| 1 | /* |
| 2 | * copyright (c) 2008 Paul Kendall <paul@kcbbs.gen.nz> |
| 3 | * |
| 4 | * This file is part of FFmpeg. |
| 5 | * |
| 6 | * FFmpeg is free software; you can redistribute it and/or |
| 7 | * modify it under the terms of the GNU Lesser General Public |
| 8 | * License as published by the Free Software Foundation; either |
| 9 | * version 2.1 of the License, or (at your option) any later version. |
| 10 | * |
| 11 | * FFmpeg is distributed in the hope that it will be useful, |
| 12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 14 | * Lesser General Public License for more details. |
| 15 | * |
| 16 | * You should have received a copy of the GNU Lesser General Public |
| 17 | * License along with FFmpeg; if not, write to the Free Software |
| 18 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 19 | */ |
| 20 | |
| 21 | /** |
| 22 | * @file latmaac.c |
| 23 | * LATM wrapped AAC decoder |
| 24 | */ |
| 25 | |
| 26 | #include <stdio.h> |
| 27 | #include <stdlib.h> |
| 28 | #include <string.h> |
| 29 | #include <math.h> |
| 30 | #include <sys/types.h> |
| 31 | |
| 32 | #include "parser.h" |
| 33 | #include "bitstream.h" |
| 34 | #include "neaacdec.h" |
| 35 | |
| 36 | extern const int aac_sample_rates[16]; |
| 37 | extern const uint8_t aac_channels[8]; |
| 38 | |
| 39 | #define min(a,b) ((a)<(b) ? (a) : (b)) |
| 40 | |
| 41 | |
| 42 | /* |
| 43 | Note: This decoder filter is intended to decode LATM streams transferred |
| 44 | in MPEG transport streams which are only supposed to contain one program. |
| 45 | To do a more complex LATM demuxing a separate LATM demuxer should be used. |
| 46 | */ |
| 47 | |
| 48 | #define AAC_NONE 0 // mode not detected (or indicated in mediatype) |
| 49 | #define AAC_LATM 1 // LATM packets (ISO/IEC 14496-3 1.7.3 Multiplex layer) |
| 50 | |
| 51 | #define SYNC_LATM 0x2b7 // 11 bits |
| 52 | |
| 53 | #define MAX_SIZE 8*1024 |
| 54 | |
| 55 | typedef struct AACConfig |
| 56 | { |
| 57 | uint8_t extra[64]; // should be way enough |
| 58 | int extrasize; |
| 59 | |
| 60 | int audioObjectType; |
| 61 | int samplingFrequencyIndex; |
| 62 | int samplingFrequency; |
| 63 | int channelConfiguration; |
| 64 | int channels; |
| 65 | } AACConfig; |
| 66 | |
| 67 | typedef struct AACParser |
| 68 | { |
| 69 | AACConfig config; |
| 70 | uint8_t frameLengthType; |
| 71 | uint16_t muxSlotLengthBytes; |
| 72 | |
| 73 | uint8_t audio_mux_version; |
| 74 | uint8_t audio_mux_version_A; |
| 75 | int taraFullness; |
| 76 | uint8_t config_crc; |
| 77 | int64_t other_data_bits; |
| 78 | |
| 79 | int mode; |
| 80 | int offset; // byte offset in "buf" buffer |
| 81 | uint8_t buf[MAX_SIZE]; // allocated buffer |
| 82 | int count; // number of bytes written in buffer |
| 83 | } AACParser; |
| 84 | |
| 85 | typedef struct AACDecoder |
| 86 | { |
| 87 | AACParser *parser; |
| 88 | faacDecHandle aac_decoder; |
| 89 | int open; |
