1 /*
2 * This file is part of FFmpeg.
3 *
4 * FFmpeg 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 * FFmpeg 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 FFmpeg; 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 "av1.h"
27 #include "av1_parse.h"
28 #include "bsf.h"
29 #include "bsf_internal.h"
30 #include "bytestream.h"
31 #include "h2645_parse.h"
32 #include "h264.h"
33 #include "hevc.h"
34 #include "vc1_common.h"
35
36 typedef struct ExtractExtradataContext {
37 const AVClass *class;
38
39 int (*extract)(AVBSFContext *ctx, AVPacket *pkt,
40 uint8_t **data, int *size);
41
42 /* AV1 specific fields */
43 AV1Packet av1_pkt;
44
45 /* H264/HEVC specific fields */
46 H2645Packet h2645_pkt;
47
48 /* AVOptions */
49 int remove;
50 } ExtractExtradataContext;
51
val_in_array(const int * arr,int len,int val)52 static int val_in_array(const int *arr, int len, int val)
53 {
54 int i;
55 for (i = 0; i < len; i++)
56 if (arr[i] == val)
57 return 1;
58 return 0;
59 }
60
extract_extradata_av1(AVBSFContext * ctx,AVPacket * pkt,uint8_t ** data,int * size)61 static int extract_extradata_av1(AVBSFContext *ctx, AVPacket *pkt,
62 uint8_t **data, int *size)
63 {
64 static const int extradata_obu_types[] = {
65 AV1_OBU_SEQUENCE_HEADER, AV1_OBU_METADATA,
66 };
67 ExtractExtradataContext *s = ctx->priv_data;
68
69 int extradata_size = 0, filtered_size = 0;
70 int nb_extradata_obu_types = FF_ARRAY_ELEMS(extradata_obu_types);
71 int i, has_seq = 0, ret = 0;
72
73 ret = ff_av1_packet_split(&s->av1_pkt, pkt->data, pkt->size, ctx);
74 if (ret < 0)
75 return ret;
76
77 for (i = 0; i < s->av1_pkt.nb_obus; i++) {
78 AV1OBU *obu = &s->av1_pkt.obus[i];
79 if (val_in_array(extradata_obu_types, nb_extradata_obu_types, obu->type)) {
80 extradata_size += obu->raw_size;
81 if (obu->type == AV1_OBU_SEQUENCE_HEADER)
82 has_seq = 1;
83 } else if (s->remove) {
84 filtered_size += obu->raw_size;
85 }
86 }
87
88 if (extradata_size && has_seq) {
89 AVBufferRef *filtered_buf = NULL;
90 PutByteContext pb_filtered_data, pb_extradata;
91 uint8_t *extradata;
92
93 if (s->remove) {
94 filtered_buf = av_buffer_alloc(filtered_size + AV_INPUT_BUFFER_PADDING_SIZE);
95 if (!filtered_buf) {
96 return AVERROR(ENOMEM);
97 }
98 memset(filtered_buf->data + filtered_size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
99 }
100
101 extradata = av_malloc(extradata_size + AV_INPUT_BUFFER_PADDING_SIZE);
102 if (!extradata) {
103 av_buffer_unref(&filtered_buf);
104 return AVERROR(ENOMEM);
105 }
106 memset(extradata + extradata_size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
107
108 *data = extradata;
109 *size = extradata_size;
110
111 bytestream2_init_writer(&pb_extradata, extradata, extradata_size);
112 if (s->remove)
113 bytestream2_init_writer(&pb_filtered_data, filtered_buf->data, filtered_size);
114
115 for (i = 0; i < s->av1_pkt.nb_obus; i++) {
116 AV1OBU *obu = &s->av1_pkt.obus[i];
117 if (val_in_array(extradata_obu_types, nb_extradata_obu_types,
118 obu->type)) {
119 bytestream2_put_bufferu(&pb_extradata, obu->raw_data, obu->raw_size);
120 } else if (s->remove) {
121 bytestream2_put_bufferu(&pb_filtered_data, obu->raw_data, obu->raw_size);
122 }
123 }
124
