FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
vf_subtitles.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2011 Baptiste Coudurier
3  * Copyright (c) 2011 Stefano Sabatini
4  * Copyright (c) 2012 Clément Bœsch
5  *
6  * This file is part of FFmpeg.
7  *
8  * FFmpeg is free software; you can redistribute it and/or
9  * modify it under the terms of the GNU Lesser General Public
10  * License as published by the Free Software Foundation; either
11  * version 2.1 of the License, or (at your option) any later version.
12  *
13  * FFmpeg is distributed in the hope that it will be useful,
14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16  * Lesser General Public License for more details.
17  *
18  * You should have received a copy of the GNU Lesser General Public
19  * License along with FFmpeg; if not, write to the Free Software
20  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21  */
22 
23 /**
24  * @file
25  * Libass subtitles burning filter.
26  *
27  * @see{http://www.matroska.org/technical/specs/subtitles/ssa.html}
28  */
29 
30 #include <ass/ass.h>
31 
32 #include "config.h"
33 #if CONFIG_SUBTITLES_FILTER
34 # include "libavcodec/avcodec.h"
35 # include "libavformat/avformat.h"
36 #endif
37 #include "libavutil/avstring.h"
38 #include "libavutil/imgutils.h"
39 #include "libavutil/opt.h"
40 #include "libavutil/parseutils.h"
41 #include "drawutils.h"
42 #include "avfilter.h"
43 #include "internal.h"
44 #include "formats.h"
45 #include "video.h"
46 
47 typedef struct {
48  const AVClass *class;
49  ASS_Library *library;
50  ASS_Renderer *renderer;
51  ASS_Track *track;
52  char *filename;
53  char *charenc;
55  uint8_t rgba_map[4];
56  int pix_step[4]; ///< steps per pixel for each plane of the main output
57  int original_w, original_h;
59 } AssContext;
60 
61 #define OFFSET(x) offsetof(AssContext, x)
62 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
63 
64 #define COMMON_OPTIONS \
65  {"filename", "set the filename of file to read", OFFSET(filename), AV_OPT_TYPE_STRING, {.str = NULL}, CHAR_MIN, CHAR_MAX, FLAGS }, \
66  {"f", "set the filename of file to read", OFFSET(filename), AV_OPT_TYPE_STRING, {.str = NULL}, CHAR_MIN, CHAR_MAX, FLAGS }, \
67  {"original_size", "set the size of the original video (used to scale fonts)", OFFSET(original_w), AV_OPT_TYPE_IMAGE_SIZE, {.str = NULL}, CHAR_MIN, CHAR_MAX, FLAGS }, \
68 
69 /* libass supports a log level ranging from 0 to 7 */
70 static const int ass_libavfilter_log_level_map[] = {
71  AV_LOG_QUIET, /* 0 */
72  AV_LOG_PANIC, /* 1 */
73  AV_LOG_FATAL, /* 2 */
74  AV_LOG_ERROR, /* 3 */
75  AV_LOG_WARNING, /* 4 */
76  AV_LOG_INFO, /* 5 */
77  AV_LOG_VERBOSE, /* 6 */
78  AV_LOG_DEBUG, /* 7 */
79 };
80 
81 static void ass_log(int ass_level, const char *fmt, va_list args, void *ctx)
82 {
83  int level = ass_libavfilter_log_level_map[ass_level];
84 
85  av_vlog(ctx, level, fmt, args);
86  av_log(ctx, level, "\n");
87 }
88 
89 static av_cold int init(AVFilterContext *ctx)
90 {
91  AssContext *ass = ctx->priv;
92 
93  if (!ass->filename) {
94  av_log(ctx, AV_LOG_ERROR, "No filename provided!\n");
