| 1 | /* |
| 2 | * Copyright (c) 2012 Justin Ruggles |
| 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 |
| 23 | * Vorbis audio parser |
| 24 | * |
| 25 | * Determines the duration for each packet. |
| 26 | */ |
| 27 | |
| 28 | #include "libavutil/log.h" |
| 29 | |
| 30 | #include "get_bits.h" |
| 31 | #include "parser.h" |
| 32 | #include "xiph.h" |
| 33 | #include "vorbis_parser_internal.h" |
| 34 | |
| 35 | static const AVClass vorbis_parser_class = { |
| 36 | .class_name = "Vorbis parser", |
| 37 | .item_name = av_default_item_name, |
| 38 | .version = LIBAVUTIL_VERSION_INT, |
| 39 | }; |
| 40 | |
| 41 | static int parse_id_header(AVVorbisParseContext *s, |
| 42 | const uint8_t *buf, int buf_size) |
| 43 | { |
| 44 | /* Id header should be 30 bytes */ |
| 45 | if (buf_size < 30) { |
| 46 | av_log(s, AV_LOG_ERROR, "Id header is too short\n"); |
| 47 | return AVERROR_INVALIDDATA; |
| 48 | } |
| 49 | |
| 50 | /* make sure this is the Id header */ |
| 51 | if (buf[0] != 1) { |
| 52 | av_log(s, AV_LOG_ERROR, "Wrong packet type in Id header\n"); |
| 53 | return AVERROR_INVALIDDATA; |
| 54 | } |
| 55 | |
| 56 | /* check for header signature */ |
| 57 | if (memcmp(&buf[1], "vorbis", 6)) { |
| 58 | av_log(s, AV_LOG_ERROR, "Invalid packet signature in Id header\n"); |
| 59 | return AVERROR_INVALIDDATA; |
| 60 | } |
| 61 | |
| 62 | if (!(buf[29] & 0x1)) { |
| 63 | av_log(s, AV_LOG_ERROR, "Invalid framing bit in Id header\n"); |
| 64 | return AVERROR_INVALIDDATA; |
| 65 | } |
| 66 | |
| 67 | s->blocksize[0] = 1 << (buf[28] & 0xF); |
| 68 | s->blocksize[1] = 1 << (buf[28] >> 4); |
| 69 | |
| 70 | return 0; |
| 71 | } |
| 72 | |
| 73 | static int parse_setup_header(AVVorbisParseContext *s, |
| 74 | const uint8_t *buf, int buf_size) |
| 75 | { |
| 76 | GetBitContext gb, gb0; |
| 77 | uint8_t *rev_buf; |
| 78 | int i, ret = 0; |
| 79 | int got_framing_bit, mode_count, got_mode_header, last_mode_count = 0; |
| 80 | |
| 81 | /* avoid overread */ |
| 82 | if (buf_size < 7) { |
| 83 | av_log(s, AV_LOG_ERROR, "Setup header is too short\n"); |
| 84 | return AVERROR_INVALIDDATA; |
| 85 | } |
| 86 | |
| 87 | /* make sure this is the Setup header */ |
| 88 | if (buf[0] != 5) { |
| 89 | av_log(s, AV_LOG_ERROR, "Wrong packet type in Setup header\n"); |
| 90 | return AVERROR_INVALIDDATA; |
| 91 | } |
| 92 | |
| 93 | /* check for header signature */ |
| 94 | if (memcmp(&buf[1], "vorbis", 6)) { |
| 95 | av_log(s, AV_LOG_ERROR, "Invalid packet signature in Setup header\n"); |
| 96 | return AVERROR_INVALIDDATA; |
| 97 | } |
| 98 | |
| 99 | /* reverse bytes so we can easily read backwards with get_bits() */ |
| 100 | if (!(rev_buf = av_malloc(buf_size))) { |
| 101 | av_log(s, AV_LOG_ERROR, "Out of memory\n"); |
| 102 | return AVERROR(ENOMEM); |
| 103 | } |
| 104 | for (i = 0; i < buf_size; i++) |
| 105 | rev_buf[i] = buf[buf_size - 1 - i]; |
| 106 | init_get_bits(&gb, rev_buf, buf_size * 8); |
| 107 | |
| 108 | got_framing_bit = 0; |