| 90 | uint32_t in_samplerate; |
| 91 | uint8_t in_channels; |
| 92 | } AACDecoder; |
| 93 | |
| 94 | typedef struct { |
| 95 | AACDecoder* decoder; |
| 96 | } FAACContext; |
| 97 | |
| 98 | static inline int64_t latm_get_value(GetBitContext *b) |
| 99 | { |
| 100 | uint8_t bytesForValue = get_bits(b, 2); |
| 101 | int64_t value = 0; |
| 102 | int i; |
| 103 | for (i=0; i<=bytesForValue; i++) { |
| 104 | value <<= 8; |
| 105 | value |= get_bits(b, 8); |
| 106 | } |
| 107 | return value; |
| 108 | } |
| 109 | |
| 110 | static void readGASpecificConfig(struct AACConfig *cfg, GetBitContext *b, PutBitContext *o) |
| 111 | { |
| 112 | int framelen_flag = get_bits(b, 1); |
| 113 | put_bits(o, 1, framelen_flag); |
| 114 | int dependsOnCoder = get_bits(b, 1); |
| 115 | put_bits(o, 1, dependsOnCoder); |
| 116 | int ext_flag; |
| 117 | int delay; |
| 118 | int layerNr; |
| 119 | |
| 120 | if (dependsOnCoder) { |
| 121 | delay = get_bits(b, 14); |
| 122 | put_bits(o, 14, delay); |
| 123 | } |
| 124 | ext_flag = get_bits(b, 1); |
| 125 | put_bits(o, 1, ext_flag); |
| 126 | if (!cfg->channelConfiguration) { |
| 127 | // program config element |
| 128 | // TODO: |
| 129 | } |
| 130 | |
| 131 | if (cfg->audioObjectType == 6 || cfg->audioObjectType == 20) { |
| 132 | layerNr = get_bits(b, 3); |
| 133 | put_bits(o, 3, layerNr); |
| 134 | } |
| 135 | if (ext_flag) { |
| 136 | if (cfg->audioObjectType == 22) { |
| 137 | skip_bits(b, 5); // numOfSubFrame |
| 138 | skip_bits(b, 11); // layer_length |
| 139 | |
| 140 | put_bits(o, 16, 0); |
| 141 | } |
| 142 | if (cfg->audioObjectType == 17 || |
| 143 | cfg->audioObjectType == 19 || |
| 144 | cfg->audioObjectType == 20 || |
| 145 | cfg->audioObjectType == 23) { |
| 146 | |
| 147 | skip_bits(b, 3); // stuff |
| 148 | put_bits(o, 3, 0); |
| 149 | } |
| 150 | |
| 151 | skip_bits(b, 1); // extflag3 |
| 152 | put_bits(o, 1, 0); |
| 153 | } |
| 154 | } |
| 155 | |
| 156 | static int readAudioSpecificConfig(struct AACConfig *cfg, GetBitContext *b) |
| 157 | { |
| 158 | PutBitContext o; |
| 159 | init_put_bits(&o, cfg->extra, sizeof(cfg->extra)); |
| 160 | |
| 161 | // returns the number of bits read |
| 162 | int ret = 0; |
| 163 | int sbr_present = -1; |
| 164 | |
| 165 | // object |
| 166 | cfg->audioObjectType = get_bits(b, 5); |
| 167 | put_bits(&o, 5, cfg->audioObjectType); |
| 168 | if (cfg->audioObjectType == 31) { |
| 169 | uint8_t n = get_bits(b, 6); |
| 170 | put_bits(&o, 6, n); |
| 171 | cfg->audioObjectType = 32 + n; |
| 172 | } |
| 173 | |
| 174 | cfg->samplingFrequencyIndex = get_bits(b, 4); |
| 175 | cfg->samplingFrequency = aac_sample_rates[cfg->samplingFrequencyIndex]; |
| 176 | put_bits(&o, 4, cfg->samplingFrequencyIndex); |
| 177 | if (cfg->samplingFrequencyIndex == 0x0f) { |
| 178 | uint32_t f = get_bits_long(b, 24); |
| 179 | put_bits(&o, 24, f); |
| 180 | cfg->samplingFrequency = f; |
| 181 | } |
