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,
66 static const unsigned char pac2_attribs[32][3] = // Color, font, ident
68 { CCCOL_WHITE, CCFONT_REGULAR, 0 }, // 0x40 || 0x60
69 { CCCOL_WHITE, CCFONT_UNDERLINED, 0 }, // 0x41 || 0x61
70 { CCCOL_GREEN, CCFONT_REGULAR, 0 }, // 0x42 || 0x62
71 { CCCOL_GREEN, CCFONT_UNDERLINED, 0 }, // 0x43 || 0x63
72 { CCCOL_BLUE, CCFONT_REGULAR, 0 }, // 0x44 || 0x64
73 { CCCOL_BLUE, CCFONT_UNDERLINED, 0 }, // 0x45 || 0x65
74 { CCCOL_CYAN, CCFONT_REGULAR, 0 }, // 0x46 || 0x66
75 { CCCOL_CYAN, CCFONT_UNDERLINED, 0 }, // 0x47 || 0x67
76 { CCCOL_RED, CCFONT_REGULAR, 0 }, // 0x48 || 0x68
77 { CCCOL_RED, CCFONT_UNDERLINED, 0 }, // 0x49 || 0x69
78 { CCCOL_YELLOW, CCFONT_REGULAR, 0 }, // 0x4a || 0x6a
79 { CCCOL_YELLOW, CCFONT_UNDERLINED, 0 }, // 0x4b || 0x6b
80 { CCCOL_MAGENTA, CCFONT_REGULAR, 0 }, // 0x4c || 0x6c
81 { CCCOL_MAGENTA, CCFONT_UNDERLINED, 0 }, // 0x4d || 0x6d
82 { CCCOL_WHITE, CCFONT_ITALICS, 0 }, // 0x4e || 0x6e
83 { CCCOL_WHITE, CCFONT_UNDERLINED_ITALICS, 0 }, // 0x4f || 0x6f
84 { CCCOL_WHITE, CCFONT_REGULAR, 0 }, // 0x50 || 0x70
85 { CCCOL_WHITE, CCFONT_UNDERLINED, 0 }, // 0x51 || 0x71
86 { CCCOL_WHITE, CCFONT_REGULAR, 4 }, // 0x52 || 0x72
87 { CCCOL_WHITE, CCFONT_UNDERLINED, 4 }, // 0x53 || 0x73
88 { CCCOL_WHITE, CCFONT_REGULAR, 8 }, // 0x54 || 0x74
89 { CCCOL_WHITE, CCFONT_UNDERLINED, 8 }, // 0x55 || 0x75
90 { CCCOL_WHITE, CCFONT_REGULAR, 12 }, // 0x56 || 0x76
91 { CCCOL_WHITE, CCFONT_UNDERLINED, 12 }, // 0x57 || 0x77
92 { CCCOL_WHITE, CCFONT_REGULAR, 16 }, // 0x58 || 0x78
93 { CCCOL_WHITE, CCFONT_UNDERLINED, 16 }, // 0x59 || 0x79
94 { CCCOL_WHITE, CCFONT_REGULAR, 20 }, // 0x5a || 0x7a
95 { CCCOL_WHITE, CCFONT_UNDERLINED, 20 }, // 0x5b || 0x7b
96 { CCCOL_WHITE, CCFONT_REGULAR, 24 }, // 0x5c || 0x7c
97 { CCCOL_WHITE, CCFONT_UNDERLINED, 24 }, // 0x5d || 0x7d
98 { CCCOL_WHITE, CCFONT_REGULAR, 28 }, // 0x5e || 0x7e
99 { CCCOL_WHITE, CCFONT_UNDERLINED, 28 } // 0x5f || 0x7f
100 /* total 32 entries */
104 /* +1 is used to compensate null character of string */
105 uint8_t characters[SCREEN_ROWS][SCREEN_COLUMNS+1];
106 uint8_t colors[SCREEN_ROWS][SCREEN_COLUMNS+1];
107 uint8_t fonts[SCREEN_ROWS][SCREEN_COLUMNS+1];
109 * Bitmask of used rows; if a bit is not set, the
110 * corresponding row is not used.
