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 ass_tb = {1, 100};
37 * 1) handle font and color completely
63 CCFONT_UNDERLINED_ITALICS,
68 CCSET_SPECIAL_AMERICAN,
69 CCSET_EXTENDED_SPANISH_FRENCH_MISC,
70 CCSET_EXTENDED_PORTUGUESE_GERMAN_DANISH,
73 static const char *charset_overrides[4][128] =
75 [CCSET_BASIC_AMERICAN] = {
88 [CCSET_SPECIAL_AMERICAN] = {
106 [CCSET_EXTENDED_SPANISH_FRENCH_MISC] = {
140 [CCSET_EXTENDED_PORTUGUESE_GERMAN_DANISH] = {
176 static const unsigned char pac2_attribs[32][3] = // Color, font, ident
178 { CCCOL_WHITE, CCFONT_REGULAR, 0 }, // 0x40 || 0x60
179 { CCCOL_WHITE, CCFONT_UNDERLINED, 0 }, // 0x41 || 0x61
180 { CCCOL_GREEN, CCFONT_REGULAR, 0 }, // 0x42 || 0x62
181 { CCCOL_GREEN, CCFONT_UNDERLINED, 0 }, // 0x43 || 0x63
182 { CCCOL_BLUE, CCFONT_REGULAR, 0 }, // 0x44 || 0x64
183 { CCCOL_BLUE, CCFONT_UNDERLINED, 0 }, // 0x45 || 0x65
184 { CCCOL_CYAN, CCFONT_REGULAR, 0 }, // 0x46 || 0x66
185 { CCCOL_CYAN, CCFONT_UNDERLINED, 0 }, // 0x47 || 0x67
186 { CCCOL_RED, CCFONT_REGULAR, 0 }, // 0x48 || 0x68
187 { CCCOL_RED, CCFONT_UNDERLINED, 0 }, // 0x49 || 0x69
188 { CCCOL_YELLOW, CCFONT_REGULAR, 0 }, // 0x4a || 0x6a
189 { CCCOL_YELLOW, CCFONT_UNDERLINED, 0 }, // 0x4b || 0x6b
190 { CCCOL_MAGENTA, CCFONT_REGULAR, 0 }, // 0x4c || 0x6c
191 { CCCOL_MAGENTA, CCFONT_UNDERLINED, 0 }, // 0x4d || 0x6d
192 { CCCOL_WHITE, CCFONT_ITALICS, 0 }, // 0x4e || 0x6e
193 { CCCOL_WHITE, CCFONT_UNDERLINED_ITALICS, 0 }, // 0x4f || 0x6f
194 { CCCOL_WHITE, CCFONT_REGULAR, 0 }, // 0x50 || 0x70
195 { CCCOL_WHITE, CCFONT_UNDERLINED, 0 }, // 0x51 || 0x71
196 { CCCOL_WHITE, CCFONT_REGULAR, 4 }, // 0x52 || 0x72
197 { CCCOL_WHITE, CCFONT_UNDERLINED, 4 }, // 0x53 || 0x73
198 { CCCOL_WHITE, CCFONT_REGULAR, 8 }, // 0x54 || 0x74
199 { CCCOL_WHITE, CCFONT_UNDERLINED, 8 }, // 0x55 || 0x75
200 { CCCOL_WHITE, CCFONT_REGULAR, 12 }, // 0x56 || 0x76
201 { CCCOL_WHITE, CCFONT_UNDERLINED, 12 }, // 0x57 || 0x77
202 { CCCOL_WHITE, CCFONT_REGULAR, 16 }, // 0x58 || 0x78
203 { CCCOL_WHITE, CCFONT_UNDERLINED, 16 }, // 0x59 || 0x79
204 { CCCOL_WHITE, CCFONT_REGULAR, 20 }, // 0x5a || 0x7a
205 { CCCOL_WHITE, CCFONT_UNDERLINED, 20 }, // 0x5b || 0x7b
206 { CCCOL_WHITE, CCFONT_REGULAR, 24 }, // 0x5c || 0x7c
207 { CCCOL_WHITE, CCFONT_UNDERLINED, 24 }, // 0x5d || 0x7d
208 { CCCOL_WHITE, CCFONT_REGULAR, 28 }, // 0x5e || 0x7e
209 { CCCOL_WHITE, CCFONT_UNDERLINED, 28 } // 0x5f || 0x7f
210 /* total 32 entries */
214 /* +1 is used to compensate null character of string */
215 uint8_t characters[SCREEN_ROWS][SCREEN_COLUMNS+1];
216 uint8_t charsets[SCREEN_ROWS][SCREEN_COLUMNS+1];
217 uint8_t colors[SCREEN_ROWS][SCREEN_COLUMNS+1];
218 uint8_t fonts[SCREEN_ROWS][SCREEN_COLUMNS+1];
220 * Bitmask of used rows; if a bit is not set, the
221 * corresponding row is not used.
