libavcodec/extract_extradata_bsf.c

   1 /*
   2  * This file is part of Libav.
   3  *
   4  * Libav is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU Lesser General Public
   6  * License as published by the Free Software Foundation; either
   7  * version 2.1 of the License, or (at your option) any later version.
   8  *
   9  * Libav is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  12  * Lesser General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU Lesser General Public
  15  * License along with Libav; if not, write to the Free Software
  16  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  17  */
  18
  19 #include <stdint.h>
  20
  21 #include "libavutil/common.h"
  22 #include "libavutil/intreadwrite.h"
  23 #include "libavutil/log.h"
  24 #include "libavutil/opt.h"
  25
  26 #include "avcodec.h"
  27 #include "bsf.h"
  28 #include "h2645_parse.h"
  29 #include "h264.h"
  30 #include "hevc.h"
  31 #include "vc1_common.h"
  32
  33 typedef struct ExtractExtradataContext {
  34     const AVClass *class;
  35
  36     int (*extract)(AVBSFContext *ctx, AVPacket *pkt,
  37                    uint8_t **data, int *size);
  38
  39     /* H264/HEVC specifc fields */
  40     H2645Packet h2645_pkt;
  41
  42     /* AVOptions */
  43     int remove;
  44 } ExtractExtradataContext;
  45
  46 static int val_in_array(const int *arr, int len, int val)
  47 {
  48     int i;
  49     for (i = 0; i < len; i++)
  50         if (arr[i] == val)
  51             return 1;
  52     return 0;
  53 }
  54
  55 static int extract_extradata_h2645(AVBSFContext *ctx, AVPacket *pkt,
  56                                    uint8_t **data, int *size)
  57 {
  58     static const int extradata_nal_types_hevc[] = {
  59         HEVC_NAL_VPS, HEVC_NAL_SPS, HEVC_NAL_PPS,
  60     };
  61     static const int extradata_nal_types_h264[] = {
  62         H264_NAL_SPS, H264_NAL_PPS,
  63     };
  64
  65     ExtractExtradataContext *s = ctx->priv_data;
  66
  67     int extradata_size = 0, filtered_size = 0;
  68     const int *extradata_nal_types;
  69     int nb_extradata_nal_types;
  70     int i, has_sps = 0, has_vps = 0, ret = 0;
  71
  72     if (ctx->par_in->codec_id == AV_CODEC_ID_HEVC) {
  73         extradata_nal_types    = extradata_nal_types_hevc;
  74         nb_extradata_nal_types = FF_ARRAY_ELEMS(extradata_nal_types_hevc);
  75     } else {
  76         extradata_nal_types    = extradata_nal_types_h264;
  77         nb_extradata_nal_types = FF_ARRAY_ELEMS(extradata_nal_types_h264);
  78     }
  79
  80     ret = ff_h2645_packet_split(&s->h2645_pkt, pkt->data, pkt->size,
  81                                 ctx, 0, 0, ctx->par_in->codec_id);
  82     if (ret < 0)
  83         return ret;
  84
  85     for (i = 0; i < s->h2645_pkt.nb_nals; i++) {
  86         H2645NAL *nal = &s->h2645_pkt.nals[i];
  87         if (val_in_array(extradata_nal_types, nb_extradata_nal_types, nal->type)) {
  88             extradata_size += nal->raw_size + 3;
  89             if (ctx->par_in->codec_id == AV_CODEC_ID_HEVC) {
  90                 if (nal->type == HEVC_NAL_SPS) has_sps = 1;
  91                 if (nal->type == HEVC_NAL_VPS) has_vps = 1;
  92             } else {
  93                 if (nal->type == H264_NAL_SPS) has_sps = 1;
  94             }
  95         } else if (s->remove) {
  96             filtered_size += nal->raw_size + 3;
  97         }
  98     }
  99
 100     if (extradata_size &&
 101         ((ctx->par_in->codec_id == AV_CODEC_ID_HEVC && has_sps && has_vps) ||
 102          (ctx->par_in->codec_id == AV_CODEC_ID_H264 && has_sps))) {
 103         AVBufferRef *filtered_buf;
 104         uint8_t *extradata, *filtered_data;
 105
 106         if (s->remove) {
 107             filtered_buf = av_buffer_alloc(filtered_size + AV_INPUT_BUFFER_PADDING_SIZE);
