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 #include "config_components.h"
34 #if CONFIG_SUBTITLES_FILTER
35 # include "libavcodec/avcodec.h"
36 # include "libavformat/avformat.h"
37 #endif
38 #include "libavutil/avstring.h"
39 #include "libavutil/imgutils.h"
40 #include "libavutil/opt.h"
41 #include "libavutil/parseutils.h"
42 #include "drawutils.h"
43 #include "avfilter.h"
44 #include "internal.h"
45 #include "formats.h"
46 #include "video.h"
47
48 typedef struct AssContext {
49 const AVClass *class;
50 ASS_Library *library;
51 ASS_Renderer *renderer;
52 ASS_Track *track;
53 char *filename;
54 char *fontsdir;
55 char *charenc;
56 char *force_style;
57 int stream_index;
58 int alpha;
59 uint8_t rgba_map[4];
60 int pix_step[4]; ///< steps per pixel for each plane of the main output
61 int original_w, original_h;
62 int shaping;
63 FFDrawContext draw;
64 } AssContext;
65
66 #define OFFSET(x) offsetof(AssContext, x)
67 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
68
69 #define COMMON_OPTIONS \
70 {"filename", "set the filename of file to read", OFFSET(filename), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, FLAGS }, \
71 {"f", "set the filename of file to read", OFFSET(filename), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, FLAGS }, \
72 {"original_size", "set the size of the original video (used to scale fonts)", OFFSET(original_w), AV_OPT_TYPE_IMAGE_SIZE, {.str = NULL}, 0, 0, FLAGS }, \
73 {"fontsdir", "set the directory containing the fonts to read", OFFSET(fontsdir), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, FLAGS }, \
74 {"alpha", "enable processing of alpha channel", OFFSET(alpha), AV_OPT_TYPE_BOOL, {.i64 = 0 }, 0, 1, FLAGS }, \
75
76 /* libass supports a log level ranging from 0 to 7 */
77 static const int ass_libavfilter_log_level_map[] = {
78 [0] = AV_LOG_FATAL, /* MSGL_FATAL */
79 [1] = AV_LOG_ERROR, /* MSGL_ERR */
80 [2] = AV_LOG_WARNING, /* MSGL_WARN */
81 [3] = AV_LOG_WARNING, /* <undefined> */
82 [4] = AV_LOG_INFO, /* MSGL_INFO */
83 [5] = AV_LOG_INFO, /* <undefined> */
84 [6] = AV_LOG_VERBOSE, /* MSGL_V */
85 [7] = AV_LOG_DEBUG, /* MSGL_DBG2 */
86 };
87
ass_log(int ass_level,const char * fmt,va_list args,void * ctx)88 static void ass_log(int ass_level, const char *fmt, va_list args, void *ctx)
89 {
90 const int ass_level_clip = av_clip(ass_level, 0,
91 FF_ARRAY_ELEMS(ass_libavfilter_log_level_map) - 1);
92 const int level = ass_libavfilter_log_level_map[ass_level_clip];
93
94 av_vlog(ctx, level, fmt, args);
95 av_log(ctx, level, "\n");
96 }
97
init(AVFilterContext * ctx)98 static av_cold int init(AVFilterContext *ctx)
99 {
100 AssContext *ass = ctx->priv;
101
102 if (!ass->filename) {
103 av_log(ctx, AV_LOG_ERROR, "No filename provided!\n");
104 return AVERROR(EINVAL);
105 }
106
107 ass->library = ass_library_init();
108 if (!ass->library) {
109 av_log(ctx, AV_LOG_ERROR, "Could not initialize libass.\n");
110 return AVERROR(EINVAL);
111 }
112 ass_set_message_cb(ass->library, ass_log, ctx);
113
114 ass_set_fonts_dir(ass->library, ass->fontsdir);
115 ass_set_extract_fonts(ass->library, 1);
116
117 ass->renderer = ass_renderer_init(ass->library);
118 if (!ass->renderer) {
119 av_log(ctx, AV_LOG_ERROR, "Could not initialize libass renderer.\n");
120 return AVERROR(EINVAL);
121 }
122
123 return 0;
124 }
125
uninit(AVFilterContext * ctx)126 static av_cold void uninit(AVFilterContext *ctx)
127 {
128 AssContext *ass = ctx->priv;
129
130 if (ass->track)
131 ass_free_track(ass->track);
132 if (ass->renderer)
133 ass_renderer_done(ass->renderer);
134 if (ass->library)
135 ass_library_done(ass->library);
136 }
137
query_formats(AVFilterContext * ctx)138 static int query_formats(AVFilterContext *ctx)
139 {
