apps/metadata/asf.c

   1 /***************************************************************************
   2  *             __________               __   ___.
   3  *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
   4  *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
   5  *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
   6  *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
   7  *                     \/            \/     \/    \/            \/
   8  *
   9  * $Id$
  10  *
  11  * Copyright (C) 2007 Dave Chapman
  12  *
  13  * This program is free software; you can redistribute it and/or
  14  * modify it under the terms of the GNU General Public License
  15  * as published by the Free Software Foundation; either version 2
  16  * of the License, or (at your option) any later version.
  17  *
  18  * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
  19  * KIND, either express or implied.
  20  *
  21  ****************************************************************************/
  22 #include <stdio.h>
  23 #include <string.h>
  24 #include <stdlib.h>
  25 #include <ctype.h>
  26 #include <inttypes.h>
  27
  28 #include "metadata.h"
  29 #include "replaygain.h"
  30 #include "debug.h"
  31 #include "rbunicode.h"
  32 #include "metadata_common.h"
  33 #include "metadata_parsers.h"
  34 #include "system.h"
  35 #include <codecs/libwma/asf.h>
  36
  37 /* TODO: Just read the GUIDs into a 16-byte array, and use memcmp to compare */
  38 struct guid_s {
  39     uint32_t v1;
  40     uint16_t v2;
  41     uint16_t v3;
  42     uint8_t  v4[8];
  43 };
  44 typedef struct guid_s guid_t;
  45
  46 struct asf_object_s {
  47     guid_t       guid;
  48     uint64_t     size;
  49     uint64_t     datalen;
  50 };
  51 typedef struct asf_object_s asf_object_t;
  52
  53 enum asf_error_e {
  54     ASF_ERROR_INTERNAL       = -1,  /* incorrect input to API calls */
  55     ASF_ERROR_OUTOFMEM       = -2,  /* some malloc inside program failed */
  56     ASF_ERROR_EOF            = -3,  /* unexpected end of file */
  57     ASF_ERROR_IO             = -4,  /* error reading or writing to file */
  58     ASF_ERROR_INVALID_LENGTH = -5,  /* length value conflict in input data */
  59     ASF_ERROR_INVALID_VALUE  = -6,  /* other value conflict in input data */
  60     ASF_ERROR_INVALID_OBJECT = -7,  /* ASF object missing or in wrong place */
  61     ASF_ERROR_OBJECT_SIZE    = -8,  /* invalid ASF object size (too small) */
  62     ASF_ERROR_SEEKABLE       = -9,  /* file not seekable */
  63     ASF_ERROR_SEEK           = -10, /* file is seekable but seeking failed */
  64     ASF_ERROR_ENCRYPTED      = -11  /* file is encrypted */
  65 };
  66
  67 static const guid_t asf_guid_null =
  68 {0x00000000, 0x0000, 0x0000, {0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00}};
  69
  70 /* top level object guids */
  71
  72 static const guid_t asf_guid_header =
  73 {0x75B22630, 0x668E, 0x11CF, {0xA6, 0xD9, 0x00, 0xAA, 0x00, 0x62, 0xCE, 0x6C}};
