2 * Closed Caption Decoding
3 * Copyright (c) 2015 Anshul Maheshwari
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/opt.h"
26 #define SCREEN_ROWS 15
27 #define SCREEN_COLUMNS 32
29 #define SET_FLAG(var, val) ( (var) |= ( 1 << (val)) )
30 #define UNSET_FLAG(var, val) ( (var) &= ~( 1 << (val)) )
31 #define CHECK_FLAG(var, val) ( (var) & ( 1 << (val)) )
33 static const AVRational ms_tb = {1, 1000};
59 CCFONT_UNDERLINED_ITALICS,
64 CCSET_SPECIAL_AMERICAN,
65 CCSET_EXTENDED_SPANISH_FRENCH_MISC,
66 CCSET_EXTENDED_PORTUGUESE_GERMAN_DANISH,
69 static const char *charset_overrides[4][128] =
71 [CCSET_BASIC_AMERICAN] = {
84 [CCSET_SPECIAL_AMERICAN] = {
102 [CCSET_EXTENDED_SPANISH_FRENCH_MISC] = {
136 [CCSET_EXTENDED_PORTUGUESE_GERMAN_DANISH] = {
172 static const unsigned char pac2_attribs[32][3] = // Color, font, ident
174 { CCCOL_WHITE, CCFONT_REGULAR, 0 }, // 0x40 || 0x60
175 { CCCOL_WHITE, CCFONT_UNDERLINED, 0 }, // 0x41 || 0x61
176 { CCCOL_GREEN, CCFONT_REGULAR, 0 }, // 0x42 || 0x62
177 { CCCOL_GREEN, CCFONT_UNDERLINED, 0 }, // 0x43 || 0x63
178 { CCCOL_BLUE, CCFONT_REGULAR, 0 }, // 0x44 || 0x64
179 { CCCOL_BLUE, CCFONT_UNDERLINED, 0 }, // 0x45 || 0x65
180 { CCCOL_CYAN, CCFONT_REGULAR, 0 }, // 0x46 || 0x66
181 { CCCOL_CYAN, CCFONT_UNDERLINED, 0 }, // 0x47 || 0x67
182 { CCCOL_RED, CCFONT_REGULAR, 0 }, // 0x48 || 0x68
183 { CCCOL_RED, CCFONT_UNDERLINED, 0 }, // 0x49 || 0x69
184 { CCCOL_YELLOW, CCFONT_REGULAR, 0 }, // 0x4a || 0x6a
185 { CCCOL_YELLOW, CCFONT_UNDERLINED, 0 }, // 0x4b || 0x6b
186 { CCCOL_MAGENTA, CCFONT_REGULAR, 0 }, // 0x4c || 0x6c
187 { CCCOL_MAGENTA, CCFONT_UNDERLINED, 0 }, // 0x4d || 0x6d
188 { CCCOL_WHITE, CCFONT_ITALICS, 0 }, // 0x4e || 0x6e
189 { CCCOL_WHITE, CCFONT_UNDERLINED_ITALICS, 0 }, // 0x4f || 0x6f
190 { CCCOL_WHITE, CCFONT_REGULAR, 0 }, // 0x50 || 0x70
191 { CCCOL_WHITE, CCFONT_UNDERLINED, 0 }, // 0x51 || 0x71
192 { CCCOL_WHITE, CCFONT_REGULAR, 4 }, // 0x52 || 0x72
193 { CCCOL_WHITE, CCFONT_UNDERLINED, 4 }, // 0x53 || 0x73
194 { CCCOL_WHITE, CCFONT_REGULAR, 8 }, // 0x54 || 0x74
195 { CCCOL_WHITE, CCFONT_UNDERLINED, 8 }, // 0x55 || 0x75
196 { CCCOL_WHITE, CCFONT_REGULAR, 12 }, // 0x56 || 0x76
197 { CCCOL_WHITE, CCFONT_UNDERLINED, 12 }, // 0x57 || 0x77
198 { CCCOL_WHITE, CCFONT_REGULAR, 16 }, // 0x58 || 0x78
199 { CCCOL_WHITE, CCFONT_UNDERLINED, 16 }, // 0x59 || 0x79
200 { CCCOL_WHITE, CCFONT_REGULAR, 20 }, // 0x5a || 0x7a
201 { CCCOL_WHITE, CCFONT_UNDERLINED, 20 }, // 0x5b || 0x7b
202 { CCCOL_WHITE, CCFONT_REGULAR, 24 }, // 0x5c || 0x7c
203 { CCCOL_WHITE, CCFONT_UNDERLINED, 24 }, // 0x5d || 0x7d
204 { CCCOL_WHITE, CCFONT_REGULAR, 28 }, // 0x5e || 0x7e
205 { CCCOL_WHITE, CCFONT_UNDERLINED, 28 } // 0x5f || 0x7f
206 /* total 32 entries */