125 if (s->remove) {
126 av_buffer_unref(&pkt->buf);
127 pkt->buf = filtered_buf;
128 pkt->data = filtered_buf->data;
129 pkt->size = filtered_size;
130 }
131 }
132
133 return 0;
134 }
135
extract_extradata_h2645(AVBSFContext * ctx,AVPacket * pkt,uint8_t ** data,int * size)136 static int extract_extradata_h2645(AVBSFContext *ctx, AVPacket *pkt,
137 uint8_t **data, int *size)
138 {
139 static const int extradata_nal_types_hevc[] = {
140 HEVC_NAL_VPS, HEVC_NAL_SPS, HEVC_NAL_PPS,
141 };
142 static const int extradata_nal_types_h264[] = {
143 H264_NAL_SPS, H264_NAL_PPS,
144 };
145
146 ExtractExtradataContext *s = ctx->priv_data;
147
148 int extradata_size = 0, filtered_size = 0;
149 const int *extradata_nal_types;
150 int nb_extradata_nal_types;
151 int i, has_sps = 0, has_vps = 0, ret = 0;
152
153 if (ctx->par_in->codec_id == AV_CODEC_ID_HEVC) {
154 extradata_nal_types = extradata_nal_types_hevc;
155 nb_extradata_nal_types = FF_ARRAY_ELEMS(extradata_nal_types_hevc);
156 } else {
157 extradata_nal_types = extradata_nal_types_h264;
158 nb_extradata_nal_types = FF_ARRAY_ELEMS(extradata_nal_types_h264);
159 }
160
161 ret = ff_h2645_packet_split(&s->h2645_pkt, pkt->data, pkt->size,
162 ctx, 0, 0, ctx->par_in->codec_id, 1, 0);
163 if (ret < 0)
164 return ret;
165
166 for (i = 0; i < s->h2645_pkt.nb_nals; i++) {
167 H2645NAL *nal = &s->h2645_pkt.nals[i];
168 if (val_in_array(extradata_nal_types, nb_extradata_nal_types, nal->type)) {
169 extradata_size += nal->raw_size + 3;
170 if (ctx->par_in->codec_id == AV_CODEC_ID_HEVC) {
171 if (nal->type == HEVC_NAL_SPS) has_sps = 1;
172 if (nal->type == HEVC_NAL_VPS) has_vps = 1;
173 } else {
174 if (nal->type == H264_NAL_SPS) has_sps = 1;
175 }
176 } else if (s->remove) {
177 filtered_size += nal->raw_size + 3;
178 }
179 }
180
181 if (extradata_size &&
182 ((ctx->par_in->codec_id == AV_CODEC_ID_HEVC && has_sps && has_vps) ||
183 (ctx->par_in->codec_id == AV_CODEC_ID_H264 && has_sps))) {
184 AVBufferRef *filtered_buf = NULL;
185 PutByteContext pb_filtered_data, pb_extradata;
186 uint8_t *extradata;
187
188 if (s->remove) {
189 filtered_buf = av_buffer_alloc(filtered_size + AV_INPUT_BUFFER_PADDING_SIZE);
190 if (!filtered_buf) {
191 return AVERROR(ENOMEM);
192 }
193 memset(filtered_buf->data + filtered_size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
194 }
195
196 extradata = av_malloc(extradata_size + AV_INPUT_BUFFER_PADDING_SIZE);
197 if (!extradata) {
198 av_buffer_unref(&filtered_buf);
199 return AVERROR(ENOMEM);
200 }
201 memset(extradata + extradata_size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
202
203 *data = extradata;
204 *size = extradata_size;
205
206 bytestream2_init_writer(&pb_extradata, extradata, extradata_size);
207 if (s->remove)
208 bytestream2_init_writer(&pb_filtered_data, filtered_buf->data, filtered_size);
209
210 for (i = 0; i < s->h2645_pkt.nb_nals; i++) {
211 H2645NAL *nal = &s->h2645_pkt.nals[i];
212 if (val_in_array(extradata_nal_types, nb_extradata_nal_types,
213 nal->type)) {
214 bytestream2_put_be24u(&pb_extradata, 1); //startcode
215 bytestream2_put_bufferu(&pb_extradata, nal->raw_data, nal->raw_size);
216 } else if (s->remove) {
217 bytestream2_put_be24u(&pb_filtered_data, 1); // startcode