| 182 | cfg->channelConfiguration = get_bits(b, 4); |
| 183 | put_bits(&o, 4, cfg->channelConfiguration); |
| 184 | cfg->channels = aac_channels[cfg->channelConfiguration]; |
| 185 | |
| 186 | if (cfg->audioObjectType == 5) { |
| 187 | sbr_present = 1; |
| 188 | |
| 189 | // TODO: parsing !!!!!!!!!!!!!!!! |
| 190 | } |
| 191 | |
| 192 | switch (cfg->audioObjectType) { |
| 193 | case 1: |
| 194 | case 2: |
| 195 | case 3: |
| 196 | case 4: |
| 197 | case 6: |
| 198 | case 7: |
| 199 | case 17: |
| 200 | case 19: |
| 201 | case 20: |
| 202 | case 21: |
| 203 | case 22: |
| 204 | case 23: |
| 205 | readGASpecificConfig(cfg, b, &o); |
| 206 | break; |
| 207 | } |
| 208 | |
| 209 | if (sbr_present == -1) { |
| 210 | if (cfg->samplingFrequency <= 24000) { |
| 211 | cfg->samplingFrequency *= 2; |
| 212 | } |
| 213 | } |
| 214 | |
| 215 | // count the extradata |
| 216 | ret = put_bits_count(&o); |
| 217 | align_put_bits(&o); |
| 218 | flush_put_bits(&o); |
| 219 | cfg->extrasize = (ret + 7) >> 3; |
| 220 | return ret; |
| 221 | } |
| 222 | |
| 223 | static void readStreamMuxConfig(struct AACParser *parser, GetBitContext *b) |
| 224 | { |
| 225 | parser->audio_mux_version_A = 0; |
| 226 | parser->audio_mux_version = get_bits(b, 1); |
| 227 | if (parser->audio_mux_version == 1) { // audioMuxVersion |
| 228 | parser->audio_mux_version_A = get_bits(b, 1); |
| 229 | } |
| 230 | |
| 231 | if (parser->audio_mux_version_A == 0) { |
| 232 | if (parser->audio_mux_version == 1) { |
| 233 | parser->taraFullness = latm_get_value(b); |
| 234 | } |
| 235 | get_bits(b, 1); // allStreamSameTimeFraming = 1 |
| 236 | get_bits(b, 6); // numSubFrames = 0 |
| 237 | get_bits(b, 4); // numPrograms = 0 |
| 238 | |
| 239 | // for each program |
| 240 | get_bits(b, 3); // numLayer = 0 |
| 241 | |
| 242 | // for each layer |
| 243 | if (parser->audio_mux_version == 0) { |
| 244 | // audio specific config. |
| 245 | readAudioSpecificConfig(&parser->config, b); |
| 246 | } else { |
| 247 | int ascLen = latm_get_value(b); |
| 248 | ascLen -= readAudioSpecificConfig(&parser->config, b); |
| 249 | |
| 250 | // fill bits |
| 251 | while (ascLen > 16) { |
| 252 | skip_bits(b, 16); |
| 253 | ascLen -= 16; |
| 254 | } |
| 255 | skip_bits(b, ascLen); |
| 256 | } |
| 257 | |
| 258 | // these are not needed... perhaps |
| 259 | int frame_length_type = get_bits(b, 3); |
| 260 | parser->frameLengthType = frame_length_type; |
| 261 | if (frame_length_type == 0) { |
| 262 | get_bits(b, 8); |
| 263 | } else if (frame_length_type == 1) { |
| 264 | get_bits(b, 9); |
| 265 | } else if (frame_length_type == 3 || |
| 266 | frame_length_type == 4 || |
| 267 | frame_length_type == 5) { |
| 268 | int celp_table_index = get_bits(b, 6); |
| 269 | } else if (frame_length_type == 6 || |
| 270 | frame_length_type == 7) { |
| 271 | int hvxc_table_index = get_bits(b, 1); |
| 272 | } |
| 273 | |
| 274 | // other data |