210 /* +1 is used to compensate null character of string */
211 uint8_t characters[SCREEN_ROWS+1][SCREEN_COLUMNS+1];
212 uint8_t charsets[SCREEN_ROWS+1][SCREEN_COLUMNS+1];
213 uint8_t colors[SCREEN_ROWS+1][SCREEN_COLUMNS+1];
214 uint8_t fonts[SCREEN_ROWS+1][SCREEN_COLUMNS+1];
216 * Bitmask of used rows; if a bit is not set, the
217 * corresponding row is not used.
218 * for setting row 1 use row | (1 << 0)
219 * for setting row 15 use row | (1 << 14)
224 typedef struct CCaptionSubContext {
227 struct Screen screen[2];
230 uint8_t cursor_column;
231 uint8_t cursor_color;
233 uint8_t cursor_charset;
239 /* visible screen time */
243 int64_t last_real_time;
246 } CCaptionSubContext;
249 static av_cold int init_decoder(AVCodecContext *avctx)
252 CCaptionSubContext *ctx = avctx->priv_data;
254 av_bprint_init(&ctx->buffer, 0, AV_BPRINT_SIZE_UNLIMITED);
255 /* taking by default roll up to 2 */
256 ctx->mode = CCMODE_ROLLUP;
258 ctx->cursor_row = 10;
259 ret = ff_ass_subtitle_header(avctx, "Monospace",
260 ASS_DEFAULT_FONT_SIZE,
262 ASS_DEFAULT_BACK_COLOR,
265 ASS_DEFAULT_UNDERLINE,
267 ASS_DEFAULT_ALIGNMENT);
275 static av_cold int close_decoder(AVCodecContext *avctx)
277 CCaptionSubContext *ctx = avctx->priv_data;
278 av_bprint_finalize(&ctx->buffer, NULL);
282 static void flush_decoder(AVCodecContext *avctx)
284 CCaptionSubContext *ctx = avctx->priv_data;
285 ctx->screen[0].row_used = 0;
286 ctx->screen[1].row_used = 0;
287 ctx->prev_cmd[0] = 0;
288 ctx->prev_cmd[1] = 0;
289 ctx->mode = CCMODE_ROLLUP;
291 ctx->cursor_row = 10;
292 ctx->cursor_column = 0;
293 ctx->cursor_font = 0;
294 ctx->cursor_color = 0;
295 ctx->cursor_charset = 0;
296 ctx->active_screen = 0;
297 ctx->last_real_time = 0;
298 ctx->screen_touched = 0;
299 ctx->buffer_changed = 0;
300 if (!(avctx->flags2 & AV_CODEC_FLAG2_RO_FLUSH_NOOP))
302 av_bprint_clear(&ctx->buffer);
306 * @param ctx closed caption context just to print log
308 static void write_char(CCaptionSubContext *ctx, struct Screen *screen, char ch)
310 uint8_t col = ctx->cursor_column;
311 char *row = screen->characters[ctx->cursor_row];
312 char *font = screen->fonts[ctx->cursor_row];
313 char *color = screen->colors[ctx->cursor_row];
314 char *charset = screen->charsets[ctx->cursor_row];
316 if (col < SCREEN_COLUMNS) {
318 font[col] = ctx->cursor_font;
319 color[col] = ctx->cursor_color;
320 charset[col] = ctx->cursor_charset;
321 ctx->cursor_charset = CCSET_BASIC_AMERICAN;
322 if (ch) ctx->cursor_column++;
325 /* We have extra space at end only for null character */
326 else if (col == SCREEN_COLUMNS && ch == 0) {
331 av_log(ctx, AV_LOG_WARNING, "Data Ignored since exceeding screen width\n");
337 * This function after validating parity bit, also remove it from data pair.