222 * for setting row 1 use row | (1 << 0)
223 * for setting row 15 use row | (1 << 14)
228 typedef struct CCaptionSubContext {
231 struct Screen screen[2];
234 uint8_t cursor_column;
235 uint8_t cursor_color;
237 uint8_t cursor_charset;
243 /* visible screen time */
247 int64_t last_real_time;
249 /* buffer to store pkt data */
252 } CCaptionSubContext;
255 static av_cold int init_decoder(AVCodecContext *avctx)
258 CCaptionSubContext *ctx = avctx->priv_data;
260 av_bprint_init(&ctx->buffer, 0, AV_BPRINT_SIZE_UNLIMITED);
261 /* taking by default roll up to 2 */
262 ctx->mode = CCMODE_ROLLUP;
264 ret = ff_ass_subtitle_header(avctx, "Monospace",
265 ASS_DEFAULT_FONT_SIZE,
267 ASS_DEFAULT_BACK_COLOR,
270 ASS_DEFAULT_UNDERLINE,
272 ASS_DEFAULT_ALIGNMENT);
276 /* allocate pkt buffer */
277 ctx->pktbuf = av_buffer_alloc(128);
279 ret = AVERROR(ENOMEM);
284 static av_cold int close_decoder(AVCodecContext *avctx)
286 CCaptionSubContext *ctx = avctx->priv_data;
287 av_bprint_finalize(&ctx->buffer, NULL);
288 av_buffer_unref(&ctx->pktbuf);
292 static void flush_decoder(AVCodecContext *avctx)
294 CCaptionSubContext *ctx = avctx->priv_data;
295 ctx->screen[0].row_used = 0;
296 ctx->screen[1].row_used = 0;
297 ctx->prev_cmd[0] = 0;
298 ctx->prev_cmd[1] = 0;
299 ctx->mode = CCMODE_ROLLUP;
302 ctx->cursor_column = 0;
303 ctx->cursor_font = 0;
304 ctx->cursor_color = 0;
305 ctx->cursor_charset = 0;
306 ctx->active_screen = 0;
307 ctx->last_real_time = 0;
308 ctx->screen_touched = 0;
309 ctx->buffer_changed = 0;
310 if (!(avctx->flags2 & AV_CODEC_FLAG2_RO_FLUSH_NOOP))
312 av_bprint_clear(&ctx->buffer);
316 * @param ctx closed caption context just to print log
318 static int write_char(CCaptionSubContext *ctx, struct Screen *screen, char ch)
320 uint8_t col = ctx->cursor_column;
321 char *row = screen->characters[ctx->cursor_row];
322 char *font = screen->fonts[ctx->cursor_row];
323 char *charset = screen->charsets[ctx->cursor_row];
325 if (col < SCREEN_COLUMNS) {
327 font[col] = ctx->cursor_font;
328 charset[col] = ctx->cursor_charset;
329 ctx->cursor_charset = CCSET_BASIC_AMERICAN;
330 if (ch) ctx->cursor_column++;
333 /* We have extra space at end only for null character */
334 else if (col == SCREEN_COLUMNS && ch == 0) {
339 av_log(ctx, AV_LOG_WARNING, "Data Ignored since exceeding screen width\n");
340 return AVERROR_INVALIDDATA;
345 * This function after validating parity bit, also remove it from data pair.
346 * The first byte doesn't pass parity, we replace it with a solid blank
347 * and process the pair.
348 * If the second byte doesn't pass parity, it returns INVALIDDATA
349 * user can ignore the whole pair and pass the other pair.
351 static int validate_cc_data_pair(uint8_t *cc_data_pair)
353 uint8_t cc_valid = (*cc_data_pair & 4) >>2;
354 uint8_t cc_type = *cc_data_pair & 3;
357 return AVERROR_INVALIDDATA;
359 // if EIA-608 data then verify parity.