| 275 | parser->other_data_bits = 0; |
| 276 | if (get_bits(b, 1)) { |
| 277 | // other data present |
| 278 | if (parser->audio_mux_version == 1) { |
| 279 | parser->other_data_bits = latm_get_value(b); |
| 280 | } else { |
| 281 | // other data not present |
| 282 | parser->other_data_bits = 0; |
| 283 | int esc, tmp; |
| 284 | do { |
| 285 | parser->other_data_bits <<= 8; |
| 286 | esc = get_bits(b, 1); |
| 287 | tmp = get_bits(b, 8); |
| 288 | parser->other_data_bits |= tmp; |
| 289 | } while (esc); |
| 290 | } |
| 291 | } |
| 292 | |
| 293 | // CRC |
| 294 | if (get_bits(b, 1)) { |
| 295 | parser->config_crc = get_bits(b, 8); |
| 296 | } |
| 297 | } else { |
| 298 | // tbd |
| 299 | } |
| 300 | } |
| 301 | |
| 302 | static void readPayloadLengthInfo(struct AACParser *parser, GetBitContext *b) |
| 303 | { |
| 304 | uint8_t tmp; |
| 305 | if (parser->frameLengthType == 0) { |
| 306 | parser->muxSlotLengthBytes = 0; |
| 307 | do { |
| 308 | tmp = get_bits(b, 8); |
| 309 | parser->muxSlotLengthBytes += tmp; |
| 310 | } while (tmp == 255); |
| 311 | } else { |
| 312 | if (parser->frameLengthType == 5 || |
| 313 | parser->frameLengthType == 7 || |
| 314 | parser->frameLengthType == 3) { |
| 315 | get_bits(b, 2); |
| 316 | } |
| 317 | } |
| 318 | } |
| 319 | |
| 320 | static void readAudioMuxElement(struct AACParser *parser, GetBitContext *b, uint8_t *payload, int *payloadsize) |
| 321 | { |
| 322 | uint8_t use_same_mux = get_bits(b, 1); |
| 323 | if (!use_same_mux) { |
| 324 | readStreamMuxConfig(parser, b); |
| 325 | } |
| 326 | |
| 327 | if (parser->audio_mux_version_A == 0) { |
| 328 | int j; |
| 329 | |
| 330 | readPayloadLengthInfo(parser, b); |
| 331 | |
| 332 | // copy data |
| 333 | for (j=0; j<parser->muxSlotLengthBytes; j++) { |
| 334 | *payload++ = get_bits(b, 8); |
| 335 | } |
| 336 | *payloadsize = parser->muxSlotLengthBytes; |
| 337 | |
| 338 | // ignore otherdata |
| 339 | } else { |
| 340 | // TBD |
| 341 | } |
| 342 | } |
| 343 | |
| 344 | static int readAudioSyncStream(struct AACParser *parser, GetBitContext *b, int size, uint8_t *payload, int *payloadsize) |
| 345 | { |
| 346 | // ISO/IEC 14496-3 Table 1.28 - Syntax of AudioMuxElement() |
| 347 | if (get_bits(b, 11) != 0x2b7) return -1; // not LATM |
| 348 | int muxlength = get_bits(b, 13); |
| 349 | |
| 350 | if (3+muxlength > size) return 0; // not enough data |
| 351 | |
| 352 | readAudioMuxElement(parser, b, payload, payloadsize); |
| 353 | |
| 354 | // we don't parse anything else here... |
| 355 | return (3+muxlength); |
| 356 | } |
| 357 | |
| 358 | |
| 359 | static void flush_buf(struct AACParser *parser, int offset) { |
| 360 | int bytes_to_flush = min(parser->count, offset); |
| 361 | int left = (parser->count - bytes_to_flush); |
| 362 | |
| 363 | if (bytes_to_flush > 0) { |
| 364 | if (left > 0) { |
| 365 | memcpy(parser->buf, parser->buf+bytes_to_flush, left); |
| 366 | parser->count = left; |