  74
  75 static const guid_t asf_guid_data =
  76 {0x75B22636, 0x668E, 0x11CF, {0xA6, 0xD9, 0x00, 0xAA, 0x00, 0x62, 0xCE, 0x6C}};
  77
  78 static const guid_t asf_guid_index =
  79 {0x33000890, 0xE5B1, 0x11CF, {0x89, 0xF4, 0x00, 0xA0, 0xC9, 0x03, 0x49, 0xCB}};
  80
  81 /* header level object guids */
  82
  83 static const guid_t asf_guid_file_properties =
  84 {0x8cabdca1, 0xa947, 0x11cf, {0x8E, 0xe4, 0x00, 0xC0, 0x0C, 0x20, 0x53, 0x65}};
  85
  86 static const guid_t asf_guid_stream_properties =
  87 {0xB7DC0791, 0xA9B7, 0x11CF, {0x8E, 0xE6, 0x00, 0xC0, 0x0C, 0x20, 0x53, 0x65}};
  88
  89 static const guid_t asf_guid_content_description =
  90 {0x75B22633, 0x668E, 0x11CF, {0xA6, 0xD9, 0x00, 0xAA, 0x00, 0x62, 0xCE, 0x6C}};
  91
  92 static const guid_t asf_guid_extended_content_description =
  93 {0xD2D0A440, 0xE307, 0x11D2, {0x97, 0xF0, 0x00, 0xA0, 0xC9, 0x5E, 0xA8, 0x50}};
  94
  95 static const guid_t asf_guid_content_encryption =
  96 {0x2211b3fb, 0xbd23, 0x11d2, {0xb4, 0xb7, 0x00, 0xa0, 0xc9, 0x55, 0xfc, 0x6e}};
  97
  98 static const guid_t asf_guid_extended_content_encryption =
  99 {0x298ae614, 0x2622, 0x4c17, {0xb9, 0x35, 0xda, 0xe0, 0x7e, 0xe9, 0x28, 0x9c}};
 100
 101 /* stream type guids */
 102
 103 static const guid_t asf_guid_stream_type_audio =
 104 {0xF8699E40, 0x5B4D, 0x11CF, {0xA8, 0xFD, 0x00, 0x80, 0x5F, 0x5C, 0x44, 0x2B}};
 105
 106 static int asf_guid_match(const guid_t *guid1, const guid_t *guid2)
 107 {
 108     if((guid1->v1 != guid2->v1) ||
 109        (guid1->v2 != guid2->v2) ||
 110        (guid1->v3 != guid2->v3) ||
 111        (memcmp(guid1->v4, guid2->v4, 8))) {
 112         return 0;
 113     }
 114
 115     return 1;
 116 }
 117
 118 /* Read the 16 byte GUID from a file */
 119 static void asf_readGUID(int fd, guid_t* guid)
 120 {
 121     read_uint32le(fd, &guid->v1);
 122     read_uint16le(fd, &guid->v2);
 123     read_uint16le(fd, &guid->v3);
 124     read(fd, guid->v4, 8);
 125 }
 126
 127 static void asf_read_object_header(asf_object_t *obj, int fd)
 128 {
 129     asf_readGUID(fd, &obj->guid);
 130     read_uint64le(fd, &obj->size);
 131     obj->datalen = 0;
 132 }
 133
 134 /* Parse an integer from the extended content object - we always
 135    convert to an int, regardless of native format.
 136 */
 137 static int asf_intdecode(int fd, int type, int length)
 138 {
 139     uint16_t tmp16;
 140     uint32_t tmp32;
 141     uint64_t tmp64;
 142
 143     if (type==3) {
 144         read_uint32le(fd, &tmp32);
 145         lseek(fd,length - 4,SEEK_CUR);
 146         return (int)tmp32;
 147     } else if (type==4) {
 148         read_uint64le(fd, &tmp64);
 149         lseek(fd,length - 8,SEEK_CUR);
 150         return (int)tmp64;
 151     } else if (type == 5) {
 152         read_uint16le(fd, &tmp16);
 153         lseek(fd,length - 2,SEEK_CUR);
 154         return (int)tmp16;
 155     }
 156
 157     return 0;
 158 }
 159
 160 /* Decode a LE utf16 string from a disk buffer into a fixed-sized
 161    utf8 buffer.