360 if (cc_type==0 || cc_type==1) {
361 if (!av_parity(cc_data_pair[2])) {
362 return AVERROR_INVALIDDATA;
364 if (!av_parity(cc_data_pair[1])) {
365 cc_data_pair[1]=0x7F;
370 if ((cc_data_pair[0] == 0xFA || cc_data_pair[0] == 0xFC || cc_data_pair[0] == 0xFD)
371 && (cc_data_pair[1] & 0x7F) == 0 && (cc_data_pair[2] & 0x7F) == 0)
372 return AVERROR_PATCHWELCOME;
375 if (cc_type == 3 || cc_type == 2)
376 return AVERROR_PATCHWELCOME;
378 /* remove parity bit */
379 cc_data_pair[1] &= 0x7F;
380 cc_data_pair[2] &= 0x7F;
385 static struct Screen *get_writing_screen(CCaptionSubContext *ctx)
389 // use Inactive screen
390 return ctx->screen + !ctx->active_screen;
395 return ctx->screen + ctx->active_screen;
397 /* It was never an option */
401 static void roll_up(CCaptionSubContext *ctx)
403 struct Screen *screen;
406 if (ctx->mode == CCMODE_TEXT)
409 screen = get_writing_screen(ctx);
411 /* +1 signify cursor_row starts from 0
412 * Can't keep lines less then row cursor pos
414 keep_lines = FFMIN(ctx->cursor_row + 1, ctx->rollup);
416 for (i = 0; i < SCREEN_ROWS; i++) {
417 if (i > ctx->cursor_row - keep_lines && i <= ctx->cursor_row)
419 UNSET_FLAG(screen->row_used, i);
422 for (i = 0; i < keep_lines && screen->row_used; i++) {
423 const int i_row = ctx->cursor_row - keep_lines + i + 1;
425 memcpy(screen->characters[i_row], screen->characters[i_row+1], SCREEN_COLUMNS);
426 memcpy(screen->colors[i_row], screen->colors[i_row+1], SCREEN_COLUMNS);
427 memcpy(screen->fonts[i_row], screen->fonts[i_row+1], SCREEN_COLUMNS);
428 memcpy(screen->charsets[i_row], screen->charsets[i_row+1], SCREEN_COLUMNS);
429 if (CHECK_FLAG(screen->row_used, i_row + 1))
430 SET_FLAG(screen->row_used, i_row);
433 UNSET_FLAG(screen->row_used, ctx->cursor_row);
436 static int capture_screen(CCaptionSubContext *ctx)
439 struct Screen *screen = ctx->screen + ctx->active_screen;
440 enum cc_font prev_font = CCFONT_REGULAR;
441 av_bprint_clear(&ctx->buffer);
443 for (i = 0; screen->row_used && i < SCREEN_ROWS; i++)
445 if (CHECK_FLAG(screen->row_used, i)) {
446 const char *row = screen->characters[i];
447 const char *font = screen->fonts[i];
448 const char *charset = screen->charsets[i];
449 const char *override;
452 /* skip leading space */
453 while (row[j] == ' ' && charset[j] == CCSET_BASIC_AMERICAN)
456 for (; j < SCREEN_COLUMNS; j++) {
457 const char *e_tag = "", *s_tag = "";
462 if (prev_font != font[j]) {
467 case CCFONT_UNDERLINED:
470 case CCFONT_UNDERLINED_ITALICS:
471 e_tag = "{\\u0}{\\i0}";
478 case CCFONT_UNDERLINED:
481 case CCFONT_UNDERLINED_ITALICS:
482 s_tag = "{\\u1}{\\i1}";
487 override = charset_overrides[(int)charset[j]][(int)row[j]];
489 av_bprintf(&ctx->buffer, "%s%s%s", e_tag, s_tag, override);
491 av_bprintf(&ctx->buffer, "%s%s%c", e_tag, s_tag, row[j]);
494 av_bprintf(&ctx->buffer, "\\N");
497 if (!av_bprint_is_complete(&ctx->buffer))
498 return AVERROR(ENOMEM);
499 if (screen->row_used && ctx->buffer.len >= 2) {
500 ctx->buffer.len -= 2;
501 ctx->buffer.str[ctx->buffer.len] = 0;
503 ctx->buffer_changed = 1;
507 static int reap_screen(CCaptionSubContext *ctx, int64_t pts)
509 ctx->start_time = ctx->startv_time;
510 ctx->startv_time = pts;
512 return capture_screen(ctx);
515 static void handle_textattr(CCaptionSubContext *ctx, uint8_t hi, uint8_t lo)
518 struct Screen *screen = get_writing_screen(ctx);
523 ctx->cursor_color = pac2_attribs[i][0];
524 ctx->cursor_font = pac2_attribs[i][1];
526 SET_FLAG(screen->row_used, ctx->cursor_row);
527 write_char(ctx, screen, ' ');
530 static void handle_pac(CCaptionSubContext *ctx, uint8_t hi, uint8_t lo)
532 static const int8_t row_map[] = {
533 11, -1, 1, 2, 3, 4, 12, 13, 14, 15, 5, 6, 7, 8, 9, 10
535 const int index = ( (hi<<1) & 0x0e) | ( (lo>>5) & 0x01 );
536 struct Screen *screen = get_writing_screen(ctx);
539 if (row_map[index] <= 0) {
540 av_log(ctx, AV_LOG_DEBUG, "Invalid pac index encountered\n");
546 ctx->cursor_row = row_map[index] - 1;
547 ctx->cursor_color = pac2_attribs[lo][0];
548 ctx->cursor_font = pac2_attribs[lo][1];
549 ctx->cursor_charset = CCSET_BASIC_AMERICAN;
550 ctx->cursor_column = 0;
551 indent = pac2_attribs[lo][2];
552 for (i = 0; i < indent; i++) {
553 write_char(ctx, screen, ' ');
558 * @param pts it is required to set end time
560 static void handle_edm(CCaptionSubContext *ctx, int64_t pts)
562 struct Screen *screen = ctx->screen + ctx->active_screen;
564 // In buffered mode, keep writing to screen until it is wiped.