| 367 | } else { |
| 368 | parser->count = 0; |
| 369 | } |
| 370 | } |
| 371 | } |
| 372 | |
| 373 | static struct AACParser *latm_create_parser() |
| 374 | { |
| 375 | struct AACParser *parser = (struct AACParser *)av_malloc(sizeof(struct AACParser)); |
| 376 | memset(parser, 0, sizeof(struct AACParser)); |
| 377 | return parser; |
| 378 | } |
| 379 | |
| 380 | static void latm_destroy_parser(struct AACParser *parser) |
| 381 | { |
| 382 | av_free(parser); |
| 383 | } |
| 384 | |
| 385 | static void latm_flush(struct AACParser *parser) |
| 386 | { |
| 387 | parser->offset = 0; |
| 388 | parser->count = 0; |
| 389 | } |
| 390 | |
| 391 | static void latm_write_data(struct AACParser *parser, uint8_t *data, int len) |
| 392 | { |
| 393 | // buffer overflow check... just ignore the data before |
| 394 | if (parser->count + len > MAX_SIZE) { |
| 395 | flush_buf(parser, parser->offset); |
| 396 | parser->offset = 0; |
| 397 | if (parser->count + len > MAX_SIZE) { |
| 398 | int to_flush = (parser->count+len) - MAX_SIZE; |
| 399 | flush_buf(parser, to_flush); |
| 400 | } |
| 401 | } |
| 402 | |
| 403 | // append data |
| 404 | memcpy(parser->buf+parser->count, data, len); |
| 405 | parser->count += len; |
| 406 | } |
| 407 | |
| 408 | static int latm_parse_packet(struct AACParser *parser, uint8_t *data, int maxsize) |
| 409 | { |
| 410 | /* |
| 411 | Return value is either number of bytes parsed or |
| 412 | -1 when failed. |
| 413 | 0 = need more data. |
| 414 | */ |
| 415 | |
| 416 | uint8_t *start = parser->buf + parser->offset; |
| 417 | int bytes = parser->count - parser->offset; |
| 418 | GetBitContext b; |
| 419 | init_get_bits(&b, start, bytes * 8); |
| 420 | |
| 421 | if (parser->mode == AAC_LATM) { |
| 422 | int outsize = 0; |
| 423 | int ret = readAudioSyncStream(parser, &b, bytes, data, &outsize); |
| 424 | |
| 425 | if (ret < 0) return -1; |
| 426 | if (ret == 0) return 0; |
| 427 | |
| 428 | // update the offset |
| 429 | parser->offset += ret; |
| 430 | return outsize; |
| 431 | } |
| 432 | |
| 433 | // check for syncwords |
| 434 | while (bytes > 2) { |
| 435 | if (show_bits(&b, 11) == SYNC_LATM) { |
| 436 | // we must parse config first... |
| 437 | int outsize = 0; |
| 438 | |
| 439 | // check if there is a complete packet available... |
| 440 | int ret = readAudioSyncStream(parser, &b, bytes, data, &outsize); |
| 441 | if (ret < 0) return -1; |
| 442 | if (ret == 0) return 0; |
| 443 | parser->offset += ret; |
| 444 | |
| 445 | parser->mode = AAC_LATM; |
| 446 | return outsize; |
| 447 | } |
| 448 | skip_bits(&b, 8); |
| 449 | parser->offset++; |
| 450 | bytes--; |
| 451 | } |
| 452 | return 0; |
| 453 | } |
| 454 | |
| 455 | static void aac_filter_close(AACDecoder *decoder) |
| 456 | { |
| 457 | if (decoder->aac_decoder) { |
| 458 | NeAACDecClose(decoder->aac_decoder); |
| 459 | decoder->aac_decoder = NULL; |
| 460 | } |