95  return AVERROR(EINVAL);
96  }
97 
98  ass->library = ass_library_init();
99  if (!ass->library) {
100  av_log(ctx, AV_LOG_ERROR, "Could not initialize libass.\n");
101  return AVERROR(EINVAL);
102  }
103  ass_set_message_cb(ass->library, ass_log, ctx);
104 
105  ass->renderer = ass_renderer_init(ass->library);
106  if (!ass->renderer) {
107  av_log(ctx, AV_LOG_ERROR, "Could not initialize libass renderer.\n");
108  return AVERROR(EINVAL);
109  }
110 
111  return 0;
112 }
113 
114 static av_cold void uninit(AVFilterContext *ctx)
115 {
116  AssContext *ass = ctx->priv;
117 
118  if (ass->track)
119  ass_free_track(ass->track);
120  if (ass->renderer)
121  ass_renderer_done(ass->renderer);
122  if (ass->library)
123  ass_library_done(ass->library);
124 }
125 
127 {
129  return 0;
130 }
131 
132 static int config_input(AVFilterLink *inlink)
133 {
134  AssContext *ass = inlink->dst->priv;
135 
136  ff_draw_init(&ass->draw, inlink->format, 0);
137 
138  ass_set_frame_size (ass->renderer, inlink->w, inlink->h);
139  if (ass->original_w && ass->original_h)
140  ass_set_aspect_ratio(ass->renderer, (double)inlink->w / inlink->h,
141  (double)ass->original_w / ass->original_h);
142 
143  return 0;
144 }
145 
146 /* libass stores an RGBA color in the format RRGGBBTT, where TT is the transparency level */
147 #define AR(c) ( (c)>>24)
148 #define AG(c) (((c)>>16)&0xFF)
149 #define AB(c) (((c)>>8) &0xFF)
150 #define AA(c) ((0xFF-c) &0xFF)
151 
152 static void overlay_ass_image(AssContext *ass, AVFrame *picref,
153  const ASS_Image *image)
154 {
155  for (; image; image = image->next) {
156  uint8_t rgba_color[] = {AR(image->color), AG(image->color), AB(image->color), AA(image->color)};
158  ff_draw_color(&ass->draw, &color, rgba_color);
159  ff_blend_mask(&ass->draw, &color,
160  picref->data, picref->linesize,
161  picref->width, picref->height,
162  image->bitmap, image->stride, image->w, image->h,
163  3, 0, image->dst_x, image->dst_y);
164  }
165 }
166 
167 static int filter_frame(AVFilterLink *inlink, AVFrame *picref)
168 {
169  AVFilterContext *ctx = inlink->dst;
170  AVFilterLink *outlink = ctx->outputs[0];
171  AssContext *ass = ctx->priv;
172  int detect_change = 0;
173  double time_ms = picref->pts * av_q2d(inlink->time_base) * 1000;
174  ASS_Image *image = ass_render_frame(ass->renderer, ass->track,
175  time_ms, &detect_change);
176 
177  if (detect_change)
178  av_log(ctx, AV_LOG_DEBUG, "Change happened at time ms:%f\n", time_ms);
179 
180  overlay_ass_image(ass, picref, image);
181 
182  return ff_filter_frame(outlink, picref);
183 }
184 
185 static const AVFilterPad ass_inputs[] = {
186  {
187  .name = "default",
188  .type = AVMEDIA_TYPE_VIDEO,
189  .filter_frame = filter_frame,
190  .config_props = config_input,
191  .needs_writable = 1,
192  },
193  { NULL }
194 };
195 
196 static const AVFilterPad ass_outputs[] = {
197  {
198  .name = "default",
199  .type = AVMEDIA_TYPE_VIDEO,
200  },
201  { NULL }
202 };
203 
204 #if CONFIG_ASS_FILTER
205 
206 static const AVOption ass_options[] = {
208  {NULL},
209 };
210 
212 
213 static av_cold int init_ass(AVFilterContext *ctx)