565 // Before wiping the display, capture contents to emit subtitle.
567 reap_screen(ctx, pts);
569 screen->row_used = 0;
571 // In realtime mode, emit an empty caption so the last one doesn't
572 // stay on the screen.
574 reap_screen(ctx, pts);
577 static void handle_eoc(CCaptionSubContext *ctx, int64_t pts)
579 // In buffered mode, we wait til the *next* EOC and
580 // reap what was already on the screen since the last EOC.
584 ctx->active_screen = !ctx->active_screen;
585 ctx->cursor_column = 0;
587 // In realtime mode, we display the buffered contents (after
588 // flipping the buffer to active above) as soon as EOC arrives.
590 reap_screen(ctx, pts);
593 static void handle_delete_end_of_row(CCaptionSubContext *ctx, char hi, char lo)
595 struct Screen *screen = get_writing_screen(ctx);
596 write_char(ctx, screen, 0);
599 static void handle_char(CCaptionSubContext *ctx, char hi, char lo, int64_t pts)
601 struct Screen *screen = get_writing_screen(ctx);
603 SET_FLAG(screen->row_used, ctx->cursor_row);
607 ctx->cursor_charset = CCSET_SPECIAL_AMERICAN;
610 if (ctx->cursor_column > 0)
611 ctx->cursor_column -= 1;
612 ctx->cursor_charset = CCSET_EXTENDED_SPANISH_FRENCH_MISC;
615 if (ctx->cursor_column > 0)
616 ctx->cursor_column -= 1;
617 ctx->cursor_charset = CCSET_EXTENDED_PORTUGUESE_GERMAN_DANISH;
620 ctx->cursor_charset = CCSET_BASIC_AMERICAN;
621 write_char(ctx, screen, hi);
626 write_char(ctx, screen, lo);
628 write_char(ctx, screen, 0);
630 if (ctx->mode != CCMODE_POPON)
631 ctx->screen_touched = 1;
634 ff_dlog(ctx, "(%c,%c)\n", hi, lo);
636 ff_dlog(ctx, "(%c)\n", hi);
639 static void process_cc608(CCaptionSubContext *ctx, int64_t pts, uint8_t hi, uint8_t lo)
641 if (hi == ctx->prev_cmd[0] && lo == ctx->prev_cmd[1]) {
642 /* ignore redundant command */
646 /* set prev command */
647 ctx->prev_cmd[0] = hi;
648 ctx->prev_cmd[1] = lo;
650 if ( (hi == 0x10 && (lo >= 0x40 && lo <= 0x5f)) ||
651 ( (hi >= 0x11 && hi <= 0x17) && (lo >= 0x40 && lo <= 0x7f) ) ) {
652 handle_pac(ctx, hi, lo);
653 } else if ( ( hi == 0x11 && lo >= 0x20 && lo <= 0x2f ) ||
654 ( hi == 0x17 && lo >= 0x2e && lo <= 0x2f) ) {
655 handle_textattr(ctx, hi, lo);
656 } else if (hi == 0x14 || hi == 0x15 || hi == 0x1c) {
659 /* resume caption loading */
660 ctx->mode = CCMODE_POPON;
663 handle_delete_end_of_row(ctx, hi, lo);
668 ctx->rollup = lo - 0x23;
669 ctx->mode = CCMODE_ROLLUP;
672 /* resume direct captioning */
673 ctx->mode = CCMODE_PAINTON;
676 /* resume text display */
677 ctx->mode = CCMODE_TEXT;
680 /* erase display memory */
681 handle_edm(ctx, pts);
684 /* carriage return */
685 ff_dlog(ctx, "carriage return\n");
687 reap_screen(ctx, pts);
689 ctx->cursor_column = 0;
692 /* erase buffered (non displayed) memory */
693 // Only in realtime mode. In buffered mode, we re-use the inactive screen
694 // for our own buffering.
