/src/ffmpeg/libavcodec/ttmlenc.c
Line | Count | Source |
1 | | /* |
2 | | * TTML subtitle encoder |
3 | | * Copyright (c) 2020 24i |
4 | | * |
5 | | * This file is part of FFmpeg. |
6 | | * |
7 | | * FFmpeg is free software; you can redistribute it and/or |
8 | | * modify it under the terms of the GNU Lesser General Public |
9 | | * License as published by the Free Software Foundation; either |
10 | | * version 2.1 of the License, or (at your option) any later version. |
11 | | * |
12 | | * FFmpeg is distributed in the hope that it will be useful, |
13 | | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
14 | | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
15 | | * Lesser General Public License for more details. |
16 | | * |
17 | | * You should have received a copy of the GNU Lesser General Public |
18 | | * License along with FFmpeg; if not, write to the Free Software |
19 | | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
20 | | */ |
21 | | |
22 | | /** |
23 | | * @file |
24 | | * TTML subtitle encoder |
25 | | * @see https://www.w3.org/TR/ttml1/ |
26 | | * @see https://www.w3.org/TR/ttml2/ |
27 | | * @see https://www.w3.org/TR/ttml-imsc/rec |
28 | | */ |
29 | | |
30 | | #include "avcodec.h" |
31 | | #include "codec_internal.h" |
32 | | #include "libavutil/bprint.h" |
33 | | #include "libavutil/internal.h" |
34 | | #include "libavutil/mem.h" |
35 | | #include "ass_split.h" |
36 | | #include "ttmlenc.h" |
37 | | |
38 | | typedef struct { |
39 | | AVCodecContext *avctx; |
40 | | ASSSplitContext *ass_ctx; |
41 | | AVBPrint buffer; |
42 | | } TTMLContext; |
43 | | |
44 | | static void ttml_text_cb(void *priv, const char *text, int len) |
45 | 0 | { |
46 | 0 | TTMLContext *s = priv; |
47 | 0 | AVBPrint cur_line; |
48 | 0 | AVBPrint *buffer = &s->buffer; |
49 | |
|
50 | 0 | av_bprint_init(&cur_line, len, AV_BPRINT_SIZE_UNLIMITED); |
51 | |
|
52 | 0 | av_bprint_append_data(&cur_line, text, len); |
53 | 0 | if (!av_bprint_is_complete(&cur_line)) { |
54 | 0 | av_log(s->avctx, AV_LOG_ERROR, |
55 | 0 | "Failed to move the current subtitle dialog to AVBPrint!\n"); |
56 | 0 | av_bprint_finalize(&cur_line, NULL); |
57 | 0 | return; |
58 | 0 | } |
59 | | |
60 | | |
61 | 0 | av_bprint_escape(buffer, cur_line.str, NULL, AV_ESCAPE_MODE_XML, |
62 | 0 | 0); |
63 | |
|
64 | 0 | av_bprint_finalize(&cur_line, NULL); |
65 | 0 | } |
66 | | |
67 | | static void ttml_new_line_cb(void *priv, int forced) |
68 | 0 | { |
69 | 0 | TTMLContext *s = priv; |
70 | |
|
71 | 0 | av_bprintf(&s->buffer, "<br/>"); |
72 | 0 | } |
73 | | |
74 | | static const ASSCodesCallbacks ttml_callbacks = { |
75 | | .text = ttml_text_cb, |
76 | | .new_line = ttml_new_line_cb, |
77 | | }; |
78 | | |
79 | | static int ttml_encode_frame(AVCodecContext *avctx, uint8_t *buf, |
80 | | int bufsize, const AVSubtitle *sub) |
81 | 0 | { |