214 {
215  AssContext *ass = ctx->priv;
216  int ret = init(ctx);
217 
218  if (ret < 0)
219  return ret;
220 
221  /* Initialize fonts */
222  ass_set_fonts(ass->renderer, NULL, NULL, 1, NULL, 1);
223 
224  ass->track = ass_read_file(ass->library, ass->filename, NULL);
225  if (!ass->track) {
226  av_log(ctx, AV_LOG_ERROR,
227  "Could not create a libass track when reading file '%s'\n",
228  ass->filename);
229  return AVERROR(EINVAL);
230  }
231  return 0;
232 }
233 
234 AVFilter ff_vf_ass = {
235  .name = "ass",
236  .description = NULL_IF_CONFIG_SMALL("Render ASS subtitles onto input video using the libass library."),
237  .priv_size = sizeof(AssContext),
238  .init = init_ass,
239  .uninit = uninit,
241  .inputs = ass_inputs,
242  .outputs = ass_outputs,
243  .priv_class = &ass_class,
244 };
245 #endif
246 
247 #if CONFIG_SUBTITLES_FILTER
248 
249 static const AVOption subtitles_options[] = {
251  {"charenc", "set input character encoding", OFFSET(charenc), AV_OPT_TYPE_STRING, {.str = NULL}, CHAR_MIN, CHAR_MAX, FLAGS},
252  {"stream_index", "set stream index", OFFSET(stream_index), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, FLAGS},
253  {"si", "set stream index", OFFSET(stream_index), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, FLAGS},
254  {NULL},
255 };
256 
257 static const char * const font_mimetypes[] = {
258  "application/x-truetype-font",
259  "application/vnd.ms-opentype",
260  "application/x-font-ttf",
261  NULL
262 };
263 
264 static int attachment_is_font(AVStream * st)
265 {
266  const AVDictionaryEntry *tag = NULL;
267  int n;
268 
269  tag = av_dict_get(st->metadata, "mimetype", NULL, AV_DICT_MATCH_CASE);
270 
271  if (tag) {
272  for (n = 0; font_mimetypes[n]; n++) {
273  if (av_strcasecmp(font_mimetypes[n], tag->value) == 0)
274  return 1;
275  }
276  }
277  return 0;
278 }
279 
280 AVFILTER_DEFINE_CLASS(subtitles);
281 
282 static av_cold int init_subtitles(AVFilterContext *ctx)
283 {
284  int j, ret, sid;
285  int k = 0;
286  AVDictionary *codec_opts = NULL;
287  AVFormatContext *fmt = NULL;
288  AVCodecContext *dec_ctx = NULL;
289  AVCodec *dec = NULL;
290  const AVCodecDescriptor *dec_desc;
291  AVStream *st;
292  AVPacket pkt;
293  AssContext *ass = ctx->priv;
294 
295  /* Init libass */
296  ret = init(ctx);
297  if (ret < 0)
298  return ret;
299  ass->track = ass_new_track(ass->library);
300  if (!ass->track) {
301  av_log(ctx, AV_LOG_ERROR, "Could not create a libass track\n");
302  return AVERROR(EINVAL);
303  }
304 
305  /* Open subtitles file */
306  ret = avformat_open_input(&fmt, ass->filename, NULL, NULL);
307  if (ret < 0) {
308  av_log(ctx, AV_LOG_ERROR, "Unable to open %s\n", ass->filename);
309  goto end;
310  }
311  ret = avformat_find_stream_info(fmt, NULL);
312  if (ret < 0)
313  goto end;
314 
315  /* Locate subtitles stream */
316  if (ass->stream_index < 0)
317  ret = av_find_best_stream(fmt, AVMEDIA_TYPE_SUBTITLE, -1, -1, NULL, 0);
318  else {
319  ret = -1;
320  if (ass->stream_index < fmt->nb_streams) {
321  for (j = 0; j < fmt->nb_streams; j++) {
322  if (fmt->streams[j]->codec->codec_type == AVMEDIA_TYPE_SUBTITLE) {