 162 */
 163
 164 static void asf_utf16LEdecode(int fd,
 165                               uint16_t utf16bytes,
 166                               unsigned char **utf8,
 167                               int* utf8bytes
 168                              )
 169 {
 170     unsigned long ucs;
 171     int n;
 172     unsigned char utf16buf[256];
 173     unsigned char* utf16 = utf16buf;
 174     unsigned char* newutf8;
 175
 176     n = read(fd, utf16buf, MIN(sizeof(utf16buf), utf16bytes));
 177     utf16bytes -= n;
 178
 179     while (n > 0) {
 180         /* Check for a surrogate pair */
 181         if (utf16[1] >= 0xD8 && utf16[1] < 0xE0) {
 182             if (n < 4) {
 183                 /* Run out of utf16 bytes, read some more */
 184                 utf16buf[0] = utf16[0];
 185                 utf16buf[1] = utf16[1];
 186
 187                 n = read(fd, utf16buf + 2, MIN(sizeof(utf16buf)-2, utf16bytes));
 188                 utf16 = utf16buf;
 189                 utf16bytes -= n;
 190                 n += 2;
 191             }
 192
 193             if (n < 4) {
 194                 /* Truncated utf16 string, abort */
 195                 break;
 196             }
 197             ucs = 0x10000 + ((utf16[0] << 10) | ((utf16[1] - 0xD8) << 18)
 198                              | utf16[2] | ((utf16[3] - 0xDC) << 8));
 199             utf16 += 4;
 200             n -= 4;
 201         } else {
 202             ucs = (utf16[0] | (utf16[1] << 8));
 203             utf16 += 2;
 204             n -= 2;
 205         }
 206
 207         if (*utf8bytes > 6) {
 208             newutf8 = utf8encode(ucs, *utf8);
 209             *utf8bytes -= (newutf8 - *utf8);
 210             *utf8 += (newutf8 - *utf8);
 211         }
 212
 213         /* We have run out of utf16 bytes, read more if available */
 214         if ((n == 0) && (utf16bytes > 0)) {
 215             n = read(fd, utf16buf, MIN(sizeof(utf16buf), utf16bytes));
 216             utf16 = utf16buf;
 217             utf16bytes -= n;
 218         }
 219     }
 220
 221     *utf8[0] = 0;
 222     --*utf8bytes;
 223
 224     if (utf16bytes > 0) {
 225         /* Skip any remaining bytes */
 226         lseek(fd, utf16bytes, SEEK_CUR);
 227     }
 228     return;
 229 }
 230
 231 static int asf_parse_header(int fd, struct mp3entry* id3,
 232                                     asf_waveformatex_t* wfx)
 233 {
 234     asf_object_t current;
 235     asf_object_t header;
 236     uint64_t datalen;
 237     int i;
 238     int fileprop = 0;
 239     uint64_t play_duration;
 240     uint16_t flags;
 241     uint32_t subobjects;
 242     uint8_t utf8buf[512];
 243     int id3buf_remaining = sizeof(id3->id3v2buf) + sizeof(id3->id3v1buf);
 244     unsigned char* id3buf = (unsigned char*)id3->id3v2buf;
 245
 246     asf_read_object_header((asf_object_t *) &header, fd);
 247
 248     //DEBUGF("header.size=%d\n",(int)header.size);
 249     if (header.size < 30) {
 250         /* invalid size for header object */
 251         return ASF_ERROR_OBJECT_SIZE;
 252     }
 253
 254     read_uint32le(fd, &subobjects);
 255
 256     /* Two reserved bytes - do we need to read them? */
 257     lseek(fd, 2, SEEK_CUR);
 258
 259     //DEBUGF("Read header - size=%d, subobjects=%d\n",(int)header.size, (int)subobjects);
 260
 261     if (subobjects > 0) {
 262         header.datalen = header.size - 30;
 263
 264         /* TODO: Check that we have datalen bytes left in the file */
 265         datalen = header.datalen;
 266
 267         for (i=0; i<(int)subobjects; i++) {
 268             //DEBUGF("Parsing header object %d - datalen=%d\n",i,(int)datalen);
 269             if (datalen < 24) {
 270                 //DEBUGF("not enough data for reading object\n");
 271                 break;
 272             }
 273
 274             asf_read_object_header(&current, fd);
 275
 276             if (current.size > datalen || current.size < 24) {
 277                 //DEBUGF("invalid object size - current.size=%d, datalen=%d\n",(int)current.size,(int)datalen);
 278                 break;
 279             }
 280
 281             if (asf_guid_match(&current.guid, &asf_guid_file_properties)) {
 282                     if (current.size < 104)
 283                         return ASF_ERROR_OBJECT_SIZE;
