avcodec/extract_extradata: Move the reference in the bsf internal buffer
[libav.git] / libavcodec / extract_extradata_bsf.c
CommitLineData
89b35a13
AK
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
33typedef struct ExtractExtradataContext {
34 const AVClass *class;
35
36 int (*extract)(AVBSFContext *ctx, AVPacket *pkt,
37 uint8_t **data, int *size);
38
39 /* AVOptions */
40 int remove;
41} ExtractExtradataContext;
42
43static int val_in_array(const int *arr, int len, int val)
44{
45 int i;
46 for (i = 0; i < len; i++)
47 if (arr[i] == val)
48 return 1;
49 return 0;
50}
51
52static int extract_extradata_h2645(AVBSFContext *ctx, AVPacket *pkt,
53 uint8_t **data, int *size)
54{
55 static const int extradata_nal_types_hevc[] = {
56 HEVC_NAL_VPS, HEVC_NAL_SPS, HEVC_NAL_PPS,
57 };
58 static const int extradata_nal_types_h264[] = {
59 H264_NAL_SPS, H264_NAL_PPS,
60 };
61
62 ExtractExtradataContext *s = ctx->priv_data;
63
64 H2645Packet h2645_pkt = { 0 };
70f11ee9 65 int extradata_size = 0, filtered_size = 0;
89b35a13
AK
66 const int *extradata_nal_types;
67 int nb_extradata_nal_types;
d7dcd825 68 int i, has_sps = 0, has_vps = 0, ret = 0;
89b35a13
AK
69
70 if (ctx->par_in->codec_id == AV_CODEC_ID_HEVC) {
71 extradata_nal_types = extradata_nal_types_hevc;
72 nb_extradata_nal_types = FF_ARRAY_ELEMS(extradata_nal_types_hevc);
73 } else {
74 extradata_nal_types = extradata_nal_types_h264;
75 nb_extradata_nal_types = FF_ARRAY_ELEMS(extradata_nal_types_h264);
76 }
77
78 ret = ff_h2645_packet_split(&h2645_pkt, pkt->data, pkt->size,
79 ctx, 0, 0, ctx->par_in->codec_id);
80 if (ret < 0)
aaf7fd16 81 goto fail;
89b35a13
AK
82
83 for (i = 0; i < h2645_pkt.nb_nals; i++) {
84 H2645NAL *nal = &h2645_pkt.nals[i];
d7dcd825 85 if (val_in_array(extradata_nal_types, nb_extradata_nal_types, nal->type)) {
89b35a13 86 extradata_size += nal->raw_size + 3;
d7dcd825
JA
87 if (ctx->par_in->codec_id == AV_CODEC_ID_HEVC) {
88 if (nal->type == HEVC_NAL_SPS) has_sps = 1;
89 if (nal->type == HEVC_NAL_VPS) has_vps = 1;
90 } else {
91 if (nal->type == H264_NAL_SPS) has_sps = 1;
92 }
70f11ee9
JA
93 } else if (s->remove) {
94 filtered_size += nal->raw_size + 3;
d7dcd825 95 }
89b35a13
AK
96 }
97
d7dcd825
JA
98 if (extradata_size &&
99 ((ctx->par_in->codec_id == AV_CODEC_ID_HEVC && has_sps && has_vps) ||
100 (ctx->par_in->codec_id == AV_CODEC_ID_H264 && has_sps))) {
89b35a13
AK
101 AVBufferRef *filtered_buf;
102 uint8_t *extradata, *filtered_data;
103
104 if (s->remove) {
70f11ee9 105 filtered_buf = av_buffer_alloc(filtered_size + AV_INPUT_BUFFER_PADDING_SIZE);
20142310
JA
106 if (!filtered_buf) {
107 ret = AVERROR(ENOMEM);
89b35a13 108 goto fail;
20142310 109 }
70f11ee9
JA
110 memset(filtered_buf->data + filtered_size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
111
89b35a13
AK
112 filtered_data = filtered_buf->data;
113 }
114
115 extradata = av_malloc(extradata_size + AV_INPUT_BUFFER_PADDING_SIZE);
116 if (!extradata) {
117 av_buffer_unref(&filtered_buf);
20142310 118 ret = AVERROR(ENOMEM);
89b35a13
AK
119 goto fail;
120 }
be65995d 121 memset(extradata + extradata_size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
89b35a13
AK
122
123 *data = extradata;
124 *size = extradata_size;
125
126 for (i = 0; i < h2645_pkt.nb_nals; i++) {
127 H2645NAL *nal = &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;
70f11ee9 144 pkt->size = filtered_size;
89b35a13
AK
145 }
146 }
147
148fail:
149 ff_h2645_packet_uninit(&h2645_pkt);
150 return ret;
151}
152
153static int extract_extradata_vc1(AVBSFContext *ctx, AVPacket *pkt,
154 uint8_t **data, int *size)
155{
156 ExtractExtradataContext *s = ctx->priv_data;
157 uint32_t state = UINT32_MAX;
158 int has_extradata = 0, extradata_size = 0;
159 int i;
160
161 for (i = 0; i < pkt->size; i++) {
162 state = (state << 8) | pkt->data[i];
163 if (IS_MARKER(state)) {
164 if (state == VC1_CODE_SEQHDR || state == VC1_CODE_ENTRYPOINT) {
165 has_extradata = 1;
166 } else if (has_extradata) {
167 extradata_size = i - 3;
168 break;
169 }
170 }
171 }
172
173 if (extradata_size) {