82 | 0 | TTMLContext *s = avctx->priv_data; |
83 | 0 | ASSDialog *dialog; |
84 | 0 | int i; |
85 | |
|
86 | 0 | av_bprint_init_for_buffer(&s->buffer, buf, bufsize); |
87 | |
|
88 | 0 | for (i=0; i<sub->num_rects; i++) { |
89 | 0 | const char *ass = sub->rects[i]->ass; |
90 | 0 | int ret; |
91 | |
|
92 | 0 | if (sub->rects[i]->type != SUBTITLE_ASS) { |
93 | 0 | av_log(avctx, AV_LOG_ERROR, "Only SUBTITLE_ASS type supported.\n"); |
94 | 0 | return AVERROR(EINVAL); |
95 | 0 | } |
96 | | |
97 | 0 | dialog = ff_ass_split_dialog(s->ass_ctx, ass); |
98 | 0 | if (!dialog) |
99 | 0 | return AVERROR(ENOMEM); |
100 | | |
101 | 0 | if (dialog->style) { |
102 | 0 | av_bprintf(&s->buffer, "<span region=\""); |
103 | 0 | av_bprint_escape(&s->buffer, dialog->style, NULL, |
104 | 0 | AV_ESCAPE_MODE_XML, |
105 | 0 | AV_ESCAPE_FLAG_XML_DOUBLE_QUOTES); |
106 | 0 | av_bprintf(&s->buffer, "\">"); |
107 | 0 | } |
108 | |
|
109 | 0 | ret = ff_ass_split_override_codes(&ttml_callbacks, s, dialog->text); |
110 | 0 | if (ret < 0) { |
111 | 0 | int log_level = (ret != AVERROR_INVALIDDATA || |
112 | 0 | avctx->err_recognition & AV_EF_EXPLODE) ? |
113 | 0 | AV_LOG_ERROR : AV_LOG_WARNING; |
114 | 0 | av_log(avctx, log_level, |
115 | 0 | "Splitting received ASS dialog text %s failed: %s\n", |
116 | 0 | dialog->text, |
117 | 0 | av_err2str(ret)); |
118 | |
|
119 | 0 | if (log_level == AV_LOG_ERROR) { |
120 | 0 | ff_ass_free_dialog(&dialog); |
121 | 0 | return ret; |
122 | 0 | } |
123 | 0 | } |
124 | | |
125 | 0 | if (dialog->style) |
126 | 0 | av_bprintf(&s->buffer, "</span>"); |
127 | |
|
128 | 0 | ff_ass_free_dialog(&dialog); |
129 | 0 | } |
130 | | |
131 | 0 | if (!s->buffer.len) |
132 | 0 | return 0; |
133 | 0 | if (!av_bprint_is_complete(&s->buffer)) { |
134 | 0 | av_log(avctx, AV_LOG_ERROR, "Buffer too small for TTML event.\n"); |
135 | 0 | return AVERROR_BUFFER_TOO_SMALL; |
136 | 0 | } |
137 | | |
138 | 0 | return s->buffer.len; |
139 | 0 | } |
140 | | |
141 | | static av_cold int ttml_encode_close(AVCodecContext *avctx) |
142 | 0 | { |
143 | 0 | TTMLContext *s = avctx->priv_data; |
144 | |
|
145 | 0 | ff_ass_split_free(s->ass_ctx); |
146 | |
|
147 | 0 | return 0; |
148 | 0 | } |
149 | | |
150 | | static const char *ttml_get_display_alignment(int alignment) |
151 | 0 | { |
152 | 0 | switch (alignment) { |
153 | 0 | case 1: |
154 | 0 | case 2: |
155 | 0 | case 3: |
156 | 0 | return "after"; |
157 | 0 | case 4: |
158 | 0 | case 5: |
159 | 0 | case 6: |
160 | 0 | return "center"; |
161 | 0 | case 7: |
162 | 0 | case 8: |
163 | 0 | case 9: |
164 | 0 | return "before"; |
165 | 0 | default: |
166 | 0 | return NULL; |
167 | 0 | } |
168 | 0 | } |
169 | | |
170 | | static const char *ttml_get_text_alignment(int alignment) |
171 | 0 | { |
172 | 0 | switch (alignment) { |
173 | 0 | case 1: |
174 | 0 | case 4: |
175 | 0 | case 7: |