 108             if (!filtered_buf) {
 109                 return AVERROR(ENOMEM);
 110             }
 111             memset(filtered_buf->data + filtered_size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
 112
 113             filtered_data = filtered_buf->data;
 114         }
 115
 116         extradata = av_malloc(extradata_size + AV_INPUT_BUFFER_PADDING_SIZE);
 117         if (!extradata) {
 118             av_buffer_unref(&filtered_buf);
 119             return AVERROR(ENOMEM);
 120         }
 121         memset(extradata + extradata_size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
 122
 123         *data = extradata;
 124         *size = extradata_size;
 125
 126         for (i = 0; i < s->h2645_pkt.nb_nals; i++) {
 127             H2645NAL *nal = &s->h2645_pkt.nals[i];
 128             if (val_in_array(extradata_nal_types, nb_extradata_nal_types,
 129                              nal->type)) {
 130                 AV_WB24(extradata, 1); // startcode
 131                 memcpy(extradata + 3, nal->raw_data, nal->raw_size);
 132                 extradata += 3 + nal->raw_size;
 133             } else if (s->remove) {
 134                 AV_WB24(filtered_data, 1); // startcode
 135                 memcpy(filtered_data + 3, nal->raw_data, nal->raw_size);
 136                 filtered_data += 3 + nal->raw_size;
 137             }
 138         }
 139
 140         if (s->remove) {
 141             av_buffer_unref(&pkt->buf);
 142             pkt->buf  = filtered_buf;
 143             pkt->data = filtered_buf->data;
 144             pkt->size = filtered_size;
 145         }
 146     }
 147
 148     return 0;
 149 }
 150
 151 static int extract_extradata_vc1(AVBSFContext *ctx, AVPacket *pkt,
 152                                  uint8_t **data, int *size)
 153 {
 154     ExtractExtradataContext *s = ctx->priv_data;
 155     uint32_t state = UINT32_MAX;
 156     int has_extradata = 0, extradata_size = 0;
 157     int i;
 158
 159     for (i = 0; i < pkt->size; i++) {
 160         state = (state << 8) | pkt->data[i];
 161         if (IS_MARKER(state)) {
 162             if (state == VC1_CODE_SEQHDR || state == VC1_CODE_ENTRYPOINT) {
 163                 has_extradata = 1;
 164             } else if (has_extradata) {
 165                 extradata_size = i - 3;
 166                 break;
 167             }
 168         }
 169     }
 170
 171     if (extradata_size) {
 172         *data = av_malloc(extradata_size + AV_INPUT_BUFFER_PADDING_SIZE);
 173         if (!*data)
 174             return AVERROR(ENOMEM);
 175
 176         memcpy(*data, pkt->data, extradata_size);
 177         memset(*data + extradata_size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
 178         *size = extradata_size;
 179
 180         if (s->remove) {
 181             pkt->data += extradata_size;
 182             pkt->size -= extradata_size;
 183         }
 184     }
 185
 186     return 0;
 187 }
 188
 189 static int extract_extradata_mpeg124(AVBSFContext *ctx, AVPacket *pkt,
 190                                      uint8_t **data, int *size)
 191 {
 192     ExtractExtradataContext *s = ctx->priv_data;
 193     int is_mpeg12 = ctx->par_in->codec_id == AV_CODEC_ID_MPEG1VIDEO ||
 194                     ctx->par_in->codec_id == AV_CODEC_ID_MPEG2VIDEO;
 195     uint32_t state = UINT32_MAX;
 196     int i;
 197
 198     for (i = 0; i < pkt->size; i++) {
 199         state = (state << 8) | pkt->data[i];
 200         if ((is_mpeg12 && state != 0x1B3 && state != 0x1B5 && state < 0x200 && state >= 0x100) ||
 201             (!is_mpeg12 && (state == 0x1B3 || state == 0x1B6))) {
 202             if (i > 3) {
 203                 *size = i - 3;
 204                 *data = av_malloc(*size + AV_INPUT_BUFFER_PADDING_SIZE);
 205                 if (!*data)