| 461 | decoder->open = 0; |
| 462 | } |
| 463 | |
| 464 | static int aac_decoder_open(AACDecoder *decoder) |
| 465 | { |
| 466 | if (decoder->aac_decoder) return 0; |
| 467 | |
| 468 | decoder->aac_decoder = NeAACDecOpen(); |
| 469 | if (!decoder->aac_decoder) return -1; |
| 470 | |
| 471 | // are we going to initialize from decoder specific info ? |
| 472 | if (decoder->parser->config.extrasize > 0) { |
| 473 | char ret = NeAACDecInit2(decoder->aac_decoder, (unsigned char*)decoder->parser->config.extra, decoder->parser->config.extrasize, &decoder->in_samplerate, &decoder->in_channels); |
| 474 | if (ret < 0) { |
| 475 | aac_filter_close(decoder); // gone wrong ? |
| 476 | return -1; |
| 477 | } |
| 478 | decoder->open = 1; |
| 479 | } else { |
| 480 | // we'll open the decoder later... |
| 481 | decoder->open = 0; |
| 482 | } |
| 483 | return 0; |
| 484 | } |
| 485 | |
| 486 | AACDecoder *aac_filter_create() |
| 487 | { |
| 488 | AACDecoder *decoder = (AACDecoder *)av_malloc(sizeof(AACDecoder)); |
| 489 | decoder->parser = latm_create_parser(); |
| 490 | decoder->aac_decoder = NULL; |
| 491 | decoder->open = 0; |
| 492 | return (void *)decoder; |
| 493 | } |
| 494 | |
| 495 | void aac_filter_destroy(AACDecoder *decoder) |
| 496 | { |
| 497 | aac_filter_close(decoder); |
| 498 | latm_destroy_parser(decoder->parser); |
| 499 | av_free(decoder); |
| 500 | } |
| 501 | |
| 502 | int aac_filter_receive(AACDecoder *decoder, void *out, int *out_size, uint8_t *data, int size) |
| 503 | { |
| 504 | uint8_t tempbuf[32*1024]; |
| 505 | int ret; |
| 506 | int consumed = size; |
| 507 | int decoded; |
| 508 | int max_size = *out_size; |
| 509 | |
| 510 | *out_size = 0; |
| 511 | |
| 512 | //------------------------------------------------------------------------- |
| 513 | // Multiplex Parsing |
| 514 | //------------------------------------------------------------------------- |
| 515 | |
| 516 | latm_write_data(decoder->parser, data, size); |
| 517 | |
| 518 | do { |
| 519 | ret = latm_parse_packet(decoder->parser, tempbuf, sizeof(tempbuf)); |
| 520 | if (ret < 0) { |
| 521 | latm_flush(decoder->parser); |
| 522 | return consumed; |
| 523 | } |
| 524 | if (ret == 0) return consumed; |
| 525 | |
| 526 | data = tempbuf; |
| 527 | size = ret; |
| 528 | |
| 529 | //------------------------------------------------------------------------- |
| 530 | // Initialize decoder (if necessary) |
| 531 | //------------------------------------------------------------------------- |
| 532 | if (!decoder->open) { |
| 533 | aac_filter_close(decoder); |
| 534 | if (decoder->parser->mode == AAC_LATM) { |
| 535 | ret = aac_decoder_open(decoder); |
| 536 | if (ret < 0) return consumed; |
| 537 | } |
| 538 | |
| 539 | if(!decoder->open) return consumed; |
| 540 | } |
| 541 | |
| 542 | //------------------------------------------------------------------------- |
| 543 | // Decode samples |
