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 {
119 struct Screen screen[2];
122 uint8_t cursor_column;
123 uint8_t cursor_color;
130 /* visible screen time */
134 /* buffer to store pkt data */
136 } CCaptionSubContext;
139 static av_cold int init_decoder(AVCodecContext *avctx)
142 CCaptionSubContext *ctx = avctx->priv_data;
144 av_bprint_init(&ctx->buffer, 0, AV_BPRINT_SIZE_UNLIMITED);
145 /* taking by default roll up to 2 */
146 ctx->mode = CCMODE_ROLLUP;
148 ret = ff_ass_subtitle_header_default(avctx);
152 /* allocate pkt buffer */
153 ctx->pktbuf = av_buffer_alloc(128);
155 ret = AVERROR(ENOMEM);
160 static av_cold int close_decoder(AVCodecContext *avctx)
162 CCaptionSubContext *ctx = avctx->priv_data;
163 av_bprint_finalize(&ctx->buffer, NULL);
164 av_buffer_unref(&ctx->pktbuf);
169 * @param ctx closed caption context just to print log
171 static int write_char(CCaptionSubContext *ctx, struct Screen *screen, char ch)
173 uint8_t col = ctx->cursor_column;
174 char *row = screen->characters[ctx->cursor_row];
175 char *font = screen->fonts[ctx->cursor_row];
177 if (col < SCREEN_COLUMNS) {
179 font[col] = ctx->cursor_font;
180 if (ch) ctx->cursor_column++;
183 /* We have extra space at end only for null character */
184 else if (col == SCREEN_COLUMNS && ch == 0) {
189 av_log(ctx, AV_LOG_WARNING, "Data Ignored since exceeding screen width\n");
190 return AVERROR_INVALIDDATA;
195 * This function after validating parity bit, also remove it from data pair.
196 * The first byte doesn't pass parity, we replace it with a solid blank
197 * and process the pair.
198 * If the second byte doesn't pass parity, it returns INVALIDDATA
199 * user can ignore the whole pair and pass the other pair.
201 static int validate_cc_data_pair(uint8_t *cc_data_pair)
203 uint8_t cc_valid = (*cc_data_pair & 4) >>2;
204 uint8_t cc_type = *cc_data_pair & 3;
207 return AVERROR_INVALIDDATA;
209 // if EIA-608 data then verify parity.
210 if (cc_type==0 || cc_type==1) {
211 if (!av_parity(cc_data_pair[2])) {
212 return AVERROR_INVALIDDATA;
214 if (!av_parity(cc_data_pair[1])) {
215 cc_data_pair[1]=0x7F;
220 if ((cc_data_pair[0] == 0xFA || cc_data_pair[0] == 0xFC || cc_data_pair[0] == 0xFD)
221 && (cc_data_pair[1] & 0x7F) == 0 && (cc_data_pair[2] & 0x7F) == 0)
222 return AVERROR_PATCHWELCOME;
225 if (cc_type == 3 || cc_type == 2)
226 return AVERROR_PATCHWELCOME;
228 /* remove parity bit */
229 cc_data_pair[1] &= 0x7F;
230 cc_data_pair[2] &= 0x7F;
235 static struct Screen *get_writing_screen(CCaptionSubContext *ctx)
239 // use Inactive screen
240 return ctx->screen + !ctx->active_screen;
245 return ctx->screen + ctx->active_screen;
247 /* It was never an option */
251 static void roll_up(CCaptionSubContext *ctx)
253 struct Screen *screen;
256 if (ctx->mode == CCMODE_TEXT)
259 screen = get_writing_screen(ctx);
261 /* +1 signify cursor_row starts from 0
262 * Can't keep lines less then row cursor pos
264 keep_lines = FFMIN(ctx->cursor_row + 1, ctx->rollup);
266 for (i = 0; i < ctx->cursor_row - keep_lines; i++)
267 UNSET_FLAG(screen->row_used, i);
270 for (i = 0; i < keep_lines && screen->row_used; i++) {
271 const int i_row = ctx->cursor_row - keep_lines + i + 1;
273 memcpy(screen->characters[i_row], screen->characters[i_row+1], SCREEN_COLUMNS);
274 memcpy(screen->colors[i_row], screen->colors[i_row+1], SCREEN_COLUMNS);