176 | 0 | return "left"; |
177 | 0 | case 2: |
178 | 0 | case 5: |
179 | 0 | case 8: |
180 | 0 | return "center"; |
181 | 0 | case 3: |
182 | 0 | case 6: |
183 | 0 | case 9: |
184 | 0 | return "right"; |
185 | 0 | default: |
186 | 0 | return NULL; |
187 | 0 | } |
188 | 0 | } |
189 | | |
190 | | static void ttml_get_origin(ASSScriptInfo script_info, ASSStyle style, |
191 | | int *origin_left, int *origin_top) |
192 | 0 | { |
193 | 0 | *origin_left = av_rescale(style.margin_l, 100, script_info.play_res_x); |
194 | 0 | *origin_top = |
195 | 0 | av_rescale((style.alignment >= 7) ? style.margin_v : 0, |
196 | 0 | 100, script_info.play_res_y); |
197 | 0 | } |
198 | | |
199 | | static void ttml_get_extent(ASSScriptInfo script_info, ASSStyle style, |
200 | | int *width, int *height) |
201 | 0 | { |
202 | 0 | *width = av_rescale(script_info.play_res_x - style.margin_r, |
203 | 0 | 100, script_info.play_res_x); |
204 | 0 | *height = av_rescale((style.alignment <= 3) ? |
205 | 0 | script_info.play_res_y - style.margin_v : |
206 | 0 | script_info.play_res_y, |
207 | 0 | 100, script_info.play_res_y); |
208 | 0 | } |
209 | | |
210 | | static int ttml_write_region(AVCodecContext *avctx, AVBPrint *buf, |
211 | | ASSScriptInfo script_info, ASSStyle style) |
212 | 0 | { |
213 | 0 | const char *display_alignment = NULL; |
214 | 0 | const char *text_alignment = NULL; |
215 | 0 | int origin_left = 0; |
216 | 0 | int origin_top = 0; |
217 | 0 | int width = 0; |
218 | 0 | int height = 0; |
219 | |
|
220 | 0 | if (!style.name) { |
221 | 0 | av_log(avctx, AV_LOG_ERROR, "Subtitle style name not set!\n"); |
222 | 0 | return AVERROR_INVALIDDATA; |
223 | 0 | } |
224 | | |
225 | 0 | if (style.font_size < 0) { |
226 | 0 | av_log(avctx, AV_LOG_ERROR, "Invalid font size for TTML: %d!\n", |
227 | 0 | style.font_size); |
228 | 0 | return AVERROR_INVALIDDATA; |
229 | 0 | } |
230 | | |
231 | 0 | if (style.margin_l < 0 || style.margin_r < 0 || style.margin_v < 0) { |
232 | 0 | av_log(avctx, AV_LOG_ERROR, |
233 | 0 | "One or more negative margin values in subtitle style: " |
234 | 0 | "left: %d, right: %d, vertical: %d!\n", |
235 | 0 | style.margin_l, style.margin_r, style.margin_v); |
236 | 0 | return AVERROR_INVALIDDATA; |
237 | 0 | } |
238 | | |
239 | 0 | display_alignment = ttml_get_display_alignment(style.alignment); |
240 | 0 | text_alignment = ttml_get_text_alignment(style.alignment); |
241 | 0 | if (!display_alignment || !text_alignment) { |
242 | 0 | av_log(avctx, AV_LOG_ERROR, |
243 | 0 | "Failed to convert ASS style alignment %d of style %s to " |
244 | 0 | "TTML display and text alignment!\n", |
245 | 0 | style.alignment, |
246 | 0 | style.name); |
247 | 0 | return AVERROR_INVALIDDATA; |
248 | 0 | } |
249 | | |
250 | 0 | ttml_get_origin(script_info, style, &origin_left, &origin_top); |