338 * The first byte doesn't pass parity, we replace it with a solid blank
339 * and process the pair.
340 * If the second byte doesn't pass parity, it returns INVALIDDATA
341 * user can ignore the whole pair and pass the other pair.
343 static int validate_cc_data_pair(uint8_t *cc_data_pair)
345 uint8_t cc_valid = (*cc_data_pair & 4) >>2;
346 uint8_t cc_type = *cc_data_pair & 3;
349 return AVERROR_INVALIDDATA;
351 // if EIA-608 data then verify parity.
352 if (cc_type==0 || cc_type==1) {
353 if (!av_parity(cc_data_pair[2])) {
354 return AVERROR_INVALIDDATA;
356 if (!av_parity(cc_data_pair[1])) {
357 cc_data_pair[1]=0x7F;
362 if ((cc_data_pair[0] == 0xFA || cc_data_pair[0] == 0xFC || cc_data_pair[0] == 0xFD)
363 && (cc_data_pair[1] & 0x7F) == 0 && (cc_data_pair[2] & 0x7F) == 0)
364 return AVERROR_PATCHWELCOME;
367 if (cc_type == 3 || cc_type == 2)
368 return AVERROR_PATCHWELCOME;
370 /* remove parity bit */
371 cc_data_pair[1] &= 0x7F;
372 cc_data_pair[2] &= 0x7F;
377 static struct Screen *get_writing_screen(CCaptionSubContext *ctx)
381 // use Inactive screen
382 return ctx->screen + !ctx->active_screen;
387 return ctx->screen + ctx->active_screen;
389 /* It was never an option */
393 static void roll_up(CCaptionSubContext *ctx)
395 struct Screen *screen;
398 if (ctx->mode == CCMODE_TEXT)
401 screen = get_writing_screen(ctx);
403 /* +1 signify cursor_row starts from 0
404 * Can't keep lines less then row cursor pos
406 keep_lines = FFMIN(ctx->cursor_row + 1, ctx->rollup);
408 for (i = 0; i < SCREEN_ROWS; i++) {
409 if (i > ctx->cursor_row - keep_lines && i <= ctx->cursor_row)
411 UNSET_FLAG(screen->row_used, i);
414 for (i = 0; i < keep_lines && screen->row_used; i++) {
415 const int i_row = ctx->cursor_row - keep_lines + i + 1;
417 memcpy(screen->characters[i_row], screen->characters[i_row+1], SCREEN_COLUMNS);
418 memcpy(screen->colors[i_row], screen->colors[i_row+1], SCREEN_COLUMNS);
419 memcpy(screen->fonts[i_row], screen->fonts[i_row+1], SCREEN_COLUMNS);
420 memcpy(screen->charsets[i_row], screen->charsets[i_row+1], SCREEN_COLUMNS);
421 if (CHECK_FLAG(screen->row_used, i_row + 1))
422 SET_FLAG(screen->row_used, i_row);
425 UNSET_FLAG(screen->row_used, ctx->cursor_row);
428 static int capture_screen(CCaptionSubContext *ctx)
431 struct Screen *screen = ctx->screen + ctx->active_screen;
432 enum cc_font prev_font = CCFONT_REGULAR;
433 enum cc_color_code prev_color = CCCOL_WHITE;
434 av_bprint_clear(&ctx->buffer);
436 for (i = 0; screen->row_used && i < SCREEN_ROWS; i++)
438 if (CHECK_FLAG(screen->row_used, i)) {
439 const char *row = screen->characters[i];
440 const char *charset = screen->charsets[i];
442 while (row[j] == ' ' && charset[j] == CCSET_BASIC_AMERICAN)
449 for (i = 0; screen->row_used && i < SCREEN_ROWS; i++)
451 if (CHECK_FLAG(screen->row_used, i)) {
452 const char *row = screen->characters[i];
453 const char *font = screen->fonts[i];
454 const char *color = screen->colors[i];