| 109 | while (get_bits_left(&gb) > 97) { |
| 110 | if (get_bits1(&gb)) { |
| 111 | got_framing_bit = get_bits_count(&gb); |
| 112 | break; |
| 113 | } |
| 114 | } |
| 115 | if (!got_framing_bit) { |
| 116 | av_log(s, AV_LOG_ERROR, "Invalid Setup header\n"); |
| 117 | ret = AVERROR_INVALIDDATA; |
| 118 | goto bad_header; |
| 119 | } |
| 120 | |
| 121 | /* Now we search backwards to find possible valid mode counts. This is not |
| 122 | * fool-proof because we could have false positive matches and read too |
| 123 | * far, but there isn't really any way to be sure without parsing through |
| 124 | * all the many variable-sized fields before the modes. This approach seems |
| 125 | * to work well in testing, and it is similar to how it is handled in |
| 126 | * liboggz. */ |
| 127 | mode_count = 0; |
| 128 | got_mode_header = 0; |
| 129 | while (get_bits_left(&gb) >= 97) { |
| 130 | if (get_bits(&gb, 8) > 63 || get_bits(&gb, 16) || get_bits(&gb, 16)) |
| 131 | break; |
| 132 | skip_bits(&gb, 1); |
| 133 | mode_count++; |
| 134 | if (mode_count > 64) |
| 135 | break; |
| 136 | gb0 = gb; |
| 137 | if (get_bits(&gb0, 6) + 1 == mode_count) { |
| 138 | got_mode_header = 1; |
| 139 | last_mode_count = mode_count; |
| 140 | } |
| 141 | } |
| 142 | if (!got_mode_header) { |
| 143 | av_log(s, AV_LOG_ERROR, "Invalid Setup header\n"); |
| 144 | ret = AVERROR_INVALIDDATA; |
| 145 | goto bad_header; |
| 146 | } |
| 147 | /* All samples I've seen use <= 2 modes, so ask for a sample if we find |
| 148 | * more than that, as it is most likely a false positive. If we get any |
| 149 | * we may need to approach this the long way and parse the whole Setup |
| 150 | * header, but I hope very much that it never comes to that. */ |
| 151 | if (last_mode_count > 2) { |
| 152 | avpriv_request_sample(s, |
| 153 | "%d modes (either a false positive or a " |
| 154 | "sample from an unknown encoder)", |
| 155 | last_mode_count); |
| 156 | } |
| 157 | /* We're limiting the mode count to 63 so that we know that the previous |
| 158 | * block flag will be in the first packet byte. */ |
| 159 | if (last_mode_count > 63) { |
| 160 | av_log(s, AV_LOG_ERROR, "Unsupported mode count: %d\n", |
| 161 | last_mode_count); |
| 162 | ret = AVERROR_INVALIDDATA; |
| 163 | goto bad_header; |
| 164 | } |
| 165 | s->mode_count = mode_count = last_mode_count; |
| 166 | /* Determine the number of bits required to code the mode and turn that |
| 167 | * into a bitmask to directly access the mode from the first frame byte. */ |
| 168 | s->mode_mask = ((1 << (av_log2(mode_count - 1) + 1)) - 1) << 1; |
| 169 | /* The previous window flag is the next bit after the mode */ |
| 170 | s->prev_mask = (s->mode_mask | 0x1) + 1; |
| 171 | |
| 172 | init_get_bits(&gb, rev_buf, buf_size * 8); |
| 173 | skip_bits_long(&gb, got_framing_bit); |
| 174 | for (i = mode_count - 1; i >= 0; i--) { |
| 175 | skip_bits_long(&gb, 40); |
| 176 | s->mode_blocksize[i] = get_bits1(&gb); |
| 177 | } |
| 178 | |
| 179 | bad_header: |