695 if (ctx->real_time) {
696 struct Screen *screen = ctx->screen + !ctx->active_screen;
697 screen->row_used = 0;
702 ff_dlog(ctx, "handle_eoc\n");
703 handle_eoc(ctx, pts);
706 ff_dlog(ctx, "Unknown command 0x%hhx 0x%hhx\n", hi, lo);
709 } else if (hi >= 0x11 && hi <= 0x13) {
710 /* Special characters */
711 handle_char(ctx, hi, lo, pts);
712 } else if (hi >= 0x20) {
713 /* Standard characters (always in pairs) */
714 handle_char(ctx, hi, lo, pts);
715 ctx->prev_cmd[0] = ctx->prev_cmd[1] = 0;
717 /* Ignoring all other non data code */
718 ff_dlog(ctx, "Unknown command 0x%hhx 0x%hhx\n", hi, lo);
722 static int decode(AVCodecContext *avctx, void *data, int *got_sub, AVPacket *avpkt)
724 CCaptionSubContext *ctx = avctx->priv_data;
725 AVSubtitle *sub = data;
726 uint8_t *bptr = NULL;
727 int len = avpkt->size;
731 if (ctx->pktbuf->size < len) {
732 ret = av_buffer_realloc(&ctx->pktbuf, len);
734 av_log(ctx, AV_LOG_WARNING, "Insufficient Memory of %d truncated to %d\n", len, ctx->pktbuf->size);
735 len = ctx->pktbuf->size;
739 memcpy(ctx->pktbuf->data, avpkt->data, len);
740 bptr = ctx->pktbuf->data;
742 for (i = 0; i < len; i += 3) {
743 uint8_t cc_type = *(bptr + i) & 3;
744 if (validate_cc_data_pair(bptr + i))
746 /* ignoring data field 1 */
750 process_cc608(ctx, avpkt->pts, *(bptr + i + 1) & 0x7f, *(bptr + i + 2) & 0x7f);
752 if (!ctx->buffer_changed)
754 ctx->buffer_changed = 0;
756 if (*ctx->buffer.str || ctx->real_time)
758 ff_dlog(ctx, "cdp writing data (%s)\n",ctx->buffer.str);
759 ret = ff_ass_add_rect(sub, ctx->buffer.str, ctx->readorder++, 0, NULL, NULL);
762 sub->pts = av_rescale_q(ctx->start_time, avctx->time_base, AV_TIME_BASE_Q);
764 sub->end_display_time = av_rescale_q(ctx->end_time - ctx->start_time,
765 avctx->time_base, av_make_q(1, 1000));
767 sub->end_display_time = -1;
768 ctx->buffer_changed = 0;
769 ctx->last_real_time = avpkt->pts;
770 ctx->screen_touched = 0;
774 if (ctx->real_time && ctx->screen_touched &&
775 avpkt->pts > ctx->last_real_time + av_rescale_q(20, ass_tb, avctx->time_base)) {
776 ctx->last_real_time = avpkt->pts;
777 ctx->screen_touched = 0;
780 ctx->buffer_changed = 0;
782 ret = ff_ass_add_rect(sub, ctx->buffer.str, ctx->readorder++, 0, NULL, NULL);
785 sub->end_display_time = -1;
788 *got_sub = sub->num_rects > 0;
792 #define OFFSET(x) offsetof(CCaptionSubContext, x)
793 #define SD AV_OPT_FLAG_SUBTITLE_PARAM | AV_OPT_FLAG_DECODING_PARAM
794 static const AVOption options[] = {
795 { "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 },
799 static const AVClass ccaption_dec_class = {
800 .class_name = "Closed caption Decoder",
801 .item_name = av_default_item_name,
803 .version = LIBAVUTIL_VERSION_INT,
806 AVCodec ff_ccaption_decoder = {
808 .long_name = NULL_IF_CONFIG_SMALL("Closed Caption (EIA-608 / CEA-708) Decoder"),
809 .type = AVMEDIA_TYPE_SUBTITLE,
810 .id = AV_CODEC_ID_EIA_608,
811 .priv_data_size = sizeof(CCaptionSubContext),
812 .init = init_decoder,
813 .close = close_decoder,
814 .flush = flush_decoder,
816 .priv_class = &ccaption_dec_class,