2 * 3GPP TS 26.245 Timed Text decoder
3 * Copyright (c) 2012 Philip Langdale <philipl@overt.org>
5 * This file is part of FFmpeg.
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.
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.
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
24 #include "libavutil/avstring.h"
25 #include "libavutil/common.h"
26 #include "libavutil/bprint.h"
27 #include "libavutil/intreadwrite.h"
28 #include "libavutil/mem.h"
30 #define STYLE_FLAG_BOLD (1<<0)
31 #define STYLE_FLAG_ITALIC (1<<1)
32 #define STYLE_FLAG_UNDERLINE (1<<2)
34 #define BOX_SIZE_INITIAL 40
36 #define STYL_BOX (1<<0)
37 #define HLIT_BOX (1<<1)
38 #define HCLR_BOX (1<<2)
39 #define TWRP_BOX (1<<3)
42 #define BOTTOM_CENTER 2
43 #define BOTTOM_RIGHT 3
45 #define MIDDLE_CENTER 5
46 #define MIDDLE_RIGHT 6
72 uint16_t style_fontID;
81 uint8_t hlit_color[4];
94 FontRecord *ftab_temp;
98 uint16_t style_entries, ftab_entries;
101 int count_s, count_f;
107 int (*decode)(const uint8_t *tsmb, MovTextContext *m, AVPacket *avpkt);
110 static void mov_text_cleanup(MovTextContext *m)
113 if (m->box_flags & STYL_BOX) {
114 for(i = 0; i < m->count_s; i++) {
121 static void mov_text_cleanup_ftab(MovTextContext *m)
125 av_freep(&m->ftab_temp->font);
126 av_freep(&m->ftab_temp);
128 for(i = 0; i < m->count_f; i++) {
129 av_freep(&m->ftab[i]->font);
130 av_freep(&m->ftab[i]);
136 static int mov_text_tx3g(AVCodecContext *avctx, MovTextContext *m)
138 uint8_t *tx3g_ptr = avctx->extradata;
139 int i, box_size, font_length;
140 int8_t v_align, h_align;
146 box_size = BOX_SIZE_INITIAL; /* Size till ftab_entries */
147 if (avctx->extradata_size < box_size)
153 h_align = *tx3g_ptr++;
154 v_align = *tx3g_ptr++;
157 m->d.alignment = TOP_LEFT;
159 m->d.alignment = MIDDLE_LEFT;
161 m->d.alignment = BOTTOM_LEFT;
165 m->d.alignment = TOP_CENTER;
167 m->d.alignment = MIDDLE_CENTER;
169 m->d.alignment = BOTTOM_CENTER;
173 m->d.alignment = TOP_RIGHT;
175 m->d.alignment = MIDDLE_RIGHT;
177 m->d.alignment = BOTTOM_RIGHT;
180 m->d.back_color = AV_RB24(tx3g_ptr);
187 style_fontID = AV_RB16(tx3g_ptr);
190 s_default.style_flag = *tx3g_ptr++;
191 m->d.bold = s_default.style_flag & STYLE_FLAG_BOLD;
192 m->d.italic = s_default.style_flag & STYLE_FLAG_ITALIC;
193 m->d.underline = s_default.style_flag & STYLE_FLAG_UNDERLINE;
195 m->d.fontsize = *tx3g_ptr++;
197 m->d.color = AV_RB24(tx3g_ptr);
205 m->ftab_entries = AV_RB16(tx3g_ptr);
208 for (i = 0; i < m->ftab_entries; i++) {
211 if (avctx->extradata_size < box_size) {
212 mov_text_cleanup_ftab(m);
216 m->ftab_temp = av_mallocz(sizeof(*m->ftab_temp));
218 mov_text_cleanup_ftab(m);
219 return AVERROR(ENOMEM);
221 m->ftab_temp->fontID = AV_RB16(tx3g_ptr);
223 font_length = *tx3g_ptr++;
225 box_size = box_size + font_length;
226 if (avctx->extradata_size < box_size) {
227 mov_text_cleanup_ftab(m);
231 m->ftab_temp->font = av_malloc(font_length + 1);
232 if (!m->ftab_temp->font) {
233 mov_text_cleanup_ftab(m);
234 return AVERROR(ENOMEM);
236 memcpy(m->ftab_temp->font, tx3g_ptr, font_length);
237 m->ftab_temp->font[font_length] = '\0';
238 av_dynarray_add(&m->ftab, &m->count_f, m->ftab_temp);
240 mov_text_cleanup_ftab(m);
241 return AVERROR(ENOMEM);
244 tx3g_ptr = tx3g_ptr + font_length;
246 for (i = 0; i < m->ftab_entries; i++) {
247 if (style_fontID == m->ftab[i]->fontID)
248 m->d.font = m->ftab[i]->font;
253 static int decode_twrp(const uint8_t *tsmb, MovTextContext *m, AVPacket *avpkt)
255 m->box_flags |= TWRP_BOX;
256 m->w.wrap_flag = *tsmb++;
260 static int decode_hlit(const uint8_t *tsmb, MovTextContext *m, AVPacket *avpkt)
262 m->box_flags |= HLIT_BOX;
263 m->h.hlit_start = AV_RB16(tsmb);
265 m->h.hlit_end = AV_RB16(tsmb);
270 static int decode_hclr(const uint8_t *tsmb, MovTextContext *m, AVPacket *avpkt)
272 m->box_flags |= HCLR_BOX;
273 memcpy(m->c.hlit_color, tsmb, 4);
278 static int decode_styl(const uint8_t *tsmb, MovTextContext *m, AVPacket *avpkt)
281 m->style_entries = AV_RB16(tsmb);
283 // A single style record is of length 12 bytes.