 206                     return AVERROR(ENOMEM);
 207
 208                 memcpy(*data, pkt->data, *size);
 209                 memset(*data + *size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
 210
 211                 if (s->remove) {
 212                     pkt->data += *size;
 213                     pkt->size -= *size;
 214                 }
 215             }
 216             break;
 217         }
 218     }
 219     return 0;
 220 }
 221
 222 static const struct {
 223     enum AVCodecID id;
 224     int (*extract)(AVBSFContext *ctx, AVPacket *pkt,
 225                    uint8_t **data, int *size);
 226 } extract_tab[] = {
 227     { AV_CODEC_ID_CAVS,       extract_extradata_mpeg124 },
 228     { AV_CODEC_ID_H264,       extract_extradata_h2645   },
 229     { AV_CODEC_ID_HEVC,       extract_extradata_h2645   },
 230     { AV_CODEC_ID_MPEG1VIDEO, extract_extradata_mpeg124 },
 231     { AV_CODEC_ID_MPEG2VIDEO, extract_extradata_mpeg124 },
 232     { AV_CODEC_ID_MPEG4,      extract_extradata_mpeg124 },
 233     { AV_CODEC_ID_VC1,        extract_extradata_vc1     },
 234 };
 235
 236 static int extract_extradata_init(AVBSFContext *ctx)
 237 {
 238     ExtractExtradataContext *s = ctx->priv_data;
 239     int i;
 240
 241     for (i = 0; i < FF_ARRAY_ELEMS(extract_tab); i++) {
 242         if (extract_tab[i].id == ctx->par_in->codec_id) {
 243             s->extract = extract_tab[i].extract;
 244             break;
 245         }
 246     }
 247     if (!s->extract)
 248         return AVERROR_BUG;
 249
 250     return 0;
 251 }
 252
 253 static int extract_extradata_filter(AVBSFContext *ctx, AVPacket *pkt)
 254 {
 255     ExtractExtradataContext *s = ctx->priv_data;
 256     uint8_t *extradata = NULL;
 257     int extradata_size;
 258     int ret = 0;
 259
 260     ret = ff_bsf_get_packet_ref(ctx, pkt);
 261     if (ret < 0)
 262         return ret;
 263
 264     ret = s->extract(ctx, pkt, &extradata, &extradata_size);
 265     if (ret < 0)
 266         goto fail;
 267
 268     if (extradata) {
 269         ret = av_packet_add_side_data(pkt, AV_PKT_DATA_NEW_EXTRADATA,
 270                                       extradata, extradata_size);
 271         if (ret < 0) {
 272             av_freep(&extradata);
 273             goto fail;
 274         }
 275     }
 276
 277     return 0;
 278
 279 fail:
 280     av_packet_unref(pkt);
 281     return ret;
 282 }
 283
 284 static void extract_extradata_close(AVBSFContext *ctx)
 285 {
 286     ExtractExtradataContext *s = ctx->priv_data;
 287     ff_h2645_packet_uninit(&s->h2645_pkt);
 288 }
 289
 290 static const enum AVCodecID codec_ids[] = {
 291     AV_CODEC_ID_CAVS,
 292     AV_CODEC_ID_H264,
 293     AV_CODEC_ID_HEVC,
 294     AV_CODEC_ID_MPEG1VIDEO,
 295     AV_CODEC_ID_MPEG2VIDEO,
 296     AV_CODEC_ID_MPEG4,
 297     AV_CODEC_ID_VC1,
 298     AV_CODEC_ID_NONE,
 299 };
 300
 301 #define OFFSET(x) offsetof(ExtractExtradataContext, x)
 302 static const AVOption options[] = {
 303     { "remove", "remove the extradata from the bitstream", OFFSET(remove), AV_OPT_TYPE_INT,
 304         { .i64 = 0 }, 0, 1 },
 305     { NULL },
 306 };
 307
 308 static const AVClass extract_extradata_class = {
 309     .class_name = "extract_extradata",
 310     .item_name  = av_default_item_name,
 311     .option     = options,
 312     .version    = LIBAVUTIL_VERSION_INT,
 313 };
 314
 315 const AVBitStreamFilter ff_extract_extradata_bsf = {
 316     .name           = "extract_extradata",
 317     .codec_ids      = codec_ids,
 318     .priv_data_size = sizeof(ExtractExtradataContext),
 319     .priv_class     = &extract_extradata_class,
 320     .init           = extract_extradata_init,
 321     .filter         = extract_extradata_filter,
 322     .close          = extract_extradata_close,
 323 };