111 * for setting row 1 use row | (1 << 0)
112 * for setting row 15 use row | (1 << 14)
117 typedef struct CCaptionSubContext {
120 struct Screen screen[2];
123 uint8_t cursor_column;
124 uint8_t cursor_color;
131 /* visible screen time */
135 int64_t last_real_time;
137 /* buffer to store pkt data */
139 } CCaptionSubContext;
142 static av_cold int init_decoder(AVCodecContext *avctx)
145 CCaptionSubContext *ctx = avctx->priv_data;
147 av_bprint_init(&ctx->buffer, 0, AV_BPRINT_SIZE_UNLIMITED);
148 /* taking by default roll up to 2 */
149 ctx->mode = CCMODE_ROLLUP;
151 ret = ff_ass_subtitle_header(avctx, "Monospace",
152 ASS_DEFAULT_FONT_SIZE,
154 ASS_DEFAULT_BACK_COLOR,
157 ASS_DEFAULT_UNDERLINE,
159 ASS_DEFAULT_ALIGNMENT);
163 /* allocate pkt buffer */
164 ctx->pktbuf = av_buffer_alloc(128);
166 ret = AVERROR(ENOMEM);
171 static av_cold int close_decoder(AVCodecContext *avctx)
173 CCaptionSubContext *ctx = avctx->priv_data;
174 av_bprint_finalize(&ctx->buffer, NULL);
175 av_buffer_unref(&ctx->pktbuf);
179 static void flush_decoder(AVCodecContext *avctx)
181 CCaptionSubContext *ctx = avctx->priv_data;
182 ctx->screen[0].row_used = 0;
183 ctx->screen[1].row_used = 0;
184 ctx->prev_cmd[0] = 0;
185 ctx->prev_cmd[1] = 0;
186 ctx->mode = CCMODE_ROLLUP;
189 ctx->cursor_column = 0;
190 ctx->cursor_font = 0;
191 ctx->cursor_color = 0;
192 ctx->active_screen = 0;
193 ctx->last_real_time = 0;
194 ctx->screen_touched = 0;
195 ctx->buffer_changed = 0;
196 av_bprint_clear(&ctx->buffer);
200 * @param ctx closed caption context just to print log
202 static int write_char(CCaptionSubContext *ctx, struct Screen *screen, char ch)
204 uint8_t col = ctx->cursor_column;
205 char *row = screen->characters[ctx->cursor_row];
206 char *font = screen->fonts[ctx->cursor_row];
208 if (col < SCREEN_COLUMNS) {
210 font[col] = ctx->cursor_font;
211 if (ch) ctx->cursor_column++;
214 /* We have extra space at end only for null character */
215 else if (col == SCREEN_COLUMNS && ch == 0) {
220 av_log(ctx, AV_LOG_WARNING, "Data Ignored since exceeding screen width\n");
221 return AVERROR_INVALIDDATA;
226 * This function after validating parity bit, also remove it from data pair.
227 * The first byte doesn't pass parity, we replace it with a solid blank
228 * and process the pair.
229 * If the second byte doesn't pass parity, it returns INVALIDDATA
230 * user can ignore the whole pair and pass the other pair.
232 static int validate_cc_data_pair(uint8_t *cc_data_pair)
234 uint8_t cc_valid = (*cc_data_pair & 4) >>2;
235 uint8_t cc_type = *cc_data_pair & 3;
238 return AVERROR_INVALIDDATA;
240 // if EIA-608 data then verify parity.