284 if (m->tracksize + m->size_var + 2 + m->style_entries * 12 > avpkt->size)
287 m->box_flags |= STYL_BOX;
288 for(i = 0; i < m->style_entries; i++) {
289 m->s_temp = av_malloc(sizeof(*m->s_temp));
292 return AVERROR(ENOMEM);
294 m->s_temp->style_start = AV_RB16(tsmb);
296 m->s_temp->style_end = AV_RB16(tsmb);
298 m->s_temp->style_fontID = AV_RB16(tsmb);
300 m->s_temp->style_flag = AV_RB8(tsmb);
302 m->s_temp->fontsize = AV_RB8(tsmb);
303 av_dynarray_add(&m->s, &m->count_s, m->s_temp);
306 return AVERROR(ENOMEM);
315 static const Box box_types[] = {
316 { MKBETAG('s','t','y','l'), 2, decode_styl },
317 { MKBETAG('h','l','i','t'), 4, decode_hlit },
318 { MKBETAG('h','c','l','r'), 4, decode_hclr },
319 { MKBETAG('t','w','r','p'), 1, decode_twrp }
322 const static size_t box_count = FF_ARRAY_ELEMS(box_types);
324 static int text_to_ass(AVBPrint *buf, const char *text, const char *text_end,
331 if (text < text_end && m->box_flags & TWRP_BOX) {
332 if (m->w.wrap_flag == 1) {
333 av_bprintf(buf, "{\\q1}"); /* End of line wrap */
335 av_bprintf(buf, "{\\q2}"); /* No wrap */
339 while (text < text_end) {
340 if (m->box_flags & STYL_BOX) {
341 for (i = 0; i < m->style_entries; i++) {
342 if (m->s[i]->style_flag && text_pos == m->s[i]->style_end) {
343 av_bprintf(buf, "{\\r}");
346 for (i = 0; i < m->style_entries; i++) {
347 if (m->s[i]->style_flag && text_pos == m->s[i]->style_start) {
348 if (m->s[i]->style_flag & STYLE_FLAG_BOLD)
349 av_bprintf(buf, "{\\b1}");
350 if (m->s[i]->style_flag & STYLE_FLAG_ITALIC)
351 av_bprintf(buf, "{\\i1}");
352 if (m->s[i]->style_flag & STYLE_FLAG_UNDERLINE)
353 av_bprintf(buf, "{\\u1}");
354 av_bprintf(buf, "{\\fs%d}", m->s[i]->fontsize);
355 for (j = 0; j < m->ftab_entries; j++) {
356 if (m->s[i]->style_fontID == m->ftab[j]->fontID)
357 av_bprintf(buf, "{\\fn%s}", m->ftab[j]->font);
362 if (m->box_flags & HLIT_BOX) {
363 if (text_pos == m->h.hlit_start) {
364 /* If hclr box is present, set the secondary color to the color
365 * specified. Otherwise, set primary color to white and secondary
366 * color to black. These colors will come from TextSampleModifier
367 * boxes in future and inverse video technique for highlight will
370 if (m->box_flags & HCLR_BOX) {
371 av_bprintf(buf, "{\\2c&H%02x%02x%02x&}", m->c.hlit_color[2],
372 m->c.hlit_color[1], m->c.hlit_color[0]);
374 av_bprintf(buf, "{\\1c&H000000&}{\\2c&HFFFFFF&}");
377 if (text_pos == m->h.hlit_end) {
378 if (m->box_flags & HCLR_BOX) {
379 av_bprintf(buf, "{\\2c&H000000&}");
381 av_bprintf(buf, "{\\1c&HFFFFFF&}{\\2c&H000000&}");
390 av_bprintf(buf, "\\N");
393 av_bprint_chars(buf, *text, 1);
403 static int mov_text_init(AVCodecContext *avctx) {
405 * TODO: Handle the default text style.