140 return ff_set_common_formats(ctx, ff_draw_supported_pixel_formats(0));
141 }
142
config_input(AVFilterLink * inlink)143 static int config_input(AVFilterLink *inlink)
144 {
145 AssContext *ass = inlink->dst->priv;
146
147 ff_draw_init(&ass->draw, inlink->format, ass->alpha ? FF_DRAW_PROCESS_ALPHA : 0);
148
149 ass_set_frame_size (ass->renderer, inlink->w, inlink->h);
150 if (ass->original_w && ass->original_h) {
151 ass_set_pixel_aspect(ass->renderer, (double)inlink->w / inlink->h /
152 ((double)ass->original_w / ass->original_h));
153 ass_set_storage_size(ass->renderer, ass->original_w, ass->original_h);
154 } else
155 ass_set_storage_size(ass->renderer, inlink->w, inlink->h);
156
157 if (ass->shaping != -1)
158 ass_set_shaper(ass->renderer, ass->shaping);
159
160 return 0;
161 }
162
163 /* libass stores an RGBA color in the format RRGGBBTT, where TT is the transparency level */
164 #define AR(c) ( (c)>>24)
165 #define AG(c) (((c)>>16)&0xFF)
166 #define AB(c) (((c)>>8) &0xFF)
167 #define AA(c) ((0xFF-(c)) &0xFF)
168
overlay_ass_image(AssContext * ass,AVFrame * picref,const ASS_Image * image)169 static void overlay_ass_image(AssContext *ass, AVFrame *picref,
170 const ASS_Image *image)
171 {
172 for (; image; image = image->next) {
173 uint8_t rgba_color[] = {AR(image->color), AG(image->color), AB(image->color), AA(image->color)};
174 FFDrawColor color;
175 ff_draw_color(&ass->draw, &color, rgba_color);
176 ff_blend_mask(&ass->draw, &color,
177 picref->data, picref->linesize,
178 picref->width, picref->height,
179 image->bitmap, image->stride, image->w, image->h,
180 3, 0, image->dst_x, image->dst_y);
181 }
182 }
183
filter_frame(AVFilterLink * inlink,AVFrame * picref)184 static int filter_frame(AVFilterLink *inlink, AVFrame *picref)
185 {
186 AVFilterContext *ctx = inlink->dst;
187 AVFilterLink *outlink = ctx->outputs[0];
188 AssContext *ass = ctx->priv;
189 int detect_change = 0;
190 double time_ms = picref->pts * av_q2d(inlink->time_base) * 1000;
191 ASS_Image *image = ass_render_frame(ass->renderer, ass->track,
192 time_ms, &detect_change);
193
194 if (detect_change)
195 av_log(ctx, AV_LOG_DEBUG, "Change happened at time ms:%f\n", time_ms);
196
197 overlay_ass_image(ass, picref, image);
198
199 return ff_filter_frame(outlink, picref);
200 }
201
202 static const AVFilterPad ass_inputs[] = {
203 {
204 .name = "default",
205 .type = AVMEDIA_TYPE_VIDEO,
206 .flags = AVFILTERPAD_FLAG_NEEDS_WRITABLE,
207 .filter_frame = filter_frame,
208 .config_props = config_input,
209 },
210 };
211
212 static const AVFilterPad ass_outputs[] = {
213 {
214 .name = "default",
215 .type = AVMEDIA_TYPE_VIDEO,
216 },
217 };
218
219 #if CONFIG_ASS_FILTER
220
221 static const AVOption ass_options[] = {
222 COMMON_OPTIONS
223 {"shaping", "set shaping engine", OFFSET(shaping), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, FLAGS, "shaping_mode"},
224 {"auto", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = -1}, INT_MIN, INT_MAX, FLAGS, "shaping_mode"},
225 {"simple", "simple shaping", 0, AV_OPT_TYPE_CONST, {.i64 = ASS_SHAPING_SIMPLE}, INT_MIN, INT_MAX, FLAGS, "shaping_mode"},
226 {"complex", "complex shaping", 0, AV_OPT_TYPE_CONST, {.i64 = ASS_SHAPING_COMPLEX}, INT_MIN, INT_MAX, FLAGS, "shaping_mode"},
227 {NULL},
228 };
229
230 AVFILTER_DEFINE_CLASS(ass);
231
init_ass(AVFilterContext * ctx)232 static av_cold int init_ass(AVFilterContext *ctx)
233 {
234 AssContext *ass = ctx->priv;
235 int ret = init(ctx);
236
237 if (ret < 0)
238 return ret;
239
240 /* Initialize fonts */
241 ass_set_fonts(ass->renderer, NULL, NULL, 1, NULL, 1);
242
243 ass->track = ass_read_file(ass->library, ass->filename, NULL);
244 if (!ass->track) {
245 av_log(ctx, AV_LOG_ERROR,
246 "Could not create a libass track when reading file '%s'\n",