455 const char *charset = screen->charsets[i];
456 const char *override;
457 int x, y, seen_char = 0;
460 /* skip leading space */
461 while (row[j] == ' ' && charset[j] == CCSET_BASIC_AMERICAN && j < tab)
464 x = ASS_DEFAULT_PLAYRESX * (0.1 + 0.0250 * j);
465 y = ASS_DEFAULT_PLAYRESY * (0.1 + 0.0533 * i);
466 av_bprintf(&ctx->buffer, "{\\an7}{\\pos(%d,%d)}", x, y);
468 for (; j < SCREEN_COLUMNS; j++) {
469 const char *e_tag = "", *s_tag = "", *c_tag = "";
474 if (prev_font != font[j]) {
479 case CCFONT_UNDERLINED:
482 case CCFONT_UNDERLINED_ITALICS:
483 e_tag = "{\\u0}{\\i0}";
490 case CCFONT_UNDERLINED:
493 case CCFONT_UNDERLINED_ITALICS:
494 s_tag = "{\\u1}{\\i1}";
498 if (prev_color != color[j]) {
501 c_tag = "{\\c&HFFFFFF&}";
504 c_tag = "{\\c&H00FF00&}";
507 c_tag = "{\\c&HFF0000&}";
510 c_tag = "{\\c&HFFFF00&}";
513 c_tag = "{\\c&H0000FF&}";
516 c_tag = "{\\c&H00FFFF&}";
519 c_tag = "{\\c&HFF00FF&}";
525 prev_color = color[j];
526 override = charset_overrides[(int)charset[j]][(int)row[j]];
528 av_bprintf(&ctx->buffer, "%s%s%s%s", e_tag, s_tag, c_tag, override);
530 } else if (row[j] == ' ' && !seen_char) {
531 av_bprintf(&ctx->buffer, "%s%s%s\\h", e_tag, s_tag, c_tag);
533 av_bprintf(&ctx->buffer, "%s%s%s%c", e_tag, s_tag, c_tag, row[j]);
538 av_bprintf(&ctx->buffer, "\\N");
541 if (!av_bprint_is_complete(&ctx->buffer))
542 return AVERROR(ENOMEM);
543 if (screen->row_used && ctx->buffer.len >= 2) {
544 ctx->buffer.len -= 2;
545 ctx->buffer.str[ctx->buffer.len] = 0;
547 ctx->buffer_changed = 1;
551 static int reap_screen(CCaptionSubContext *ctx, int64_t pts)
553 ctx->start_time = ctx->startv_time;
554 ctx->startv_time = pts;
556 return capture_screen(ctx);
559 static void handle_textattr(CCaptionSubContext *ctx, uint8_t hi, uint8_t lo)
562 struct Screen *screen = get_writing_screen(ctx);
567 ctx->cursor_color = pac2_attribs[i][0];
568 ctx->cursor_font = pac2_attribs[i][1];
570 SET_FLAG(screen->row_used, ctx->cursor_row);
571 write_char(ctx, screen, ' ');
574 static void handle_pac(CCaptionSubContext *ctx, uint8_t hi, uint8_t lo)
576 static const int8_t row_map[] = {
577 11, -1, 1, 2, 3, 4, 12, 13, 14, 15, 5, 6, 7, 8, 9, 10
579 const int index = ( (hi<<1) & 0x0e) | ( (lo>>5) & 0x01 );
580 struct Screen *screen = get_writing_screen(ctx);
583 if (row_map[index] <= 0) {
584 av_log(ctx, AV_LOG_DEBUG, "Invalid pac index encountered\n");
590 ctx->cursor_row = row_map[index] - 1;
591 ctx->cursor_color = pac2_attribs[lo][0];
592 ctx->cursor_font = pac2_attribs[lo][1];
593 ctx->cursor_charset = CCSET_BASIC_AMERICAN;
594 ctx->cursor_column = 0;
595 indent = pac2_attribs[lo][2];
596 for (i = 0; i < indent; i++) {
597 write_char(ctx, screen, ' ');
602 * @param pts it is required to set end time
604 static int handle_edm(CCaptionSubContext *ctx, int64_t pts)
606 struct Screen *screen = ctx->screen + ctx->active_screen;
609 // In buffered mode, keep writing to screen until it is wiped.