275 memcpy(screen->fonts[i_row], screen->fonts[i_row+1], SCREEN_COLUMNS);
276 if (CHECK_FLAG(screen->row_used, i_row + 1))
277 SET_FLAG(screen->row_used, i_row);
280 UNSET_FLAG(screen->row_used, ctx->cursor_row);
283 static int capture_screen(CCaptionSubContext *ctx)
287 struct Screen *screen = ctx->screen + ctx->active_screen;
288 enum cc_font prev_font = CCFONT_REGULAR;
289 av_bprint_clear(&ctx->buffer);
291 for (i = 0; screen->row_used && i < SCREEN_ROWS; i++)
293 if (CHECK_FLAG(screen->row_used, i)) {
294 const char *row = screen->characters[i];
295 const char *font = screen->fonts[i];
298 /* skip leading space */
299 while (row[j] == ' ')
302 for (; j < SCREEN_COLUMNS; j++) {
303 const char *e_tag = "", *s_tag = "";
308 if (prev_font != font[j]) {
313 case CCFONT_UNDERLINED:
316 case CCFONT_UNDERLINED_ITALICS:
317 e_tag = "{/u0}{/i0}";
324 case CCFONT_UNDERLINED:
327 case CCFONT_UNDERLINED_ITALICS:
328 s_tag = "{/u1}{/i1}";
334 av_bprintf(&ctx->buffer, "%s%s%c", e_tag, s_tag, row[j]);
337 av_bprintf(&ctx->buffer, "\\N");
338 ret = av_bprint_is_complete(&ctx->buffer);
340 ret = AVERROR(ENOMEM);
346 if (screen->row_used && ctx->buffer.len >= 2) {
347 ctx->buffer.len -= 2;
348 ctx->buffer.str[ctx->buffer.len] = 0;
350 ctx->buffer_changed = 1;
354 static int reap_screen(CCaptionSubContext *ctx, int64_t pts)
356 ctx->start_time = ctx->startv_time;
357 ctx->startv_time = pts;
359 return capture_screen(ctx);
362 static void handle_textattr(CCaptionSubContext *ctx, uint8_t hi, uint8_t lo)
365 struct Screen *screen = get_writing_screen(ctx);
370 ctx->cursor_color = pac2_attribs[i][0];
371 ctx->cursor_font = pac2_attribs[i][1];
373 SET_FLAG(screen->row_used, ctx->cursor_row);
374 write_char(ctx, screen, ' ');
377 static void handle_pac(CCaptionSubContext *ctx, uint8_t hi, uint8_t lo)
379 static const int8_t row_map[] = {
380 11, -1, 1, 2, 3, 4, 12, 13, 14, 15, 5, 6, 7, 8, 9, 10
382 const int index = ( (hi<<1) & 0x0e) | ( (lo>>5) & 0x01 );
383 struct Screen *screen = get_writing_screen(ctx);
386 if (row_map[index] <= 0) {
387 av_log(ctx, AV_LOG_DEBUG, "Invalid pac index encountered\n");
393 ctx->cursor_row = row_map[index] - 1;
394 ctx->cursor_color = pac2_attribs[lo][0];
395 ctx->cursor_font = pac2_attribs[lo][1];
396 ctx->cursor_column = 0;
397 indent = pac2_attribs[lo][2];
398 for (i = 0; i < indent; i++) {
399 write_char(ctx, screen, ' ');
404 * @param pts it is required to set end time
406 static void handle_edm(CCaptionSubContext *ctx, int64_t pts)
408 struct Screen *screen = ctx->screen + ctx->active_screen;
410 reap_screen(ctx, pts);
411 screen->row_used = 0;
414 static void handle_eoc(CCaptionSubContext *ctx, int64_t pts)
417 ctx->active_screen = !ctx->active_screen;
418 ctx->cursor_column = 0;
421 static void handle_delete_end_of_row(CCaptionSubContext *ctx, char hi, char lo)
423 struct Screen *screen = get_writing_screen(ctx);
424 write_char(ctx, screen, 0);
427 static void handle_char(CCaptionSubContext *ctx, char hi, char lo, int64_t pts)
429 struct Screen *screen = get_writing_screen(ctx);
431 SET_FLAG(screen->row_used, ctx->cursor_row);
433 write_char(ctx, screen, hi);
436 write_char(ctx, screen, lo);
438 write_char(ctx, screen, 0);
440 /* reset prev command since character can repeat */
441 ctx->prev_cmd[0] = 0;
442 ctx->prev_cmd[1] = 0;