 284
 285                     if (fileprop) {
 286                         /* multiple file properties objects not allowed */
 287                         return ASF_ERROR_INVALID_OBJECT;
 288                     }
 289
 290                     fileprop = 1;
 291                     /* All we want is the play duration - uint64_t at offset 40 */
 292                     lseek(fd, 40, SEEK_CUR);
 293
 294                     read_uint64le(fd, &play_duration);
 295                     id3->length = play_duration / 10000;
 296
 297                     //DEBUGF("****** length = %lums\n", id3->length);
 298
 299                     /* Read the packet size - uint32_t at offset 68 */
 300                     lseek(fd, 20, SEEK_CUR);
 301                     read_uint32le(fd, &wfx->packet_size);
 302
 303                     /* Skip bytes remaining in object */
 304                     lseek(fd, current.size - 24 - 72, SEEK_CUR);
 305             } else if (asf_guid_match(&current.guid, &asf_guid_stream_properties)) {
 306                     guid_t guid;
 307                     uint32_t propdatalen;
 308
 309                     if (current.size < 78)
 310                         return ASF_ERROR_OBJECT_SIZE;
 311
 312 #if 0
 313                     asf_byteio_getGUID(&guid, current->data);
 314                     datalen = asf_byteio_getDWLE(current->data + 40);
 315                     flags = asf_byteio_getWLE(current->data + 48);
 316 #endif
 317
 318                     asf_readGUID(fd, &guid);
 319
 320                     lseek(fd, 24, SEEK_CUR);
 321                     read_uint32le(fd, &propdatalen);
 322                     lseek(fd, 4, SEEK_CUR);
 323                     read_uint16le(fd, &flags);
 324
 325                     if (!asf_guid_match(&guid, &asf_guid_stream_type_audio)) {
 326                         //DEBUGF("Found stream properties for non audio stream, skipping\n");
 327                         lseek(fd,current.size - 24 - 50,SEEK_CUR);
 328                     } else if (wfx->audiostream == -1) {
 329                         lseek(fd, 4, SEEK_CUR);
 330                         //DEBUGF("Found stream properties for audio stream %d\n",flags&0x7f);
 331
 332                         if (propdatalen < 18) {
 333                             return ASF_ERROR_INVALID_LENGTH;
 334                         }
 335
 336 #if 0
 337                         if (asf_byteio_getWLE(data + 16) > datalen - 16) {
 338                             return ASF_ERROR_INVALID_LENGTH;
 339                         }
 340 #endif
 341                         read_uint16le(fd, &wfx->codec_id);
 342                         read_uint16le(fd, &wfx->channels);
 343                         read_uint32le(fd, &wfx->rate);
 344                         read_uint32le(fd, &wfx->bitrate);
 345                         wfx->bitrate *= 8;
 346                         read_uint16le(fd, &wfx->blockalign);
 347                         read_uint16le(fd, &wfx->bitspersample);
 348                         read_uint16le(fd, &wfx->datalen);
 349
 350                         /* Round bitrate to the nearest kbit */
 351                         id3->bitrate = (wfx->bitrate + 500) / 1000;
 352                         id3->frequency = wfx->rate;
 353
 354                         if (wfx->codec_id == ASF_CODEC_ID_WMAV1) {
 355                             read(fd, wfx->data, 4);
 356                             lseek(fd,current.size - 24 - 72 - 4,SEEK_CUR);
 357                             wfx->audiostream = flags&0x7f;
 358                         } else if (wfx->codec_id == ASF_CODEC_ID_WMAV2) {
 359                             read(fd, wfx->data, 6);
 360                             lseek(fd,current.size - 24 - 72 - 6,SEEK_CUR);
 361                             wfx->audiostream = flags&0x7f;
 362                         } else {
 363                             DEBUGF("Unsupported WMA codec (Pro, Lossless, Voice, etc)\n");
 364                             lseek(fd,current.size - 24 - 72,SEEK_CUR);
 365                         }
 366
 367                     }
 368             } else if (asf_guid_match(&current.guid, &asf_guid_content_description)) {