241 if (cc_type==0 || cc_type==1) {
242 if (!av_parity(cc_data_pair[2])) {
243 return AVERROR_INVALIDDATA;
245 if (!av_parity(cc_data_pair[1])) {
246 cc_data_pair[1]=0x7F;
251 if ((cc_data_pair[0] == 0xFA || cc_data_pair[0] == 0xFC || cc_data_pair[0] == 0xFD)
252 && (cc_data_pair[1] & 0x7F) == 0 && (cc_data_pair[2] & 0x7F) == 0)
253 return AVERROR_PATCHWELCOME;
256 if (cc_type == 3 || cc_type == 2)
257 return AVERROR_PATCHWELCOME;
259 /* remove parity bit */
260 cc_data_pair[1] &= 0x7F;
261 cc_data_pair[2] &= 0x7F;
266 static struct Screen *get_writing_screen(CCaptionSubContext *ctx)
270 // use Inactive screen
271 return ctx->screen + !ctx->active_screen;
276 return ctx->screen + ctx->active_screen;
278 /* It was never an option */
282 static void roll_up(CCaptionSubContext *ctx)
284 struct Screen *screen;
287 if (ctx->mode == CCMODE_TEXT)
290 screen = get_writing_screen(ctx);
292 /* +1 signify cursor_row starts from 0
293 * Can't keep lines less then row cursor pos
295 keep_lines = FFMIN(ctx->cursor_row + 1, ctx->rollup);
297 for (i = 0; i < SCREEN_ROWS; i++) {
298 if (i > ctx->cursor_row - keep_lines && i <= ctx->cursor_row)
300 UNSET_FLAG(screen->row_used, i);
303 for (i = 0; i < keep_lines && screen->row_used; i++) {
304 const int i_row = ctx->cursor_row - keep_lines + i + 1;
306 memcpy(screen->characters[i_row], screen->characters[i_row+1], SCREEN_COLUMNS);
307 memcpy(screen->colors[i_row], screen->colors[i_row+1], SCREEN_COLUMNS);
308 memcpy(screen->fonts[i_row], screen->fonts[i_row+1], SCREEN_COLUMNS);
309 if (CHECK_FLAG(screen->row_used, i_row + 1))
310 SET_FLAG(screen->row_used, i_row);
313 UNSET_FLAG(screen->row_used, ctx->cursor_row);
316 static int capture_screen(CCaptionSubContext *ctx)
319 struct Screen *screen = ctx->screen + ctx->active_screen;
320 enum cc_font prev_font = CCFONT_REGULAR;
321 av_bprint_clear(&ctx->buffer);
323 for (i = 0; screen->row_used && i < SCREEN_ROWS; i++)
325 if (CHECK_FLAG(screen->row_used, i)) {
326 const char *row = screen->characters[i];
327 const char *font = screen->fonts[i];
330 /* skip leading space */
331 while (row[j] == ' ')
334 for (; j < SCREEN_COLUMNS; j++) {
335 const char *e_tag = "", *s_tag = "";
340 if (prev_font != font[j]) {
345 case CCFONT_UNDERLINED:
348 case CCFONT_UNDERLINED_ITALICS:
349 e_tag = "{\\u0}{\\i0}";
356 case CCFONT_UNDERLINED:
359 case CCFONT_UNDERLINED_ITALICS:
360 s_tag = "{\\u1}{\\i1}";
365 av_bprintf(&ctx->buffer, "%s%s%c", e_tag, s_tag, row[j]);
367 av_bprintf(&ctx->buffer, "\\N");
370 if (!av_bprint_is_complete(&ctx->buffer))
371 return AVERROR(ENOMEM);
372 if (screen->row_used && ctx->buffer.len >= 2) {
373 ctx->buffer.len -= 2;
374 ctx->buffer.str[ctx->buffer.len] = 0;
376 ctx->buffer_changed = 1;
380 static int reap_screen(CCaptionSubContext *ctx, int64_t pts)
382 ctx->start_time = ctx->startv_time;
383 ctx->startv_time = pts;
385 return capture_screen(ctx);
388 static void handle_textattr(CCaptionSubContext *ctx, uint8_t hi, uint8_t lo)
391 struct Screen *screen = get_writing_screen(ctx);
396 ctx->cursor_color = pac2_attribs[i][0];
397 ctx->cursor_font = pac2_attribs[i][1];
399 SET_FLAG(screen->row_used, ctx->cursor_row);
400 write_char(ctx, screen, ' ');
403 static void handle_pac(CCaptionSubContext *ctx, uint8_t hi, uint8_t lo)
405 static const int8_t row_map[] = {
406 11, -1, 1, 2, 3, 4, 12, 13, 14, 15, 5, 6, 7, 8, 9, 10
408 const int index = ( (hi<<1) & 0x0e) | ( (lo>>5) & 0x01 );
409 struct Screen *screen = get_writing_screen(ctx);
412 if (row_map[index] <= 0) {
413 av_log(ctx, AV_LOG_DEBUG, "Invalid pac index encountered\n");
419 ctx->cursor_row = row_map[index] - 1;
420 ctx->cursor_color = pac2_attribs[lo][0];
421 ctx->cursor_font = pac2_attribs[lo][1];
422 ctx->cursor_column = 0;
423 indent = pac2_attribs[lo][2];
424 for (i = 0; i < indent; i++) {
425 write_char(ctx, screen, ' ');
430 * @param pts it is required to set end time
432 static void handle_edm(CCaptionSubContext *ctx, int64_t pts)
434 struct Screen *screen = ctx->screen + ctx->active_screen;
436 // In buffered mode, keep writing to screen until it is wiped.