218 bytestream2_put_bufferu(&pb_filtered_data, nal->raw_data, nal->raw_size);
219 }
220 }
221
222 if (s->remove) {
223 av_buffer_unref(&pkt->buf);
224 pkt->buf = filtered_buf;
225 pkt->data = filtered_buf->data;
226 pkt->size = filtered_size;
227 }
228 }
229
230 return 0;
231 }
232
extract_extradata_vc1(AVBSFContext * ctx,AVPacket * pkt,uint8_t ** data,int * size)233 static int extract_extradata_vc1(AVBSFContext *ctx, AVPacket *pkt,
234 uint8_t **data, int *size)
235 {
236 ExtractExtradataContext *s = ctx->priv_data;
237 const uint8_t *ptr = pkt->data, *end = pkt->data + pkt->size;
238 uint32_t state = UINT32_MAX;
239 int has_extradata = 0, extradata_size = 0;
240
241 while (ptr < end) {
242 ptr = avpriv_find_start_code(ptr, end, &state);
243 if (state == VC1_CODE_SEQHDR || state == VC1_CODE_ENTRYPOINT) {
244 has_extradata = 1;
245 } else if (has_extradata && IS_MARKER(state)) {
246 extradata_size = ptr - 4 - pkt->data;
247 break;
248 }
249 }
250
251 if (extradata_size) {
252 *data = av_malloc(extradata_size + AV_INPUT_BUFFER_PADDING_SIZE);
253 if (!*data)
254 return AVERROR(ENOMEM);
255
256 memcpy(*data, pkt->data, extradata_size);
257 memset(*data + extradata_size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
258 *size = extradata_size;
259
260 if (s->remove) {
261 pkt->data += extradata_size;
262 pkt->size -= extradata_size;
263 }
264 }
265
266 return 0;
267 }
268
extract_extradata_mpeg12(AVBSFContext * ctx,AVPacket * pkt,uint8_t ** data,int * size)269 static int extract_extradata_mpeg12(AVBSFContext *ctx, AVPacket *pkt,
270 uint8_t **data, int *size)
271 {
272 ExtractExtradataContext *s = ctx->priv_data;
273 uint32_t state = UINT32_MAX;
274 int i, found = 0;
275
276 for (i = 0; i < pkt->size; i++) {
277 state = (state << 8) | pkt->data[i];
278 if (state == 0x1B3)
279 found = 1;
280 else if (found && state != 0x1B5 && state < 0x200 && state >= 0x100) {
281 if (i > 3) {
282 *size = i - 3;
283 *data = av_malloc(*size + AV_INPUT_BUFFER_PADDING_SIZE);
284 if (!*data)
285 return AVERROR(ENOMEM);
286
287 memcpy(*data, pkt->data, *size);
288 memset(*data + *size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
289
290 if (s->remove) {
291 pkt->data += *size;
292 pkt->size -= *size;
293 }
294 }
295 break;
296 }
297 }
298 return 0;
299 }
300
extract_extradata_mpeg4(AVBSFContext * ctx,AVPacket * pkt,uint8_t ** data,int * size)301 static int extract_extradata_mpeg4(AVBSFContext *ctx, AVPacket *pkt,
302 uint8_t **data, int *size)
303 {
304 ExtractExtradataContext *s = ctx->priv_data;
305 const uint8_t *ptr = pkt->data, *end = pkt->data + pkt->size;
306 uint32_t state = UINT32_MAX;
307
308 while (ptr < end) {
309 ptr = avpriv_find_start_code(ptr, end, &state);
310 if (state == 0x1B3 || state == 0x1B6) {
311 if (ptr - pkt->data > 4) {
312 *size = ptr - 4 - pkt->data;
313 *data = av_malloc(*size + AV_INPUT_BUFFER_PADDING_SIZE);
314 if (!*data)
315 return AVERROR(ENOMEM);
316
317 memcpy(*data, pkt->data, *size);
318 memset(*data + *size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
319
320 if (s->remove) {
321 pkt->data += *size;
322 pkt->size -= *size;
323 }
324 }
325 break;
326 }
327 }
328 return 0;