610 // Before wiping the display, capture contents to emit subtitle.
612 ret = reap_screen(ctx, pts);
614 screen->row_used = 0;
616 // In realtime mode, emit an empty caption so the last one doesn't
617 // stay on the screen.
619 ret = reap_screen(ctx, pts);
624 static int handle_eoc(CCaptionSubContext *ctx, int64_t pts)
628 ctx->active_screen = !ctx->active_screen;
630 // In buffered mode, we wait til the *next* EOC and
631 // reap what was already on the screen since the last EOC.
633 ret = handle_edm(ctx, pts);
635 ctx->cursor_column = 0;
637 // In realtime mode, we display the buffered contents (after
638 // flipping the buffer to active above) as soon as EOC arrives.
640 ret = reap_screen(ctx, pts);
645 static void handle_delete_end_of_row(CCaptionSubContext *ctx)
647 struct Screen *screen = get_writing_screen(ctx);
648 write_char(ctx, screen, 0);
651 static void handle_char(CCaptionSubContext *ctx, char hi, char lo)
653 struct Screen *screen = get_writing_screen(ctx);
655 SET_FLAG(screen->row_used, ctx->cursor_row);
659 ctx->cursor_charset = CCSET_SPECIAL_AMERICAN;
662 if (ctx->cursor_column > 0)
663 ctx->cursor_column -= 1;
664 ctx->cursor_charset = CCSET_EXTENDED_SPANISH_FRENCH_MISC;
667 if (ctx->cursor_column > 0)
668 ctx->cursor_column -= 1;
669 ctx->cursor_charset = CCSET_EXTENDED_PORTUGUESE_GERMAN_DANISH;
672 ctx->cursor_charset = CCSET_BASIC_AMERICAN;
673 write_char(ctx, screen, hi);
678 write_char(ctx, screen, lo);
680 write_char(ctx, screen, 0);
682 if (ctx->mode != CCMODE_POPON)
683 ctx->screen_touched = 1;
686 ff_dlog(ctx, "(%c,%c)\n", hi, lo);
688 ff_dlog(ctx, "(%c)\n", hi);
691 static int process_cc608(CCaptionSubContext *ctx, int64_t pts, uint8_t hi, uint8_t lo)
695 if (hi == ctx->prev_cmd[0] && lo == ctx->prev_cmd[1]) {
699 /* set prev command */
700 ctx->prev_cmd[0] = hi;
701 ctx->prev_cmd[1] = lo;
703 if ( (hi == 0x10 && (lo >= 0x40 && lo <= 0x5f)) ||
704 ( (hi >= 0x11 && hi <= 0x17) && (lo >= 0x40 && lo <= 0x7f) ) ) {
705 handle_pac(ctx, hi, lo);
706 } else if ( ( hi == 0x11 && lo >= 0x20 && lo <= 0x2f ) ||
707 ( hi == 0x17 && lo >= 0x2e && lo <= 0x2f) ) {
708 handle_textattr(ctx, hi, lo);
709 } else if (hi == 0x14 || hi == 0x15 || hi == 0x1c) {
712 /* resume caption loading */
713 ctx->mode = CCMODE_POPON;
716 handle_delete_end_of_row(ctx);
721 ctx->rollup = lo - 0x23;
722 ctx->mode = CCMODE_ROLLUP;
725 /* resume direct captioning */
726 ctx->mode = CCMODE_PAINTON;
729 /* resume text display */
730 ctx->mode = CCMODE_TEXT;
733 /* erase display memory */
734 handle_edm(ctx, pts);
737 /* carriage return */
738 ff_dlog(ctx, "carriage return\n");
740 ret = reap_screen(ctx, pts);
742 ctx->cursor_column = 0;
745 /* erase buffered (non displayed) memory */
746 // Only in realtime mode. In buffered mode, we re-use the inactive screen
747 // for our own buffering.