437 // Before wiping the display, capture contents to emit subtitle.
439 reap_screen(ctx, pts);
441 screen->row_used = 0;
443 // In realtime mode, emit an empty caption so the last one doesn't
444 // stay on the screen.
446 reap_screen(ctx, pts);
449 static void handle_eoc(CCaptionSubContext *ctx, int64_t pts)
451 // In buffered mode, we wait til the *next* EOC and
452 // reap what was already on the screen since the last EOC.
456 ctx->active_screen = !ctx->active_screen;
457 ctx->cursor_column = 0;
459 // In realtime mode, we display the buffered contents (after
460 // flipping the buffer to active above) as soon as EOC arrives.
462 reap_screen(ctx, pts);
465 static void handle_delete_end_of_row(CCaptionSubContext *ctx, char hi, char lo)
467 struct Screen *screen = get_writing_screen(ctx);
468 write_char(ctx, screen, 0);
471 static void handle_char(CCaptionSubContext *ctx, char hi, char lo, int64_t pts)
473 struct Screen *screen = get_writing_screen(ctx);
475 SET_FLAG(screen->row_used, ctx->cursor_row);
477 write_char(ctx, screen, hi);
480 write_char(ctx, screen, lo);
482 write_char(ctx, screen, 0);
484 if (ctx->mode != CCMODE_POPON)
485 ctx->screen_touched = 1;
487 /* reset prev command since character can repeat */
488 ctx->prev_cmd[0] = 0;
489 ctx->prev_cmd[1] = 0;
491 ff_dlog(ctx, "(%c,%c)\n", hi, lo);
493 ff_dlog(ctx, "(%c)\n", hi);
496 static void process_cc608(CCaptionSubContext *ctx, int64_t pts, uint8_t hi, uint8_t lo)
498 if (hi == ctx->prev_cmd[0] && lo == ctx->prev_cmd[1]) {
499 /* ignore redundant command */
500 } else if ( (hi == 0x10 && (lo >= 0x40 && lo <= 0x5f)) ||
501 ( (hi >= 0x11 && hi <= 0x17) && (lo >= 0x40 && lo <= 0x7f) ) ) {
502 handle_pac(ctx, hi, lo);
503 } else if ( ( hi == 0x11 && lo >= 0x20 && lo <= 0x2f ) ||
504 ( hi == 0x17 && lo >= 0x2e && lo <= 0x2f) ) {
505 handle_textattr(ctx, hi, lo);
506 } else if (hi == 0x14 || hi == 0x15 || hi == 0x1c) {
509 /* resume caption loading */
510 ctx->mode = CCMODE_POPON;
513 handle_delete_end_of_row(ctx, hi, lo);
518 ctx->rollup = lo - 0x23;
519 ctx->mode = CCMODE_ROLLUP;
522 /* resume direct captioning */
523 ctx->mode = CCMODE_PAINTON;
526 /* resume text display */
527 ctx->mode = CCMODE_TEXT;
530 /* erase display memory */
531 handle_edm(ctx, pts);
534 /* carriage return */
535 ff_dlog(ctx, "carriage return\n");
537 reap_screen(ctx, pts);
539 ctx->cursor_column = 0;
542 /* erase buffered (non displayed) memory */
543 // Only in realtime mode. In buffered mode, we re-use the inactive screen
544 // for our own buffering.