174 *data = av_malloc(extradata_size + AV_INPUT_BUFFER_PADDING_SIZE);
175 if (!*data)
176 return AVERROR(ENOMEM);
177
178 memcpy(*data, pkt->data, extradata_size);
be65995d 179 memset(*data + extradata_size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
89b35a13
AK
180 *size = extradata_size;
181
182 if (s->remove) {
183 pkt->data += extradata_size;
184 pkt->size -= extradata_size;
185 }
186 }
187
188 return 0;
189}
190
191static int extract_extradata_mpeg124(AVBSFContext *ctx, AVPacket *pkt,
192 uint8_t **data, int *size)
193{
194 ExtractExtradataContext *s = ctx->priv_data;
195 int is_mpeg12 = ctx->par_in->codec_id == AV_CODEC_ID_MPEG1VIDEO ||
196 ctx->par_in->codec_id == AV_CODEC_ID_MPEG2VIDEO;
197 uint32_t state = UINT32_MAX;
198 int i;
199
200 for (i = 0; i < pkt->size; i++) {
201 state = (state << 8) | pkt->data[i];
202 if ((is_mpeg12 && state != 0x1B3 && state != 0x1B5 && state < 0x200 && state >= 0x100) ||
203 (!is_mpeg12 && (state == 0x1B3 || state == 0x1B6))) {
204 if (i > 3) {
205 *size = i - 3;
206 *data = av_malloc(*size + AV_INPUT_BUFFER_PADDING_SIZE);
207 if (!*data)
208 return AVERROR(ENOMEM);
209
210 memcpy(*data, pkt->data, *size);
be65995d 211 memset(*data + *size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
89b35a13
AK
212
213 if (s->remove) {
214 pkt->data += *size;
215 pkt->size -= *size;
216 }
217 }
218 break;
219 }
220 }
221 return 0;
222}
223
224static const struct {
225 enum AVCodecID id;
226 int (*extract)(AVBSFContext *ctx, AVPacket *pkt,
227 uint8_t **data, int *size);
228} extract_tab[] = {
229 { AV_CODEC_ID_CAVS, extract_extradata_mpeg124 },
230 { AV_CODEC_ID_H264, extract_extradata_h2645 },
231 { AV_CODEC_ID_HEVC, extract_extradata_h2645 },
232 { AV_CODEC_ID_MPEG1VIDEO, extract_extradata_mpeg124 },
233 { AV_CODEC_ID_MPEG2VIDEO, extract_extradata_mpeg124 },
234 { AV_CODEC_ID_MPEG4, extract_extradata_mpeg124 },
235 { AV_CODEC_ID_VC1, extract_extradata_vc1 },
236};
237
238static int extract_extradata_init(AVBSFContext *ctx)
239{
240 ExtractExtradataContext *s = ctx->priv_data;
241 int i;
242
243 for (i = 0; i < FF_ARRAY_ELEMS(extract_tab); i++) {
244 if (extract_tab[i].id == ctx->par_in->codec_id) {
245 s->extract = extract_tab[i].extract;
246 break;
247 }
248 }
249 if (!s->extract)
250 return AVERROR_BUG;
251
252 return 0;
253}
254
6442d8ba 255static int extract_extradata_filter(AVBSFContext *ctx, AVPacket *pkt)
89b35a13
AK
256{
257 ExtractExtradataContext *s = ctx->priv_data;
89b35a13
AK
258 uint8_t *extradata = NULL;
259 int extradata_size;
260 int ret = 0;
261
6442d8ba 262 ret = ff_bsf_get_packet_ref(ctx, pkt);
89b35a13
AK
263 if (ret < 0)
264 return ret;
265
6442d8ba 266 ret = s->extract(ctx, pkt, &extradata, &extradata_size);
89b35a13
AK
267 if (ret < 0)
268 goto fail;
269
270 if (extradata) {
6442d8ba 271 ret = av_packet_add_side_data(pkt, AV_PKT_DATA_NEW_EXTRADATA,
89b35a13
AK
272 extradata, extradata_size);
273 if (ret < 0) {
274 av_freep(&extradata);
275 goto fail;
276 }
277 }
278
6442d8ba 279 return 0;
89b35a13
AK
280
281fail:
6442d8ba 282 av_packet_unref(pkt);
89b35a13
AK
283 return ret;
284}
285
286static const enum AVCodecID codec_ids[] = {
287 AV_CODEC_ID_CAVS,
288 AV_CODEC_ID_H264,
289 AV_CODEC_ID_HEVC,
290 AV_CODEC_ID_MPEG1VIDEO,
291 AV_CODEC_ID_MPEG2VIDEO,
292 AV_CODEC_ID_MPEG4,
293 AV_CODEC_ID_VC1,
294 AV_CODEC_ID_NONE,
295};
296
297#define OFFSET(x) offsetof(ExtractExtradataContext, x)
298static const AVOption options[] = {
299 { "remove", "remove the extradata from the bitstream", OFFSET(remove), AV_OPT_TYPE_INT,
300 { .i64 = 0 }, 0, 1 },
301 { NULL },
302};
303
304static const AVClass extract_extradata_class = {
305 .class_name = "extract_extradata",
306 .item_name = av_default_item_name,
307 .option = options,
308 .version = LIBAVUTIL_VERSION_INT,
309};
310
311const AVBitStreamFilter ff_extract_extradata_bsf = {
312 .name = "extract_extradata",
313 .codec_ids = codec_ids,
314 .priv_data_size = sizeof(ExtractExtradataContext),
315 .priv_class = &extract_extradata_class,
316 .init = extract_extradata_init,
317 .filter = extract_extradata_filter,
318};