247 ass->filename);
248 return AVERROR(EINVAL);
249 }
250 return 0;
251 }
252
253 const AVFilter ff_vf_ass = {
254 .name = "ass",
255 .description = NULL_IF_CONFIG_SMALL("Render ASS subtitles onto input video using the libass library."),
256 .priv_size = sizeof(AssContext),
257 .init = init_ass,
258 .uninit = uninit,
259 FILTER_INPUTS(ass_inputs),
260 FILTER_OUTPUTS(ass_outputs),
261 FILTER_QUERY_FUNC(query_formats),
262 .priv_class = &ass_class,
263 };
264 #endif
265
266 #if CONFIG_SUBTITLES_FILTER
267
268 static const AVOption subtitles_options[] = {
269 COMMON_OPTIONS
270 {"charenc", "set input character encoding", OFFSET(charenc), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, FLAGS},
271 {"stream_index", "set stream index", OFFSET(stream_index), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, FLAGS},
272 {"si", "set stream index", OFFSET(stream_index), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, FLAGS},
273 {"force_style", "force subtitle style", OFFSET(force_style), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, FLAGS},
274 {NULL},
275 };
276
277 static const char * const font_mimetypes[] = {
278 "font/ttf",
279 "font/otf",
280 "font/sfnt",
281 "font/woff",
282 "font/woff2",
283 "application/font-sfnt",
284 "application/font-woff",
285 "application/x-truetype-font",
286 "application/vnd.ms-opentype",
287 "application/x-font-ttf",
288 NULL
289 };
290
attachment_is_font(AVStream * st)291 static int attachment_is_font(AVStream * st)
292 {
293 const AVDictionaryEntry *tag = NULL;
294 int n;
295
296 tag = av_dict_get(st->metadata, "mimetype", NULL, AV_DICT_MATCH_CASE);
297
298 if (tag) {
299 for (n = 0; font_mimetypes[n]; n++) {
300 if (av_strcasecmp(font_mimetypes[n], tag->value) == 0)
301 return 1;
302 }
303 }
304 return 0;
305 }
306
307 AVFILTER_DEFINE_CLASS(subtitles);
308
init_subtitles(AVFilterContext * ctx)309 static av_cold int init_subtitles(AVFilterContext *ctx)
310 {
311 int j, ret, sid;
312 int k = 0;
313 AVDictionary *codec_opts = NULL;
314 AVFormatContext *fmt = NULL;
315 AVCodecContext *dec_ctx = NULL;
316 const AVCodec *dec;
317 const AVCodecDescriptor *dec_desc;
318 AVStream *st;
319 AVPacket pkt;
320 AssContext *ass = ctx->priv;
321
322 /* Init libass */
323 ret = init(ctx);
324 if (ret < 0)
325 return ret;
326 ass->track = ass_new_track(ass->library);
327 if (!ass->track) {
328 av_log(ctx, AV_LOG_ERROR, "Could not create a libass track\n");
329 return AVERROR(EINVAL);
330 }
331
332 /* Open subtitles file */
333 ret = avformat_open_input(&fmt, ass->filename, NULL, NULL);
334 if (ret < 0) {
335 av_log(ctx, AV_LOG_ERROR, "Unable to open %s\n", ass->filename);
336 goto end;
337 }
338 ret = avformat_find_stream_info(fmt, NULL);
339 if (ret < 0)
340 goto end;
341
342 /* Locate subtitles stream */
343 if (ass->stream_index < 0)
344 ret = av_find_best_stream(fmt, AVMEDIA_TYPE_SUBTITLE, -1, -1, NULL, 0);
345 else {
346 ret = -1;
347 if (ass->stream_index < fmt->nb_streams) {
348 for (j = 0; j < fmt->nb_streams; j++) {
349 if (fmt->streams[j]->codecpar->codec_type == AVMEDIA_TYPE_SUBTITLE) {
350 if (ass->stream_index == k) {
351 ret = j;
352 break;
353 }
354 k++;
355 }
356 }
357 }
358 }
359
360 if (ret < 0) {
361 av_log(ctx, AV_LOG_ERROR, "Unable to locate subtitle stream in %s\n",
362 ass->filename);
363 goto end;
364 }
365 sid = ret;
366 st = fmt->streams[sid];
367
368 /* Load attached fonts */
369 for (j = 0; j < fmt->nb_streams; j++) {
370 AVStream *st = fmt->streams[j];
371 if (st->codecpar->codec_type == AVMEDIA_TYPE_ATTACHMENT &&
372 attachment_is_font(st)) {
373 const AVDictionaryEntry *tag = NULL;
374 tag = av_dict_get(st->metadata, "filename", NULL,
375 AV_DICT_MATCH_CASE);
376
377 if (tag) {
