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