git.sesse.net Git - ffmpeg/blob - libavcodec/srtdec.c

   1 /*
   2  * SubRip subtitle decoder
   3  * Copyright (c) 2010  Aurelien Jacobs <aurel@gnuage.org>
   4  *
   5  * This file is part of FFmpeg.
   6  *
   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.
  11  *
  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.
  16  *
  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
  20  */
  21
  22 #include "libavutil/avstring.h"
  23 #include "libavutil/common.h"
  24 #include "libavutil/intreadwrite.h"
  25 #include "libavutil/parseutils.h"
  26 #include "avcodec.h"
  27 #include "ass.h"
  28
  29 static int html_color_parse(AVCodecContext *avctx, const char *str)
  30 {
  31     uint8_t rgba[4];
  32     if (av_parse_color(rgba, str, strcspn(str, "\" >"), avctx) < 0)
  33         return -1;
  34     return rgba[0] | rgba[1] << 8 | rgba[2] << 16;
  35 }
  36
  37 enum {
  38     PARAM_UNKNOWN = -1,
  39     PARAM_SIZE,
  40     PARAM_COLOR,
  41     PARAM_FACE,
  42     PARAM_NUMBER
  43 };
  44
  45 typedef struct {
  46     char tag[128];
  47     char param[PARAM_NUMBER][128];
  48 } SrtStack;
  49
  50 static const char *srt_to_ass(AVCodecContext *avctx, char *out, char *out_end,
  51                               const char *in, int x1, int y1, int x2, int y2)
  52 {
  53     char c, *param, buffer[128], tmp[128];
  54     int len, tag_close, sptr = 1, line_start = 1, an = 0, end = 0;
  55     SrtStack stack[16];
  56
  57     stack[0].tag[0] = 0;
  58     strcpy(stack[0].param[PARAM_SIZE],  "{\\fs}");
  59     strcpy(stack[0].param[PARAM_COLOR], "{\\c}");
  60     strcpy(stack[0].param[PARAM_FACE],  "{\\fn}");
  61
  62     if (x1 >= 0 && y1 >= 0) {
  63         if (x2 >= 0 && y2 >= 0 && (x2 != x1 || y2 != y1))
  64             snprintf(out, out_end-out,
  65                             "{\\an1}{\\move(%d,%d,%d,%d)}", x1, y1, x2, y2);
  66         else
  67             snprintf(out, out_end-out, "{\\an1}{\\pos(%d,%d)}", x1, y1);
  68         out += strlen(out);
  69     }
  70
  71     for (; out < out_end && !end && *in; in++) {
  72         switch (*in) {
  73         case '\r':
  74             break;
  75         case '\n':
  76             if (line_start) {
  77                 end = 1;
  78                 break;
  79             }
  80             while (out[-1] == ' ')
  81                 out--;
  82             snprintf(out, out_end-out, "\\N");
  83             if(out<out_end) out += strlen(out);
  84             line_start = 1;
  85             break;
  86         case ' ':
  87             if (!line_start)
  88                 *out++ = *in;
  89             break;
  90         case '{':    /* skip all {\xxx} substrings except for {\an%d}
  91                         and all microdvd like styles such as {Y:xxx} */
  92             an += sscanf(in, "{\\an%*1u}%c", &c) == 1;
  93             if ((an != 1 && sscanf(in, "{\\%*[^}]}%n%c", &len, &c) > 0) ||
  94                 sscanf(in, "{%*1[CcFfoPSsYy]:%*[^}]}%n%c", &len, &c) > 0) {
  95                 in += len - 1;
  96             } else
  97                 *out++ = *in;
  98             break;
  99         case '<':
 100             tag_close = in[1] == '/';
 101             if (sscanf(in+tag_close+1, "%127[^>]>%n%c", buffer, &len,&c) >= 2) {
 102                 if ((param = strchr(buffer, ' ')))
 103                     *param++ = 0;
 104                 if ((!tag_close && sptr < FF_ARRAY_ELEMS(stack)) ||
 105                     ( tag_close && sptr > 0 && !strcmp(stack[sptr-1].tag, buffer))) {
 106                     int i, j, unknown = 0;
 107                     in += len + tag_close;
 108                     if (!tag_close)
 109                         memset(stack+sptr, 0, sizeof(*stack));
 110                     if (!strcmp(buffer, "font")) {
 111                         if (tag_close) {
 112                             for (i=PARAM_NUMBER-1; i>=0; i--)