 369                     /* Object contains five 16-bit string lengths, followed by the five strings:
 370                        title, artist, copyright, description, rating
 371                      */
 372                     uint16_t strlength[5];
 373                     int i;
 374
 375                     //DEBUGF("Found GUID_CONTENT_DESCRIPTION - size=%d\n",(int)(current.size - 24));
 376
 377                     /* Read the 5 string lengths - number of bytes included trailing zero */
 378                     for (i=0; i<5; i++) {
 379                         read_uint16le(fd, &strlength[i]);
 380                         //DEBUGF("strlength = %u\n",strlength[i]);
 381                     }
 382
 383                     if (strlength[0] > 0) {  /* 0 - Title */
 384                         id3->title = id3buf;
 385                         asf_utf16LEdecode(fd, strlength[0], &id3buf, &id3buf_remaining);
 386                     }
 387
 388                     if (strlength[1] > 0) {  /* 1 - Artist */
 389                         id3->artist = id3buf;
 390                         asf_utf16LEdecode(fd, strlength[1], &id3buf, &id3buf_remaining);
 391                     }
 392
 393                     lseek(fd, strlength[2], SEEK_CUR); /* 2 - copyright */
 394
 395                     if (strlength[3] > 0) {  /* 3 - description */
 396                         id3->comment = id3buf;
 397                         asf_utf16LEdecode(fd, strlength[3], &id3buf, &id3buf_remaining);
 398                     }
 399
 400                     lseek(fd, strlength[4], SEEK_CUR); /* 4 - rating */
 401             } else if (asf_guid_match(&current.guid, &asf_guid_extended_content_description)) {
 402                     uint16_t count;
 403                     int i;
 404                     int bytesleft = current.size - 24;
 405                     //DEBUGF("Found GUID_EXTENDED_CONTENT_DESCRIPTION\n");
 406
 407                     read_uint16le(fd, &count);
 408                     bytesleft -= 2;
 409                     //DEBUGF("extended metadata count = %u\n",count);
 410
 411                     for (i=0; i < count; i++) {
 412                         uint16_t length, type;
 413                         unsigned char* utf8 = utf8buf;
 414                         int utf8length = 512;
 415
 416                         read_uint16le(fd, &length);
 417                         asf_utf16LEdecode(fd, length, &utf8, &utf8length);
 418                         bytesleft -= 2 + length;
 419
 420                         read_uint16le(fd, &type);
 421                         read_uint16le(fd, &length);
 422
 423                         if (!strcmp("WM/TrackNumber",utf8buf)) {
 424                             if (type == 0) {
 425                                 id3->track_string = id3buf;
 426                                 asf_utf16LEdecode(fd, length, &id3buf, &id3buf_remaining);
 427                                 id3->tracknum = atoi(id3->track_string);
 428                             } else if ((type >=2) && (type <= 5)) {
 429                                 id3->tracknum = asf_intdecode(fd, type, length);
 430                             } else {
 431                                 lseek(fd, length, SEEK_CUR);
 432                             }
 433                         } else if ((!strcmp("WM/Genre",utf8buf)) && (type == 0)) {
 434                             id3->genre_string = id3buf;
 435                             asf_utf16LEdecode(fd, length, &id3buf, &id3buf_remaining);
 436                         } else if ((!strcmp("WM/AlbumTitle",utf8buf)) && (type == 0)) {
 437                             id3->album = id3buf;
 438                             asf_utf16LEdecode(fd, length, &id3buf, &id3buf_remaining);
 439                         } else if ((!strcmp("WM/AlbumArtist",utf8buf)) && (type == 0)) {
 440                             id3->albumartist = id3buf;
 441                             asf_utf16LEdecode(fd, length, &id3buf, &id3buf_remaining);
 442                         } else if ((!strcmp("WM/Composer",utf8buf)) && (type == 0)) {
 443                             id3->composer = id3buf;
 444                             asf_utf16LEdecode(fd, length, &id3buf, &id3buf_remaining);
 445                         } else if (!strcmp("WM/Year",utf8buf)) {