444 ff_dlog(ctx, "(%c,%c)\n", hi, lo);
446 ff_dlog(ctx, "(%c)\n", hi);
449 static void process_cc608(CCaptionSubContext *ctx, int64_t pts, uint8_t hi, uint8_t lo)
451 if (hi == ctx->prev_cmd[0] && lo == ctx->prev_cmd[1]) {
452 /* ignore redundant command */
453 } else if ( (hi == 0x10 && (lo >= 0x40 && lo <= 0x5f)) ||
454 ( (hi >= 0x11 && hi <= 0x17) && (lo >= 0x40 && lo <= 0x7f) ) ) {
455 handle_pac(ctx, hi, lo);
456 } else if ( ( hi == 0x11 && lo >= 0x20 && lo <= 0x2f ) ||
457 ( hi == 0x17 && lo >= 0x2e && lo <= 0x2f) ) {
458 handle_textattr(ctx, hi, lo);
459 } else if (hi == 0x14 || hi == 0x15 || hi == 0x1c) {
462 /* resume caption loading */
463 ctx->mode = CCMODE_POPON;
466 handle_delete_end_of_row(ctx, hi, lo);
470 ctx->mode = CCMODE_ROLLUP;
474 ctx->mode = CCMODE_ROLLUP;
478 ctx->mode = CCMODE_ROLLUP;
481 /* resume direct captioning */
482 ctx->mode = CCMODE_PAINTON;
485 /* resume text display */
486 ctx->mode = CCMODE_TEXT;
489 /* erase display memory */
490 handle_edm(ctx, pts);
493 /* carriage return */
494 ff_dlog(ctx, "carriage return\n");
495 reap_screen(ctx, pts);
497 ctx->cursor_column = 0;
501 ff_dlog(ctx, "handle_eoc\n");
502 handle_eoc(ctx, pts);
505 ff_dlog(ctx, "Unknown command 0x%hhx 0x%hhx\n", hi, lo);
508 } else if (hi >= 0x20) {
509 /* Standard characters (always in pairs) */
510 handle_char(ctx, hi, lo, pts);
512 /* Ignoring all other non data code */
513 ff_dlog(ctx, "Unknown command 0x%hhx 0x%hhx\n", hi, lo);
516 /* set prev command */
517 ctx->prev_cmd[0] = hi;
518 ctx->prev_cmd[1] = lo;
521 static int decode(AVCodecContext *avctx, void *data, int *got_sub, AVPacket *avpkt)
523 CCaptionSubContext *ctx = avctx->priv_data;
524 AVSubtitle *sub = data;
525 uint8_t *bptr = NULL;
526 int len = avpkt->size;
530 if (ctx->pktbuf->size < len) {
531 ret = av_buffer_realloc(&ctx->pktbuf, len);
533 av_log(ctx, AV_LOG_WARNING, "Insufficient Memory of %d truncated to %d\n", len, ctx->pktbuf->size);
534 len = ctx->pktbuf->size;
538 memcpy(ctx->pktbuf->data, avpkt->data, len);
539 bptr = ctx->pktbuf->data;
541 for (i = 0; i < len; i += 3) {
542 uint8_t cc_type = *(bptr + i) & 3;
543 if (validate_cc_data_pair(bptr + i))
545 /* ignoring data field 1 */
549 process_cc608(ctx, avpkt->pts, *(bptr + i + 1) & 0x7f, *(bptr + i + 2) & 0x7f);
550 if (ctx->buffer_changed && *ctx->buffer.str)
552 int start_time = av_rescale_q(ctx->start_time, avctx->time_base, ass_tb);
553 int end_time = av_rescale_q(ctx->end_time, avctx->time_base, ass_tb);
554 ff_dlog(ctx, "cdp writing data (%s)\n",ctx->buffer.str);
555 ret = ff_ass_add_rect_bprint(sub, &ctx->buffer, start_time, end_time - start_time);
558 sub->pts = av_rescale_q(ctx->start_time, avctx->time_base, AV_TIME_BASE_Q);
559 ctx->buffer_changed = 0;
563 *got_sub = sub->num_rects > 0;
567 static const AVOption options[] = {
571 static const AVClass ccaption_dec_class = {
572 .class_name = "Closed caption Decoder",
573 .item_name = av_default_item_name,
575 .version = LIBAVUTIL_VERSION_INT,
578 AVCodec ff_ccaption_decoder = {
580 .long_name = NULL_IF_CONFIG_SMALL("Closed Caption (EIA-608 / CEA-708) Decoder"),
581 .type = AVMEDIA_TYPE_SUBTITLE,
582 .id = AV_CODEC_ID_EIA_608,
583 .priv_data_size = sizeof(CCaptionSubContext),
584 .init = init_decoder,
585 .close = close_decoder,
587 .priv_class = &ccaption_dec_class,