748 if (ctx->real_time) {
749 struct Screen *screen = ctx->screen + !ctx->active_screen;
750 screen->row_used = 0;
755 ff_dlog(ctx, "handle_eoc\n");
756 ret = handle_eoc(ctx, pts);
759 ff_dlog(ctx, "Unknown command 0x%hhx 0x%hhx\n", hi, lo);
762 } else if (hi >= 0x11 && hi <= 0x13) {
763 /* Special characters */
764 handle_char(ctx, hi, lo);
765 } else if (hi >= 0x20) {
766 /* Standard characters (always in pairs) */
767 handle_char(ctx, hi, lo);
768 ctx->prev_cmd[0] = ctx->prev_cmd[1] = 0;
769 } else if (hi == 0x17 && lo >= 0x21 && lo <= 0x23) {
771 /* Tab offsets (spacing) */
772 for (i = 0; i < lo - 0x20; i++) {
773 handle_char(ctx, ' ', 0);
776 /* Ignoring all other non data code */
777 ff_dlog(ctx, "Unknown command 0x%hhx 0x%hhx\n", hi, lo);
783 static int decode(AVCodecContext *avctx, void *data, int *got_sub, AVPacket *avpkt)
785 CCaptionSubContext *ctx = avctx->priv_data;
786 AVSubtitle *sub = data;
787 const int64_t start_time = sub->pts;
788 uint8_t *bptr = NULL;
789 int len = avpkt->size;
795 for (i = 0; i < len; i += 3) {
796 uint8_t cc_type = *(bptr + i) & 3;
797 if (validate_cc_data_pair(bptr + i))
799 /* ignoring data field 1 */
803 ret = process_cc608(ctx, start_time, *(bptr + i + 1) & 0x7f, *(bptr + i + 2) & 0x7f);
808 if (!ctx->buffer_changed)
810 ctx->buffer_changed = 0;
812 if (ctx->buffer.str[0] || ctx->real_time) {
813 ff_dlog(ctx, "cdp writing data (%s)\n", ctx->buffer.str);
814 ret = ff_ass_add_rect(sub, ctx->buffer.str, ctx->readorder++, 0, NULL, NULL);
817 sub->pts = ctx->start_time;
819 sub->end_display_time = av_rescale_q(ctx->end_time - ctx->start_time,
820 AV_TIME_BASE_Q, ms_tb);
822 sub->end_display_time = -1;
823 ctx->buffer_changed = 0;
824 ctx->last_real_time = sub->pts;
825 ctx->screen_touched = 0;
829 if (ctx->real_time && ctx->screen_touched &&
830 sub->pts > ctx->last_real_time + av_rescale_q(200, ms_tb, AV_TIME_BASE_Q)) {
831 ctx->last_real_time = sub->pts;
832 ctx->screen_touched = 0;
835 ctx->buffer_changed = 0;
837 ret = ff_ass_add_rect(sub, ctx->buffer.str, ctx->readorder++, 0, NULL, NULL);
840 sub->end_display_time = -1;
843 *got_sub = sub->num_rects > 0;
847 #define OFFSET(x) offsetof(CCaptionSubContext, x)
848 #define SD AV_OPT_FLAG_SUBTITLE_PARAM | AV_OPT_FLAG_DECODING_PARAM
849 static const AVOption options[] = {
850 { "real_time", "emit subtitle events as they are decoded for real-time display", OFFSET(real_time), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, SD },
854 static const AVClass ccaption_dec_class = {
855 .class_name = "Closed caption Decoder",
856 .item_name = av_default_item_name,
858 .version = LIBAVUTIL_VERSION_INT,
861 AVCodec ff_ccaption_decoder = {
863 .long_name = NULL_IF_CONFIG_SMALL("Closed Caption (EIA-608 / CEA-708)"),
864 .type = AVMEDIA_TYPE_SUBTITLE,
865 .id = AV_CODEC_ID_EIA_608,
866 .priv_data_size = sizeof(CCaptionSubContext),
867 .init = init_decoder,
868 .close = close_decoder,
869 .flush = flush_decoder,
871 .priv_class = &ccaption_dec_class,