
| 761 | if hasattr(self, 'start') and self.start: |
|---|
| 762 | self.keys.append('start') |
|---|
| 763 | |
|---|
| 764 | # fill in values for support functions: |
|---|
| 765 | self.__seek_size__ = 10000000 # 10 MB |
|---|
| 766 | self.__sample_size__ = 100000 # 100 k scanning |
|---|
| 767 | self.__search__ = self._find_timer_TS_ |
|---|
| 768 | self.filename = file.name |
|---|
| 769 | |
|---|
| 770 | # get length of the file |
|---|
| 771 | self.length = self.get_length() |
|---|
| 772 | return 1 |
|---|
| 773 | |
|---|
| 774 | |
|---|
| 775 | def _find_timer_TS_(self, buffer): |
|---|
| 776 | c = 0 |
|---|
| 777 | |
|---|
| 778 | while c + TS_PACKET_LENGTH < len(buffer): |
|---|
| 779 | if ord(buffer[c]) == ord(buffer[c+TS_PACKET_LENGTH]) == TS_SYNC: |
|---|
| 780 | break |
|---|
| 781 | c += 1 |
|---|
| 782 | else: |
|---|
| 783 | return -1 |
|---|
| 784 | |
|---|
| 785 | while c + TS_PACKET_LENGTH < len(buffer): |
|---|
| 786 | start = ord(buffer[c+1]) & 0x40 |
|---|
| 787 | if not start: |
|---|
| 788 | c += TS_PACKET_LENGTH |
|---|
| 789 | continue |
|---|
| 790 | |
|---|
| 791 | tsid = ((ord(buffer[c+1]) & 0x3F) << 8) + ord(buffer[c+2]) |
|---|
| 792 | adapt = (ord(buffer[c+3]) & 0x30) >> 4 |
|---|
| 793 | |
|---|
| 794 | offset = 4 |
|---|
| 795 | if adapt & 0x02: |
|---|
| 796 | # meta info present, skip it for now |
|---|
| 797 | offset += ord(buffer[c+offset]) + 1 |
|---|
| 798 | |
|---|
| 799 | if adapt & 0x01: |
|---|
| 800 | timestamp = self.ReadPESHeader(c+offset, buffer[c+offset:], tsid)[1] |
|---|
| 801 | return c + offset + timestamp |
|---|
| 802 | c += TS_PACKET_LENGTH |
|---|
| 803 | return -1 |
|---|
| 804 | |
|---|
| 805 | |
|---|
| 806 | |
|---|
| 807 | # Support functions ============================================== |
|---|
| 808 | |
|---|
| 809 | def get_endpos(self): |
|---|
| 810 | """ |
|---|
| 811 | get the last timestamp of the mpeg, return -1 if this is not possible |
|---|
| 812 | """ |
|---|
| 813 | if not hasattr(self, 'filename') or not hasattr(self, 'start'): |
|---|
| 814 | return -1 |
|---|
| 815 | |
|---|
| 816 | file = open(self.filename) |
|---|
| 817 | file.seek(os.stat(self.filename)[stat.ST_SIZE]-self.__sample_size__) |
|---|
| 818 | buffer = file.read(self.__sample_size__) |
|---|
| 819 | |
|---|
| 820 | end = -1 |
|---|
| 821 | while 1: |
|---|
| 822 | pos = self.__search__(buffer) |
|---|
| 823 | if pos == -1: |
|---|
| 824 | break |
|---|
| 825 | end = self.get_time(buffer[pos:]) |
|---|
| 826 | buffer = buffer[pos+100:] |
|---|
| 827 | |
|---|
| 828 | file.close() |
|---|
| 829 | return end |
|---|
| 830 | |
|---|
| 831 | |
|---|
| 832 | def get_length(self): |
|---|
| 833 | """ |
|---|
| 834 | get the length in seconds, return -1 if this is not possible |
|---|
| 835 | """ |
|---|
| 836 | end = self.get_endpos() |
|---|
| 837 | if end == -1: |
|---|
| 838 | return -1 |
|---|
| 839 | if self.start > end: |
|---|
| 840 | return int(((long(1) << 33) - 1 ) / 90000) - self.start + end |
|---|
| 841 | return end - self.start |
|---|
| 842 | |
|---|
| 843 | |
|---|
| 844 | def seek(self, end_time): |
|---|
| 845 | """ |
|---|
| 846 | Return the byte position in the file where the time position |
|---|
| 847 | is 'pos' seconds. Return 0 if this is not possible |
|---|
| 848 | """ |
|---|
| 849 | if not hasattr(self, 'filename') or not hasattr(self, 'start'): |
|---|
| 850 | return 0 |
|---|
| 851 | |
|---|
| 852 | file = open(self.filename) |
|---|
| 853 | seek_to = 0 |
|---|
| 854 | |
|---|
| 855 | while 1: |
|---|
| 856 | file.seek(self.__seek_size__, 1) |
|---|
| 857 | buffer = file.read(self.__sample_size__) |
|---|
| 858 | if len(buffer) < 10000: |
|---|
| 859 | break |
|---|
| 860 | pos = self.__search__(buffer) |
|---|
| 861 | if pos != -1: |
|---|
| 862 | # found something |
|---|
| 863 | if self.get_time(buffer[pos:]) >= end_time: |
|---|
| 864 | # too much, break |
|---|
| 865 | break |
|---|
| 866 | # that wasn't enough |
|---|
| 867 | seek_to = file.tell() |
|---|
| 868 | |
|---|
| 869 | file.close() |
|---|
| 870 | return seek_to |
|---|
| 871 | |
|---|
| 872 | |
|---|
| 873 | def __scan__(self): |
|---|
| 874 | """ |
|---|
| 875 | scan file for timestamps (may take a long time) |
|---|
| 876 | """ |
|---|
| 877 | if not hasattr(self, 'filename') or not hasattr(self, 'start'): |
|---|
| 878 | return 0 |
|---|
| 879 | file = open(self.filename) |
|---|
| 880 | print 'scanning file...' |
|---|
| 881 | while 1: |
|---|
| 882 | file.seek(self.__seek_size__ * 10, 1) |
|---|
| 883 | buffer = file.read(self.__sample_size__) |
|---|
| 884 | if len(buffer) < 10000: |
|---|
| 885 | break |
|---|
| 886 | pos = self.__search__(buffer) |
|---|
| 887 | if pos == -1: |
|---|
| 888 | continue |
|---|
| 889 | print self.get_time(buffer[pos:]) |
|---|
| 890 | |
|---|
| 891 | file.close() |
|---|
| 892 | print 'done' |
|---|
| 893 | print |
|---|
| 894 | |
|---|
| 895 | |
|---|
| 896 | |
|---|
| 897 | mmpython.registertype( 'video/mpeg', ('mpeg','mpg','mp4', 'ts'), mediainfo.TYPE_AV, MpegInfo ) |
|---|