251 | 0 | ttml_get_extent(script_info, style, &width, &height); |
252 | |
|
253 | 0 | av_bprintf(buf, " <region xml:id=\""); |
254 | 0 | av_bprint_escape(buf, style.name, NULL, AV_ESCAPE_MODE_XML, |
255 | 0 | AV_ESCAPE_FLAG_XML_DOUBLE_QUOTES); |
256 | 0 | av_bprintf(buf, "\"\n"); |
257 | |
|
258 | 0 | av_bprintf(buf, " tts:origin=\"%d%% %d%%\"\n", |
259 | 0 | origin_left, origin_top); |
260 | 0 | av_bprintf(buf, " tts:extent=\"%d%% %d%%\"\n", |
261 | 0 | width, height); |
262 | |
|
263 | 0 | av_bprintf(buf, " tts:displayAlign=\""); |
264 | 0 | av_bprint_escape(buf, display_alignment, NULL, AV_ESCAPE_MODE_XML, |
265 | 0 | AV_ESCAPE_FLAG_XML_DOUBLE_QUOTES); |
266 | 0 | av_bprintf(buf, "\"\n"); |
267 | |
|
268 | 0 | av_bprintf(buf, " tts:textAlign=\""); |
269 | 0 | av_bprint_escape(buf, text_alignment, NULL, AV_ESCAPE_MODE_XML, |
270 | 0 | AV_ESCAPE_FLAG_XML_DOUBLE_QUOTES); |
271 | 0 | av_bprintf(buf, "\"\n"); |
272 | | |
273 | | // if we set cell resolution to our script reference resolution, |
274 | | // then a single line is a single "point" on our canvas. Thus, by setting |
275 | | // our font size to font size in cells, we should gain a similar enough |
276 | | // scale without resorting to explicit pixel based font sizing, which is |
277 | | // frowned upon in the TTML community. |
278 | 0 | av_bprintf(buf, " tts:fontSize=\"%dc\"\n", |
279 | 0 | style.font_size); |
280 | |
|
281 | 0 | if (style.font_name) { |
282 | 0 | av_bprintf(buf, " tts:fontFamily=\""); |
283 | 0 | av_bprint_escape(buf, style.font_name, NULL, AV_ESCAPE_MODE_XML, |
284 | 0 | AV_ESCAPE_FLAG_XML_DOUBLE_QUOTES); |
285 | 0 | av_bprintf(buf, "\"\n"); |
286 | 0 | } |
287 | |
|
288 | 0 | av_bprintf(buf, " tts:overflow=\"visible\" />\n"); |
289 | |
|
290 | 0 | return 0; |
291 | 0 | } |
292 | | |
293 | | static int ttml_write_header_content(AVCodecContext *avctx) |
294 | 0 | { |
295 | 0 | TTMLContext *s = avctx->priv_data; |
296 | 0 | ASS *ass = (ASS *)s->ass_ctx; |
297 | 0 | ASSScriptInfo script_info = ass->script_info; |
298 | 0 | const size_t base_extradata_size = TTMLENC_EXTRADATA_SIGNATURE_SIZE + 1 + |
299 | 0 | AV_INPUT_BUFFER_PADDING_SIZE; |
300 | 0 | size_t additional_extradata_size = 0; |
301 | 0 | int ret; |
302 | |
|
303 | 0 | if (script_info.play_res_x <= 0 || script_info.play_res_y <= 0) { |
304 | 0 | av_log(avctx, AV_LOG_ERROR, |
305 | 0 | "Invalid subtitle reference resolution %dx%d!\n", |
306 | 0 | script_info.play_res_x, script_info.play_res_y); |
307 | 0 | return AVERROR_INVALIDDATA; |
308 | 0 | } |
309 | | |
310 | 0 | av_bprint_init(&s->buffer, 0, INT_MAX - base_extradata_size); |
311 | | |
312 | | // write the first string in extradata, attributes in the base "tt" element. |
313 | 0 | av_bprintf(&s->buffer, TTML_DEFAULT_NAMESPACING); |
314 | | // the cell resolution is in character cells, so not exactly 1:1 against |