545 if (ctx->real_time) {
546 struct Screen *screen = ctx->screen + !ctx->active_screen;
547 screen->row_used = 0;
552 ff_dlog(ctx, "handle_eoc\n");
553 handle_eoc(ctx, pts);
556 ff_dlog(ctx, "Unknown command 0x%hhx 0x%hhx\n", hi, lo);
559 } else if (hi >= 0x20) {
560 /* Standard characters (always in pairs) */
561 handle_char(ctx, hi, lo, pts);
563 /* Ignoring all other non data code */
564 ff_dlog(ctx, "Unknown command 0x%hhx 0x%hhx\n", hi, lo);
567 /* set prev command */
568 ctx->prev_cmd[0] = hi;
569 ctx->prev_cmd[1] = lo;
572 static int decode(AVCodecContext *avctx, void *data, int *got_sub, AVPacket *avpkt)
574 CCaptionSubContext *ctx = avctx->priv_data;
575 AVSubtitle *sub = data;
576 uint8_t *bptr = NULL;
577 int len = avpkt->size;
581 if (ctx->pktbuf->size < len) {
582 ret = av_buffer_realloc(&ctx->pktbuf, len);
584 av_log(ctx, AV_LOG_WARNING, "Insufficient Memory of %d truncated to %d\n", len, ctx->pktbuf->size);
585 len = ctx->pktbuf->size;
589 memcpy(ctx->pktbuf->data, avpkt->data, len);
590 bptr = ctx->pktbuf->data;
592 for (i = 0; i < len; i += 3) {
593 uint8_t cc_type = *(bptr + i) & 3;
594 if (validate_cc_data_pair(bptr + i))
596 /* ignoring data field 1 */
600 process_cc608(ctx, avpkt->pts, *(bptr + i + 1) & 0x7f, *(bptr + i + 2) & 0x7f);
602 if (!ctx->buffer_changed)
604 ctx->buffer_changed = 0;
606 if (*ctx->buffer.str || ctx->real_time)
608 int64_t sub_pts = ctx->real_time ? avpkt->pts : ctx->start_time;
609 int start_time = av_rescale_q(sub_pts, avctx->time_base, ass_tb);
611 if (!ctx->real_time) {
612 int end_time = av_rescale_q(ctx->end_time, avctx->time_base, ass_tb);
613 duration = end_time - start_time;
615 ff_dlog(ctx, "cdp writing data (%s)\n",ctx->buffer.str);
616 ret = ff_ass_add_rect_bprint(sub, &ctx->buffer, start_time, duration);
619 sub->pts = av_rescale_q(sub_pts, avctx->time_base, AV_TIME_BASE_Q);
620 ctx->buffer_changed = 0;
621 ctx->last_real_time = avpkt->pts;
622 ctx->screen_touched = 0;
626 if (ctx->real_time && ctx->screen_touched &&
627 avpkt->pts > ctx->last_real_time + av_rescale_q(20, ass_tb, avctx->time_base)) {
628 ctx->last_real_time = avpkt->pts;
629 ctx->screen_touched = 0;
632 ctx->buffer_changed = 0;
634 int start_time = av_rescale_q(avpkt->pts, avctx->time_base, ass_tb);
635 ret = ff_ass_add_rect_bprint(sub, &ctx->buffer, start_time, -1);
638 sub->pts = av_rescale_q(avpkt->pts, avctx->time_base, AV_TIME_BASE_Q);
641 *got_sub = sub->num_rects > 0;
645 #define OFFSET(x) offsetof(CCaptionSubContext, x)
646 #define SD AV_OPT_FLAG_SUBTITLE_PARAM | AV_OPT_FLAG_DECODING_PARAM
647 static const AVOption options[] = {
648 { "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 },
652 static const AVClass ccaption_dec_class = {
653 .class_name = "Closed caption Decoder",
654 .item_name = av_default_item_name,
656 .version = LIBAVUTIL_VERSION_INT,
659 AVCodec ff_ccaption_decoder = {
661 .long_name = NULL_IF_CONFIG_SMALL("Closed Caption (EIA-608 / CEA-708) Decoder"),
662 .type = AVMEDIA_TYPE_SUBTITLE,
663 .id = AV_CODEC_ID_EIA_608,
664 .priv_data_size = sizeof(CCaptionSubContext),
665 .init = init_decoder,
666 .close = close_decoder,
667 .flush = flush_decoder,
669 .priv_class = &ccaption_dec_class,