| 544 | //------------------------------------------------------------------------- |
| 545 | NeAACDecFrameInfo info; |
| 546 | void *buf = NeAACDecDecode(decoder->aac_decoder, &info, data, size); |
| 547 | |
| 548 | if (buf) { |
| 549 | decoder->in_samplerate = info.samplerate; |
| 550 | decoder->in_channels = info.channels; |
| 551 | |
| 552 | //--------------------------------------------------------------------- |
| 553 | // Deliver decoded samples |
| 554 | //--------------------------------------------------------------------- |
| 555 | |
| 556 | // kram dekoduje 16-bit. my vypustame 16-bit. takze by to malo byt okej |
| 557 | decoded = info.samples * sizeof(short); |
| 558 | |
| 559 | // napraskame tam sample |
| 560 | *out_size += decoded; |
| 561 | if(*out_size > max_size) { |
| 562 | av_log(NULL, AV_LOG_ERROR, "overflow!\n"); |
| 563 | } else { |
| 564 | memcpy(out, buf, decoded); |
| 565 | out = (unsigned char *)out + decoded; |
| 566 | } |
| 567 | } else { |
| 568 | // need more data |
| 569 | break; |
| 570 | } |
| 571 | |
| 572 | } while (1); // decode all packets |
| 573 | return consumed; |
| 574 | } |
| 575 | |
| 576 | void aac_filter_getinfo(AACDecoder *decoder, int *sample_rate, int *channels) |
| 577 | { |
| 578 | if(!decoder->open) return; |
| 579 | *sample_rate = decoder->in_samplerate; |
| 580 | *channels = decoder->in_channels; |
| 581 | } |
| 582 | |
| 583 | static int faac_decode_init(AVCodecContext *avctx) |
| 584 | { |
| 585 | FAACContext *s = avctx->priv_data; |
| 586 | avctx->frame_size = 360; |
| 587 | avctx->sample_rate = 48000; |
| 588 | avctx->channels = 2; |
| 589 | avctx->bit_rate = 8192 * 8 * avctx->sample_rate / avctx->frame_size; |
| 590 | s->decoder = aac_filter_create(); |
| 591 | return 0; |
| 592 | } |
| 593 | |
| 594 | static int faac_decode_frame(AVCodecContext *avctx, |
| 595 | void *data, int *data_size, |
| 596 | uint8_t *buf, int buf_size) |
| 597 | { |
| 598 | FAACContext *s = avctx->priv_data; |
| 599 | int ret; |
| 600 | |
| 601 | if (s->decoder == NULL) faac_decode_init(avctx); |
| 602 | ret = aac_filter_receive(s->decoder, data, data_size, buf, buf_size); |
| 603 | aac_filter_getinfo(s->decoder, &(avctx->sample_rate), &(avctx->channels)); |
| 604 | return ret; |
| 605 | } |
| 606 | |
| 607 | static int faac_decode_end(AVCodecContext *avctx) |
| 608 | { |
| 609 | FAACContext *s = avctx->priv_data; |
| 610 | if(s->decoder != NULL) { |
| 611 | aac_filter_destroy(s->decoder); |
| 612 | } |
| 613 | return 0; |
| 614 | } |
| 615 | |
| 616 | |
| 617 | |
| 618 | #define LATM_HEADER 0x56e000 // 0x2b7 (11 bits) |
| 619 | #define LATM_MASK 0xFFE000 // top 11 bits |
| 620 | #define LATM_SIZE_MASK 0x001FFF // bottom 13 bits |
| 621 | |
| 622 | typedef struct LATMParseContext{ |
| 623 | ParseContext pc; |
| 624 | int count; |
| 625 | } LATMParseContext; |
| 626 | |
| 627 | /** |
| 628 | * finds the end of the current frame in the bitstream. |