323  if (ass->stream_index == k) {
324  ret = j;
325  break;
326  }
327  k++;
328  }
329  }
330  }
331  }
332 
333  if (ret < 0) {
334  av_log(ctx, AV_LOG_ERROR, "Unable to locate subtitle stream in %s\n",
335  ass->filename);
336  goto end;
337  }
338  sid = ret;
339  st = fmt->streams[sid];
340 
341  /* Load attached fonts */
342  for (j = 0; j < fmt->nb_streams; j++) {
343  AVStream *st = fmt->streams[j];
345  attachment_is_font(st)) {
346  const AVDictionaryEntry *tag = NULL;
347  tag = av_dict_get(st->metadata, "filename", NULL,
349 
350  if (tag) {
351  av_log(ctx, AV_LOG_DEBUG, "Loading attached font: %s\n",
352  tag->value);
353  ass_add_font(ass->library, tag->value,
354  st->codec->extradata,
355  st->codec->extradata_size);
356  } else {
357  av_log(ctx, AV_LOG_WARNING,
358  "Font attachment has no filename, ignored.\n");
359  }
360  }
361  }
362 
363  /* Initialize fonts */
364  ass_set_fonts(ass->renderer, NULL, NULL, 1, NULL, 1);
365 
366  /* Open decoder */
367  dec_ctx = st->codec;
368  dec = avcodec_find_decoder(dec_ctx->codec_id);
369  if (!dec) {
370  av_log(ctx, AV_LOG_ERROR, "Failed to find subtitle codec %s\n",
371  avcodec_get_name(dec_ctx->codec_id));
372  return AVERROR(EINVAL);
373  }
374  dec_desc = avcodec_descriptor_get(dec_ctx->codec_id);
375  if (dec_desc && !(dec_desc->props & AV_CODEC_PROP_TEXT_SUB)) {
376  av_log(ctx, AV_LOG_ERROR,
377  "Only text based subtitles are currently supported\n");
378  return AVERROR_PATCHWELCOME;
379  }
380  if (ass->charenc)
381  av_dict_set(&codec_opts, "sub_charenc", ass->charenc, 0);
382  ret = avcodec_open2(dec_ctx, dec, &codec_opts);
383  if (ret < 0)
384  goto end;
385 
386  /* Decode subtitles and push them into the renderer (libass) */
387  if (dec_ctx->subtitle_header)
388  ass_process_codec_private(ass->track,
389  dec_ctx->subtitle_header,
390  dec_ctx->subtitle_header_size);
391  av_init_packet(&pkt);
392  pkt.data = NULL;
393  pkt.size = 0;
394  while (av_read_frame(fmt, &pkt) >= 0) {
395  int i, got_subtitle;
396  AVSubtitle sub = {0};
397 
398  if (pkt.stream_index == sid) {
399  ret = avcodec_decode_subtitle2(dec_ctx, &sub, &got_subtitle, &pkt);
400  if (ret < 0) {
401  av_log(ctx, AV_LOG_WARNING, "Error decoding: %s (ignored)\n",
402  av_err2str(ret));
403  } else if (got_subtitle) {
404  for (i = 0; i < sub.num_rects; i++) {
405  char *ass_line = sub.rects[i]->ass;
406  if (!ass_line)
407  break;
408  ass_process_data(ass->track, ass_line, strlen(ass_line));
409  }
410  }
411  }
412  av_free_packet(&pkt);
413  avsubtitle_free(&sub);
414  }
415 
416 end:
417  av_dict_free(&codec_opts);
418  if (dec_ctx)
419  avcodec_close(dec_ctx);
420  if (fmt)
421  avformat_close_input(&fmt);
422  return ret;
423 }
424 
425 AVFilter ff_vf_subtitles = {
426  .name = "subtitles",
427  .description = NULL_IF_CONFIG_SMALL("Render text subtitles onto input video using the libass library."),
428  .priv_size = sizeof(AssContext),
429  .init = init_subtitles,
430  .uninit = uninit,
432  .inputs = ass_inputs,
433  .outputs = ass_outputs,
434  .priv_class = &subtitles_class,
435 };
436 #endif