315 | | // a pixel based resolution, but as the tts:extent in the root |
316 | | // "tt" element is frowned upon (and disallowed in the EBU-TT profile), |
317 | | // we mimic the reference resolution by setting it as the cell resolution. |
318 | 0 | av_bprintf(&s->buffer, " ttp:cellResolution=\"%d %d\"\n", |
319 | 0 | script_info.play_res_x, script_info.play_res_y); |
320 | 0 | av_bprint_chars(&s->buffer, '\0', 1); |
321 | | |
322 | | // write the second string in extradata, head element containing the styles |
323 | 0 | av_bprintf(&s->buffer, " <head>\n"); |
324 | 0 | av_bprintf(&s->buffer, " <layout>\n"); |
325 | |
|
326 | 0 | for (int i = 0; i < ass->styles_count; i++) { |
327 | 0 | ret = ttml_write_region(avctx, &s->buffer, script_info, |
328 | 0 | ass->styles[i]); |
329 | 0 | if (ret < 0) |
330 | 0 | goto fail; |
331 | 0 | } |
332 | | |
333 | 0 | av_bprintf(&s->buffer, " </layout>\n"); |
334 | 0 | av_bprintf(&s->buffer, " </head>\n"); |
335 | 0 | av_bprint_chars(&s->buffer, '\0', 1); |
336 | |
|
337 | 0 | if (!av_bprint_is_complete(&s->buffer)) { |
338 | 0 | ret = AVERROR(ENOMEM); |
339 | 0 | goto fail; |
340 | 0 | } |
341 | | |
342 | 0 | additional_extradata_size = s->buffer.len; |
343 | |
|
344 | 0 | if (!(avctx->extradata = |
345 | 0 | av_mallocz(base_extradata_size + additional_extradata_size))) { |
346 | 0 | ret = AVERROR(ENOMEM); |
347 | 0 | goto fail; |
348 | 0 | } |
349 | | |
350 | 0 | avctx->extradata_size = |
351 | 0 | TTMLENC_EXTRADATA_SIGNATURE_SIZE + additional_extradata_size; |
352 | 0 | memcpy(avctx->extradata, TTMLENC_EXTRADATA_SIGNATURE, |
353 | 0 | TTMLENC_EXTRADATA_SIGNATURE_SIZE); |
354 | |
|
355 | 0 | memcpy(avctx->extradata + TTMLENC_EXTRADATA_SIGNATURE_SIZE, |
356 | 0 | s->buffer.str, additional_extradata_size); |
357 | |
|
358 | 0 | ret = 0; |
359 | 0 | fail: |
360 | 0 | av_bprint_finalize(&s->buffer, NULL); |
361 | |
|
362 | 0 | return ret; |
363 | 0 | } |
364 | | |
365 | | static av_cold int ttml_encode_init(AVCodecContext *avctx) |
366 | 0 | { |
367 | 0 | TTMLContext *s = avctx->priv_data; |
368 | 0 | int ret = AVERROR_BUG; |
369 | 0 | s->avctx = avctx; |
370 | |
|
371 | 0 | if (!(s->ass_ctx = ff_ass_split(avctx->subtitle_header))) { |
372 | 0 | return AVERROR_INVALIDDATA; |
373 | 0 | } |
374 | | |
375 | 0 | if ((ret = ttml_write_header_content(avctx)) < 0) { |
376 | 0 | return ret; |
377 | 0 | } |
378 | | |
379 | 0 | return 0; |
380 | 0 | } |
381 | | |
382 | | const FFCodec ff_ttml_encoder = { |
383 | | .p.name = "ttml", |
384 | | CODEC_LONG_NAME("TTML subtitle"), |
385 | | .p.type = AVMEDIA_TYPE_SUBTITLE, |
386 | | .p.id = AV_CODEC_ID_TTML, |
387 | | .priv_data_size = sizeof(TTMLContext), |
388 | | .init = ttml_encode_init, |
389 | | FF_CODEC_ENCODE_SUB_CB(ttml_encode_frame), |
390 | | .close = ttml_encode_close, |
391 | | .caps_internal = FF_CODEC_CAP_INIT_CLEANUP, |
392 | | }; |