 446                             if (type == 0) {
 447                                 id3->year_string = id3buf;
 448                                 asf_utf16LEdecode(fd, length, &id3buf, &id3buf_remaining);
 449                                 id3->year = atoi(id3->year_string);
 450                             } else if ((type >=2) && (type <= 5)) {
 451                                 id3->year = asf_intdecode(fd, type, length);
 452                             } else {
 453                                 lseek(fd, length, SEEK_CUR);
 454                             }
 455                         } else if (!strncmp("replaygain_", utf8buf, 11)) {
 456                             char* value = id3buf;
 457                             int buf_len = id3buf_remaining;
 458                             int len;
 459                             asf_utf16LEdecode(fd, length, &id3buf, &id3buf_remaining);
 460                             len = parse_replaygain(utf8buf, value, id3,
 461                                 value, buf_len);
 462
 463                             if (len == 0) {
 464                                 /* Don't need to keep the value */
 465                                 id3buf = value;
 466                                 id3buf_remaining = buf_len;
 467                             }
 468                         } else if (!strcmp("MusicBrainz/Track Id", utf8buf)) {
 469                             id3->mb_track_id = id3buf;
 470                             asf_utf16LEdecode(fd, length, &id3buf, &id3buf_remaining);
 471                         } else {
 472                             lseek(fd, length, SEEK_CUR);
 473                         }
 474                         bytesleft -= 4 + length;
 475                     }
 476
 477                     lseek(fd, bytesleft, SEEK_CUR);
 478             } else if (asf_guid_match(&current.guid, &asf_guid_content_encryption)
 479                 || asf_guid_match(&current.guid, &asf_guid_extended_content_encryption)) {
 480                 //DEBUGF("File is encrypted\n");
 481                 return ASF_ERROR_ENCRYPTED;
 482             } else {
 483                 //DEBUGF("Skipping %d bytes of object\n",(int)(current.size - 24));
 484                 lseek(fd,current.size - 24,SEEK_CUR);
 485             }
 486
 487             //DEBUGF("Parsed object - size = %d\n",(int)current.size);
 488             datalen -= current.size;
 489         }
 490
 491         if (i != (int)subobjects || datalen != 0) {
 492             //DEBUGF("header data doesn't match given subobject count\n");
 493             return ASF_ERROR_INVALID_VALUE;
 494         }
 495
 496         //DEBUGF("%d subobjects read successfully\n", i);
 497     }
 498
 499 #if 0
 500     tmp = asf_parse_header_validate(file, &header);
 501     if (tmp < 0) {
 502         /* header read ok but doesn't validate correctly */
 503         return tmp;
 504     }
 505 #endif
 506
 507     //DEBUGF("header validated correctly\n");
 508
 509     return 0;
 510 }
 511
 512 bool get_asf_metadata(int fd, struct mp3entry* id3)
 513 {
 514     int res;
 515     asf_object_t obj;
 516     asf_waveformatex_t wfx;
 517
 518     wfx.audiostream = -1;
 519
 520     res = asf_parse_header(fd, id3, &wfx);
 521
 522     if (res < 0) {
 523         DEBUGF("ASF: parsing error - %d\n",res);
 524         return false;
 525     }
 526
 527     if (wfx.audiostream == -1) {
 528         DEBUGF("ASF: No WMA streams found\n");
 529         return false;
 530     }
 531
 532     asf_read_object_header(&obj, fd);
 533
 534     if (!asf_guid_match(&obj.guid, &asf_guid_data)) {
 535         DEBUGF("ASF: No data object found\n");
 536         return false;
 537     }
 538
 539     /* Store the current file position - no need to parse the header
 540        again in the codec.  The +26 skips the rest of the data object
 541        header.
 542      */
 543     id3->first_frame_offset = lseek(fd, 0, SEEK_CUR) + 26;
 544     id3->filesize = filesize(fd);
 545     /* We copy the wfx struct to the MP3 TOC field in the id3 struct so
 546        the codec doesn't need to parse the header object again */
 547     memcpy(id3->toc, &wfx, sizeof(wfx));
 548
 549     return true;
 550 }