 113                                 if (stack[sptr-1].param[i][0])
 114                                     for (j=sptr-2; j>=0; j--)
 115                                         if (stack[j].param[i][0]) {
 116                                             snprintf(out, out_end-out,
 117                                                             "%s", stack[j].param[i]);
 118                                             if(out<out_end) out += strlen(out);
 119                                             break;
 120                                         }
 121                         } else {
 122                             while (param) {
 123                                 if (!strncmp(param, "size=", 5)) {
 124                                     unsigned font_size;
 125                                     param += 5 + (param[5] == '"');
 126                                     if (sscanf(param, "%u", &font_size) == 1) {
 127                                         snprintf(stack[sptr].param[PARAM_SIZE],
 128                                              sizeof(stack[0].param[PARAM_SIZE]),
 129                                              "{\\fs%u}", font_size);
 130                                     }
 131                                 } else if (!strncmp(param, "color=", 6)) {
 132                                     param += 6 + (param[6] == '"');
 133                                     snprintf(stack[sptr].param[PARAM_COLOR],
 134                                          sizeof(stack[0].param[PARAM_COLOR]),
 135                                          "{\\c&H%X&}",
 136                                          html_color_parse(avctx, param));
 137                                 } else if (!strncmp(param, "face=", 5)) {
 138                                     param += 5 + (param[5] == '"');
 139                                     len = strcspn(param,
 140                                                   param[-1] == '"' ? "\"" :" ");
 141                                     av_strlcpy(tmp, param,
 142                                                FFMIN(sizeof(tmp), len+1));
 143                                     param += len;
 144                                     snprintf(stack[sptr].param[PARAM_FACE],
 145                                              sizeof(stack[0].param[PARAM_FACE]),
 146                                              "{\\fn%s}", tmp);
 147                                 }
 148                                 if ((param = strchr(param, ' ')))
 149                                     param++;
 150                             }
 151                             for (i=0; i<PARAM_NUMBER; i++)
 152                                 if (stack[sptr].param[i][0]) {
 153                                     snprintf(out, out_end-out,
 154                                                     "%s", stack[sptr].param[i]);
 155                                     if(out<out_end) out += strlen(out);
 156                                 }
 157                         }
 158                     } else if (!buffer[1] && strspn(buffer, "bisu") == 1) {
 159                         snprintf(out, out_end-out,
 160                                         "{\\%c%d}", buffer[0], !tag_close);
 161                         if(out<out_end) out += strlen(out);
 162                     } else {
 163                         unknown = 1;
 164                         snprintf(tmp, sizeof(tmp), "</%s>", buffer);
 165                     }
 166                     if (tag_close) {
 167                         sptr--;
 168                     } else if (unknown && !strstr(in, tmp)) {
 169                         in -= len + tag_close;
 170                         *out++ = *in;
 171                     } else
 172                         av_strlcpy(stack[sptr++].tag, buffer,
 173                                    sizeof(stack[0].tag));