| 180 | av_free(rev_buf); |
| 181 | return ret; |
| 182 | } |
| 183 | |
| 184 | static int vorbis_parse_init(AVVorbisParseContext *s, |
| 185 | const uint8_t *extradata, int extradata_size) |
| 186 | { |
| 187 | uint8_t *header_start[3]; |
| 188 | int header_len[3]; |
| 189 | int ret; |
| 190 | |
| 191 | s->class = &vorbis_parser_class; |
| 192 | s->extradata_parsed = 1; |
| 193 | |
| 194 | if ((ret = avpriv_split_xiph_headers(extradata, |
| 195 | extradata_size, 30, |
| 196 | header_start, header_len)) < 0) { |
| 197 | av_log(s, AV_LOG_ERROR, "Extradata corrupt.\n"); |
| 198 | return ret; |
| 199 | } |
| 200 | |
| 201 | if ((ret = parse_id_header(s, header_start[0], header_len[0])) < 0) |
| 202 | return ret; |
| 203 | |
| 204 | if ((ret = parse_setup_header(s, header_start[2], header_len[2])) < 0) |
| 205 | return ret; |
| 206 | |
| 207 | s->valid_extradata = 1; |
| 208 | s->previous_blocksize = s->blocksize[s->mode_blocksize[0]]; |
| 209 | |
| 210 | return 0; |
| 211 | } |
| 212 | |
| 213 | int av_vorbis_parse_frame_flags(AVVorbisParseContext *s, const uint8_t *buf, |
| 214 | int buf_size, int *flags) |
| 215 | { |
| 216 | int duration = 0; |
| 217 | |
| 218 | if (s->valid_extradata && buf_size > 0) { |
| 219 | int mode, current_blocksize; |
| 220 | int previous_blocksize = s->previous_blocksize; |
| 221 | |
| 222 | if (buf[0] & 1) { |
| 223 | /* If the user doesn't care about special packets, it's a bad one. */ |
| 224 | if (!flags) |
| 225 | goto bad_packet; |
| 226 | |
| 227 | /* Set the flag for which kind of special packet it is. */ |
| 228 | if (buf[0] == 1) |
| 229 | *flags |= VORBIS_FLAG_HEADER; |
| 230 | else if (buf[0] == 3) |
| 231 | *flags |= VORBIS_FLAG_COMMENT; |
| 232 | else |
| 233 | goto bad_packet; |
| 234 | |
| 235 | /* Special packets have no duration. */ |
| 236 | return 0; |
| 237 | |
| 238 | bad_packet: |
| 239 | av_log(s, AV_LOG_ERROR, "Invalid packet\n"); |
| 240 | return AVERROR_INVALIDDATA; |
| 241 | } |
| 242 | if (s->mode_count == 1) |
| 243 | mode = 0; |
| 244 | else |
| 245 | mode = (buf[0] & s->mode_mask) >> 1; |
| 246 | if (mode >= s->mode_count) { |
| 247 | av_log(s, AV_LOG_ERROR, "Invalid mode in packet\n"); |
| 248 | return AVERROR_INVALIDDATA; |
| 249 | } |
| 250 | if(s->mode_blocksize[mode]){ |
| 251 | int flag = !!(buf[0] & s->prev_mask); |
| 252 | previous_blocksize = s->blocksize[flag]; |
| 253 | } |
| 254 | current_blocksize = s->blocksize[s->mode_blocksize[mode]]; |
| 255 | duration = (previous_blocksize + current_blocksize) >> 2; |
| 256 | s->previous_blocksize = current_blocksize; |
| 257 | } |
| 258 | |
| 259 | return duration; |
| 260 | } |
| 261 | |
| 262 | int av_vorbis_parse_frame(AVVorbisParseContext *s, const uint8_t *buf, |
| 263 | int buf_size) |
| 264 | { |
| 265 | return av_vorbis_parse_frame_flags(s, buf, buf_size, NULL); |
| 266 | } |
| 267 | |
| 268 | void av_vorbis_parse_reset(AVVorbisParseContext *s) |
| 269 | { |
| 270 | if (s->valid_extradata) |
