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