 174                     break;
 175                 }
 176             }
 177         default:
 178             *out++ = *in;
 179             break;
 180         }
 181         if (*in != ' ' && *in != '\r' && *in != '\n')
 182             line_start = 0;
 183     }
 184
 185     out = FFMIN(out, out_end-3);
 186     while (!strncmp(out-2, "\\N", 2))
 187         out -= 2;
 188     while (out[-1] == ' ')
 189         out--;
 190     snprintf(out, out_end-out, "\r\n");
 191     return in;
 192 }
 193
 194 static const char *read_ts(const char *buf, int *ts_start, int *ts_end,
 195                            int *x1, int *y1, int *x2, int *y2)
 196 {
 197     int i, hs, ms, ss, he, me, se;
 198
 199     for (i=0; i<2; i++) {
 200         /* try to read timestamps in either the first or second line */
 201         int c = sscanf(buf, "%d:%2d:%2d%*1[,.]%3d --> %d:%2d:%2d%*1[,.]%3d"
 202                        "%*[ ]X1:%u X2:%u Y1:%u Y2:%u",
 203                        &hs, &ms, &ss, ts_start, &he, &me, &se, ts_end,
 204                        x1, x2, y1, y2);
 205         buf += strcspn(buf, "\n") + 1;
 206         if (c >= 8) {
 207             *ts_start = 100*(ss + 60*(ms + 60*hs)) + *ts_start/10;
 208             *ts_end   = 100*(se + 60*(me + 60*he)) + *ts_end  /10;
 209             return buf;
 210         }
 211     }
 212     return NULL;
 213 }
 214
 215 static int srt_decode_frame(AVCodecContext *avctx,
 216                             void *data, int *got_sub_ptr, AVPacket *avpkt)
 217 {
 218     AVSubtitle *sub = data;
 219     int ts_start, ts_end, x1 = -1, y1 = -1, x2 = -1, y2 = -1;
 220     char buffer[2048];
 221     const char *ptr = avpkt->data;
 222     const char *end = avpkt->data + avpkt->size;
 223     int size;
 224     const uint8_t *p = av_packet_get_side_data(avpkt, AV_PKT_DATA_SUBTITLE_POSITION, &size);
 225
 226     if (p && size == 16) {
 227         x1 = AV_RL32(p     );
 228         y1 = AV_RL32(p +  4);
 229         x2 = AV_RL32(p +  8);
 230         y2 = AV_RL32(p + 12);
 231     }
 232
 233     if (avpkt->size <= 0)
 234         return avpkt->size;
 235
 236     while (ptr < end && *ptr) {
 237         if (avctx->codec->id == AV_CODEC_ID_SRT) {
 238             ptr = read_ts(ptr, &ts_start, &ts_end, &x1, &y1, &x2, &y2);
 239             if (!ptr)
 240                 break;
 241         } else {
 242             // Do final divide-by-10 outside rescale to force rounding down.
 243             ts_start = av_rescale_q(avpkt->pts,
 244                                     avctx->time_base,
 245                                     (AVRational){1,100});
 246             ts_end   = av_rescale_q(avpkt->pts + avpkt->duration,
 247                                     avctx->time_base,
 248                                     (AVRational){1,100});
 249         }
 250         ptr = srt_to_ass(avctx, buffer, buffer+sizeof(buffer), ptr,
 251                          x1, y1, x2, y2);
 252         ff_ass_add_rect(sub, buffer, ts_start, ts_end-ts_start, 0);
 253     }
 254
 255     *got_sub_ptr = sub->num_rects > 0;
 256     return avpkt->size;
 257 }
 258
 259 #if CONFIG_SRT_DECODER
 260 /* deprecated decoder */
 261 AVCodec ff_srt_decoder = {
 262     .name         = "srt",
 263     .long_name    = NULL_IF_CONFIG_SMALL("SubRip subtitle with embedded timing"),
 264     .type         = AVMEDIA_TYPE_SUBTITLE,
 265     .id           = AV_CODEC_ID_SRT,
 266     .init         = ff_ass_subtitle_header_default,
 267     .decode       = srt_decode_frame,
 268 };
 269 #endif
 270
 271 #if CONFIG_SUBRIP_DECODER
 272 AVCodec ff_subrip_decoder = {
 273     .name         = "subrip",
 274     .long_name    = NULL_IF_CONFIG_SMALL("SubRip subtitle"),
 275     .type         = AVMEDIA_TYPE_SUBTITLE,
 276     .id           = AV_CODEC_ID_SUBRIP,
 277     .init         = ff_ass_subtitle_header_default,
 278     .decode       = srt_decode_frame,
 279 };
 280 #endif