378 av_log(ctx, AV_LOG_DEBUG, "Loading attached font: %s\n",
379 tag->value);
380 ass_add_font(ass->library, tag->value,
381 st->codecpar->extradata,
382 st->codecpar->extradata_size);
383 } else {
384 av_log(ctx, AV_LOG_WARNING,
385 "Font attachment has no filename, ignored.\n");
386 }
387 }
388 }
389
390 /* Initialize fonts */
391 ass_set_fonts(ass->renderer, NULL, NULL, 1, NULL, 1);
392
393 /* Open decoder */
394 dec = avcodec_find_decoder(st->codecpar->codec_id);
395 if (!dec) {
396 av_log(ctx, AV_LOG_ERROR, "Failed to find subtitle codec %s\n",
397 avcodec_get_name(st->codecpar->codec_id));
398 ret = AVERROR_DECODER_NOT_FOUND;
399 goto end;
400 }
401 dec_desc = avcodec_descriptor_get(st->codecpar->codec_id);
402 if (dec_desc && !(dec_desc->props & AV_CODEC_PROP_TEXT_SUB)) {
403 av_log(ctx, AV_LOG_ERROR,
404 "Only text based subtitles are currently supported\n");
405 ret = AVERROR_PATCHWELCOME;
406 goto end;
407 }
408 if (ass->charenc)
409 av_dict_set(&codec_opts, "sub_charenc", ass->charenc, 0);
410
411 dec_ctx = avcodec_alloc_context3(dec);
412 if (!dec_ctx) {
413 ret = AVERROR(ENOMEM);
414 goto end;
415 }
416
417 ret = avcodec_parameters_to_context(dec_ctx, st->codecpar);
418 if (ret < 0)
419 goto end;
420
421 /*
422 * This is required by the decoding process in order to rescale the
423 * timestamps: in the current API the decoded subtitles have their pts
424 * expressed in AV_TIME_BASE, and thus the lavc internals need to know the
425 * stream time base in order to achieve the rescaling.
426 *
427 * That API is old and needs to be reworked to match behaviour with A/V.
428 */
429 dec_ctx->pkt_timebase = st->time_base;
430
431 ret = avcodec_open2(dec_ctx, NULL, &codec_opts);
432 if (ret < 0)
433 goto end;
434
435 if (ass->force_style) {
436 char **list = NULL;
437 char *temp = NULL;
438 char *ptr = av_strtok(ass->force_style, ",", &temp);
439 int i = 0;
440 while (ptr) {
441 av_dynarray_add(&list, &i, ptr);
442 if (!list) {
443 ret = AVERROR(ENOMEM);
444 goto end;
445 }
446 ptr = av_strtok(NULL, ",", &temp);
447 }
448 av_dynarray_add(&list, &i, NULL);
449 if (!list) {
450 ret = AVERROR(ENOMEM);
451 goto end;
452 }
453 ass_set_style_overrides(ass->library, list);
454 av_free(list);
455 }
456 /* Decode subtitles and push them into the renderer (libass) */
457 if (dec_ctx->subtitle_header)
458 ass_process_codec_private(ass->track,
459 dec_ctx->subtitle_header,
460 dec_ctx->subtitle_header_size);
461 while (av_read_frame(fmt, &pkt) >= 0) {
462 int i, got_subtitle;
463 AVSubtitle sub = {0};
464
465 if (pkt.stream_index == sid) {
466 ret = avcodec_decode_subtitle2(dec_ctx, &sub, &got_subtitle, &pkt);
467 if (ret < 0) {
468 av_log(ctx, AV_LOG_WARNING, "Error decoding: %s (ignored)\n",
469 av_err2str(ret));
470 } else if (got_subtitle) {
471 const int64_t start_time = av_rescale_q(sub.pts, AV_TIME_BASE_Q, av_make_q(1, 1000));
472 const int64_t duration = sub.end_display_time;
473 for (i = 0; i < sub.num_rects; i++) {
474 char *ass_line = sub.rects[i]->ass;
475 if (!ass_line)
476 break;
477 ass_process_chunk(ass->track, ass_line, strlen(ass_line),
478 start_time, duration);
479 }
480 }
481 }
482 av_packet_unref(&pkt);
483 avsubtitle_free(&sub);
484 }
485
486 end:
487 av_dict_free(&codec_opts);
488 avcodec_free_context(&dec_ctx);
489 avformat_close_input(&fmt);
490 return ret;
491 }
492
493 const AVFilter ff_vf_subtitles = {
494 .name = "subtitles",
495 .description = NULL_IF_CONFIG_SMALL("Render text subtitles onto input video using the libass library."),
496 .priv_size = sizeof(AssContext),
497 .init = init_subtitles,
498 .uninit = uninit,
499 FILTER_INPUTS(ass_inputs),
500 FILTER_OUTPUTS(ass_outputs),
501 FILTER_QUERY_FUNC(query_formats),
502 .priv_class = &subtitles_class,
503 };
504 #endif
505