| 271 | s->previous_blocksize = s->blocksize[0]; |
| 272 | } |
| 273 | |
| 274 | void av_vorbis_parse_free(AVVorbisParseContext **s) |
| 275 | { |
| 276 | av_freep(s); |
| 277 | } |
| 278 | |
| 279 | AVVorbisParseContext *av_vorbis_parse_init(const uint8_t *extradata, |
| 280 | int extradata_size) |
| 281 | { |
| 282 | AVVorbisParseContext *s = av_mallocz(sizeof(*s)); |
| 283 | int ret; |
| 284 | |
| 285 | if (!s) |
| 286 | return NULL; |
| 287 | |
| 288 | ret = vorbis_parse_init(s, extradata, extradata_size); |
| 289 | if (ret < 0) { |
| 290 | av_vorbis_parse_free(&s); |
| 291 | return NULL; |
| 292 | } |
| 293 | |
| 294 | return s; |
| 295 | } |
| 296 | |
| 297 | #if LIBAVCODEC_VERSION_MAJOR < 57 |
| 298 | int avpriv_vorbis_parse_extradata(AVCodecContext *avctx, AVVorbisParseContext *s) |
| 299 | { |
| 300 | return vorbis_parse_init(s, avctx->extradata, avctx->extradata_size); |
| 301 | } |
| 302 | void avpriv_vorbis_parse_reset(AVVorbisParseContext *s) |
| 303 | { |
| 304 | av_vorbis_parse_reset(s); |
| 305 | } |
| 306 | int avpriv_vorbis_parse_frame(AVVorbisParseContext *s, const uint8_t *buf, |
| 307 | int buf_size) |
| 308 | { |
| 309 | return av_vorbis_parse_frame(s, buf, buf_size); |
| 310 | } |
| 311 | int avpriv_vorbis_parse_frame_flags(AVVorbisParseContext *s, const uint8_t *buf, |
| 312 | int buf_size, int *flags) |
| 313 | { |
| 314 | return av_vorbis_parse_frame_flags(s, buf, buf_size, flags); |
| 315 | } |
| 316 | #endif |
| 317 | |
| 318 | #if CONFIG_VORBIS_PARSER |
| 319 | |
| 320 | typedef struct VorbisParseContext { |
| 321 | AVVorbisParseContext *vp; |
| 322 | } VorbisParseContext; |
| 323 | |
| 324 | static int vorbis_parse(AVCodecParserContext *s1, AVCodecContext *avctx, |
| 325 | const uint8_t **poutbuf, int *poutbuf_size, |
| 326 | const uint8_t *buf, int buf_size) |
| 327 | { |
| 328 | VorbisParseContext *s = s1->priv_data; |
| 329 | int duration; |
| 330 | |
| 331 | if (!s->vp && avctx->extradata && avctx->extradata_size) { |
| 332 | s->vp = av_vorbis_parse_init(avctx->extradata, avctx->extradata_size); |
| 333 | } |
| 334 | if (!s->vp) |
| 335 | goto end; |
| 336 | |
| 337 | if ((duration = av_vorbis_parse_frame(s->vp, buf, buf_size)) >= 0) |
| 338 | s1->duration = duration; |
| 339 | |
| 340 | end: |
| 341 | /* always return the full packet. this parser isn't doing any splitting or |
| 342 | combining, only packet analysis */ |
| 343 | *poutbuf = buf; |
| 344 | *poutbuf_size = buf_size; |
| 345 | return buf_size; |
| 346 | } |
| 347 | |
| 348 | static void vorbis_parser_close(AVCodecParserContext *ctx) |
| 349 | { |
| 350 | VorbisParseContext *s = ctx->priv_data; |
| 351 | av_vorbis_parse_free(&s->vp); |
| 352 | } |
| 353 | |
| 354 | AVCodecParser ff_vorbis_parser = { |
| 355 | .codec_ids = { AV_CODEC_ID_VORBIS }, |
| 356 | .priv_data_size = sizeof(VorbisParseContext), |
| 357 | .parser_parse = vorbis_parse, |
| 358 | .parser_close = vorbis_parser_close, |
| 359 | }; |
| 360 | #endif /* CONFIG_VORBIS_PARSER */ |