329 }
330
331 static const struct {
332 enum AVCodecID id;
333 int (*extract)(AVBSFContext *ctx, AVPacket *pkt,
334 uint8_t **data, int *size);
335 } extract_tab[] = {
336 { AV_CODEC_ID_AV1, extract_extradata_av1 },
337 { AV_CODEC_ID_AVS2, extract_extradata_mpeg4 },
338 { AV_CODEC_ID_CAVS, extract_extradata_mpeg4 },
339 { AV_CODEC_ID_H264, extract_extradata_h2645 },
340 { AV_CODEC_ID_HEVC, extract_extradata_h2645 },
341 { AV_CODEC_ID_MPEG1VIDEO, extract_extradata_mpeg12 },
342 { AV_CODEC_ID_MPEG2VIDEO, extract_extradata_mpeg12 },
343 { AV_CODEC_ID_MPEG4, extract_extradata_mpeg4 },
344 { AV_CODEC_ID_VC1, extract_extradata_vc1 },
345 };
346
extract_extradata_init(AVBSFContext * ctx)347 static int extract_extradata_init(AVBSFContext *ctx)
348 {
349 ExtractExtradataContext *s = ctx->priv_data;
350 int i;
351
352 for (i = 0; i < FF_ARRAY_ELEMS(extract_tab); i++) {
353 if (extract_tab[i].id == ctx->par_in->codec_id) {
354 s->extract = extract_tab[i].extract;
355 break;
356 }
357 }
358 if (!s->extract)
359 return AVERROR_BUG;
360
361 return 0;
362 }
363
extract_extradata_filter(AVBSFContext * ctx,AVPacket * pkt)364 static int extract_extradata_filter(AVBSFContext *ctx, AVPacket *pkt)
365 {
366 ExtractExtradataContext *s = ctx->priv_data;
367 uint8_t *extradata = NULL;
368 int extradata_size;
369 int ret = 0;
370
371 ret = ff_bsf_get_packet_ref(ctx, pkt);
372 if (ret < 0)
373 return ret;
374
375 ret = s->extract(ctx, pkt, &extradata, &extradata_size);
376 if (ret < 0)
377 goto fail;
378
379 if (extradata) {
380 ret = av_packet_add_side_data(pkt, AV_PKT_DATA_NEW_EXTRADATA,
381 extradata, extradata_size);
382 if (ret < 0) {
383 av_freep(&extradata);
384 goto fail;
385 }
386 }
387
388 return 0;
389
390 fail:
391 av_packet_unref(pkt);
392 return ret;
393 }
394
extract_extradata_close(AVBSFContext * ctx)395 static void extract_extradata_close(AVBSFContext *ctx)
396 {
397 ExtractExtradataContext *s = ctx->priv_data;
398 ff_av1_packet_uninit(&s->av1_pkt);
399 ff_h2645_packet_uninit(&s->h2645_pkt);
400 }
401
402 static const enum AVCodecID codec_ids[] = {
403 AV_CODEC_ID_AV1,
404 AV_CODEC_ID_AVS2,
405 AV_CODEC_ID_CAVS,
406 AV_CODEC_ID_H264,
407 AV_CODEC_ID_HEVC,
408 AV_CODEC_ID_MPEG1VIDEO,
409 AV_CODEC_ID_MPEG2VIDEO,
410 AV_CODEC_ID_MPEG4,
411 AV_CODEC_ID_VC1,
412 AV_CODEC_ID_NONE,
413 };
414
415 #define OFFSET(x) offsetof(ExtractExtradataContext, x)
416 #define FLAGS (AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_BSF_PARAM)
417 static const AVOption options[] = {
418 { "remove", "remove the extradata from the bitstream", OFFSET(remove), AV_OPT_TYPE_INT,
419 { .i64 = 0 }, 0, 1, FLAGS },
420 { NULL },
421 };
422
423 static const AVClass extract_extradata_class = {
424 .class_name = "extract_extradata",
425 .item_name = av_default_item_name,
426 .option = options,
427 .version = LIBAVUTIL_VERSION_INT,
428 };
429
430 const AVBitStreamFilter ff_extract_extradata_bsf = {
431 .name = "extract_extradata",
432 .codec_ids = codec_ids,
433 .priv_data_size = sizeof(ExtractExtradataContext),
434 .priv_class = &extract_extradata_class,
435 .init = extract_extradata_init,
436 .filter = extract_extradata_filter,
437 .close = extract_extradata_close,
438 };
439