406 * NB: Most players ignore styles completely, with the result that
407 * it's very common to find files where the default style is broken
408 * and respecting it results in a worse experience than ignoring it.
411 MovTextContext *m = avctx->priv_data;
412 ret = mov_text_tx3g(avctx, m);
414 return ff_ass_subtitle_header(avctx, m->d.font, m->d.fontsize, m->d.color,
415 m->d.back_color, m->d.bold, m->d.italic,
416 m->d.underline, m->d.alignment);
418 return ff_ass_subtitle_header_default(avctx);
421 static int mov_text_decode_frame(AVCodecContext *avctx,
422 void *data, int *got_sub_ptr, AVPacket *avpkt)
424 AVSubtitle *sub = data;
425 MovTextContext *m = avctx->priv_data;
426 int ret, ts_start, ts_end;
428 char *ptr = avpkt->data;
430 int text_length, tsmb_type, ret_tsmb;
434 if (!ptr || avpkt->size < 2)
435 return AVERROR_INVALIDDATA;
438 * A packet of size two with value zero is an empty subtitle
439 * used to mark the end of the previous non-empty subtitle.
440 * We can just drop them here as we have duration information
441 * already. If the value is non-zero, then it's technically a
444 if (avpkt->size == 2)
445 return AV_RB16(ptr) == 0 ? 0 : AVERROR_INVALIDDATA;
448 * The first two bytes of the packet are the length of the text string
449 * In complex cases, there are style descriptors appended to the string
450 * so we can't just assume the packet size is the string size.
452 text_length = AV_RB16(ptr);
453 end = ptr + FFMIN(2 + text_length, avpkt->size);
456 ts_start = av_rescale_q(avpkt->pts,
458 (AVRational){1,100});
459 ts_end = av_rescale_q(avpkt->pts + avpkt->duration,
461 (AVRational){1,100});
464 m->tracksize = 2 + text_length;
465 m->style_entries = 0;
468 // Note that the spec recommends lines be no longer than 2048 characters.
469 av_bprint_init(&buf, 0, AV_BPRINT_SIZE_UNLIMITED);
470 if (text_length + 2 != avpkt->size) {
471 while (m->tracksize + 8 <= avpkt->size) {
472 // A box is a minimum of 8 bytes.
473 tsmb = ptr + m->tracksize - 2;
474 tsmb_size = AV_RB32(tsmb);
476 tsmb_type = AV_RB32(tsmb);
479 if (tsmb_size == 1) {
480 if (m->tracksize + 16 > avpkt->size)
482 tsmb_size = AV_RB64(tsmb);
487 //size_var is equal to 8 or 16 depending on the size of box
489 if (m->tracksize + tsmb_size > avpkt->size)
492 for (size_t i = 0; i < box_count; i++) {
493 if (tsmb_type == box_types[i].type) {
494 if (m->tracksize + m->size_var + box_types[i].base_size > avpkt->size)
496 ret_tsmb = box_types[i].decode(tsmb, m, avpkt);
501 m->tracksize = m->tracksize + tsmb_size;
503 text_to_ass(&buf, ptr, end, m);
506 text_to_ass(&buf, ptr, end, m);
508 ret = ff_ass_add_rect_bprint(sub, &buf, ts_start, ts_end - ts_start);
509 av_bprint_finalize(&buf, NULL);
512 *got_sub_ptr = sub->num_rects > 0;
516 static int mov_text_decode_close(AVCodecContext *avctx)
518 MovTextContext *m = avctx->priv_data;
519 mov_text_cleanup_ftab(m);
523 AVCodec ff_movtext_decoder = {
525 .long_name = NULL_IF_CONFIG_SMALL("3GPP Timed Text subtitle"),
526 .type = AVMEDIA_TYPE_SUBTITLE,
527 .id = AV_CODEC_ID_MOV_TEXT,
528 .priv_data_size = sizeof(MovTextContext),
529 .init = mov_text_init,
530 .decode = mov_text_decode_frame,
531 .close = mov_text_decode_close,