| 629 | * @return the position of the first byte of the next frame, or -1 |
| 630 | */ |
| 631 | static int latm_find_frame_end(AVCodecParserContext *s1, const uint8_t *buf, |
| 632 | int buf_size) { |
| 633 | LATMParseContext *s = s1->priv_data; |
| 634 | ParseContext *pc = &s->pc; |
| 635 | int pic_found, i; |
| 636 | uint32_t state; |
| 637 | |
| 638 | pic_found = pc->frame_start_found; |
| 639 | state = pc->state; |
| 640 | |
| 641 | i = 0; |
| 642 | if(!pic_found){ |
| 643 | for(i=0; i<buf_size; i++){ |
| 644 | state = (state<<8) | buf[i]; |
| 645 | if((state & LATM_MASK) == LATM_HEADER){ |
| 646 | i++; |
| 647 | s->count = - i; |
| 648 | pic_found=1; |
| 649 | break; |
| 650 | } |
| 651 | } |
| 652 | } |
| 653 | |
| 654 | if(pic_found){ |
| 655 | /* EOF considered as end of frame */ |
| 656 | if (buf_size == 0) |
| 657 | return 0; |
| 658 | if((state & LATM_SIZE_MASK) - s->count <= buf_size) { |
| 659 | pc->frame_start_found = 0; |
| 660 | pc->state = -1; |
| 661 | return (state & LATM_SIZE_MASK) - s->count; |
| 662 | } |
| 663 | } |
| 664 | |
| 665 | s->count += buf_size; |
| 666 | pc->frame_start_found = pic_found; |
| 667 | pc->state = state; |
| 668 | return END_NOT_FOUND; |
| 669 | } |
| 670 | |
| 671 | static int latm_parse(AVCodecParserContext *s1, |
| 672 | AVCodecContext *avctx, |
| 673 | const uint8_t **poutbuf, int *poutbuf_size, |
| 674 | const uint8_t *buf, int buf_size) |
| 675 | { |
| 676 | LATMParseContext *s = s1->priv_data; |
| 677 | ParseContext *pc = &s->pc; |
| 678 | int next; |
| 679 | |
| 680 | if(s1->flags & PARSER_FLAG_COMPLETE_FRAMES){ |
| 681 | next = buf_size; |
| 682 | }else{ |
| 683 | next = latm_find_frame_end(s1, buf, buf_size); |
| 684 | |
| 685 | if (ff_combine_frame(pc, next, &buf, &buf_size) < 0) { |
| 686 | *poutbuf = NULL; |
| 687 | *poutbuf_size = 0; |
| 688 | return buf_size; |
| 689 | } |
| 690 | } |
| 691 | *poutbuf = buf; |
| 692 | *poutbuf_size = buf_size; |
| 693 | return next; |
| 694 | } |
| 695 | |
| 696 | static int latm_split(AVCodecContext *avctx, |
| 697 | const uint8_t *buf, int buf_size) |
| 698 | { |
| 699 | int i; |
| 700 | uint32_t state= -1; |
| 701 | |
| 702 | for(i=0; i<buf_size; i++){ |
| 703 | state= (state<<8) | buf[i]; |
| 704 | if((state & LATM_MASK) == LATM_HEADER) |
| 705 | return i-2; |
| 706 | } |
| 707 | return 0; |
| 708 | } |
| 709 | |
| 710 | AVCodec libfaad2_decoder = { |
| 711 | .name = "AAC/LATM", |
| 712 | .type = CODEC_TYPE_AUDIO, |
| 713 | .id = CODEC_ID_AAC_LATM, |
| 714 | .priv_data_size = sizeof (FAACContext), |
| 715 | .init = faac_decode_init, |
| 716 | .close = faac_decode_end, |
| 717 | .decode = faac_decode_frame, |
| 718 | }; |
| 719 | |
| 720 | AVCodecParser latm_parser = { |
| 721 | { CODEC_ID_AAC_LATM }, |
| 722 | sizeof(LATMParseContext), |
| 723 | NULL, |
| 724 | latm_parse, |
| 725 | ff_parse_close, |
| 726 | latm_split, |
| 727 | }; |