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