avcodec/ass: add ASS trailing \r\n in one single place
[ffmpeg.git] / 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 void rstrip_spaces_buf(AVBPrint *buf)
51 {
52     while (buf->len > 0 && buf->str[buf->len - 1] == ' ')
53         buf->str[--buf->len] = 0;
54 }
55
56 static void srt_to_ass(AVCodecContext *avctx, AVBPrint *dst,
57                        const char *in, int x1, int y1, int x2, int y2)
58 {
59     char *param, buffer[128], tmp[128];
60     int len, tag_close, sptr = 1, line_start = 1, an = 0, end = 0;
61     SrtStack stack[16];
62
63     stack[0].tag[0] = 0;
64     strcpy(stack[0].param[PARAM_SIZE],  "{\\fs}");
65     strcpy(stack[0].param[PARAM_COLOR], "{\\c}");
66     strcpy(stack[0].param[PARAM_FACE],  "{\\fn}");
67
68     if (x1 >= 0 && y1 >= 0) {
69         if (x2 >= 0 && y2 >= 0 && (x2 != x1 || y2 != y1))
70             av_bprintf(dst, "{\\an1}{\\move(%d,%d,%d,%d)}", x1, y1, x2, y2);
71         else
72             av_bprintf(dst, "{\\an1}{\\pos(%d,%d)}", x1, y1);
73     }
74
75     for (; !end && *in; in++) {
76         switch (*in) {
77         case '\r':
78             break;
79         case '\n':
80             if (line_start) {
81                 end = 1;
82                 break;
83             }
84             rstrip_spaces_buf(dst);
85             av_bprintf(dst, "\\N");
86             line_start = 1;
87             break;
88         case ' ':
89             if (!line_start)
90                 av_bprint_chars(dst, *in, 1);
91             break;
92         case '{':    /* skip all {\xxx} substrings except for {\an%d}
93                         and all microdvd like styles such as {Y:xxx} */
94             len = 0;
95             an += sscanf(in, "{\\an%*1u}%n", &len) >= 0 && len > 0;
96             if ((an != 1 && (len = 0, sscanf(in, "{\\%*[^}]}%n", &len) >= 0 && len > 0)) ||
97                 (len = 0, sscanf(in, "{%*1[CcFfoPSsYy]:%*[^}]}%n", &len) >= 0 && len > 0)) {
98                 in += len - 1;
99             } else
100                 av_bprint_chars(dst, *in, 1);
101             break;
102         case '<':
103             tag_close = in[1] == '/';
104             len = 0;
105             if (sscanf(in+tag_close+1, "%127[^>]>%n", buffer, &len) >= 1 && len > 0) {
106                 if ((param = strchr(buffer, ' ')))
107                     *param++ = 0;
108                 if ((!tag_close && sptr < FF_ARRAY_ELEMS(stack)) ||
109                     ( tag_close && sptr > 0 && !strcmp(stack[sptr-1].tag, buffer))) {
110                     int i, j, unknown = 0;
111                     in += len + tag_close;
112                     if (!tag_close)
113                         memset(stack+sptr, 0, sizeof(*stack));
114                     if (!strcmp(buffer, "font")) {
115                         if (tag_close) {
116                             for (i=PARAM_NUMBER-1; i>=0; i--)
117                                 if (stack[sptr-1].param[i][0])
118                                     for (j=sptr-2; j>=0; j--)
119                                         if (stack[j].param[i][0]) {
120                                             av_bprintf(dst, "%s", stack[j].param[i]);
121                                             break;
122                                         }
123                         } else {
124                             while (param) {
125                                 if (!strncmp(param, "size=", 5)) {
126                                     unsigned font_size;
127                                     param += 5 + (param[5] == '"');
128                                     if (sscanf(param, "%u", &font_size) == 1) {
129                                         snprintf(stack[sptr].param[PARAM_SIZE],
130                                              sizeof(stack[0].param[PARAM_SIZE]),
131                                              "{\\fs%u}", font_size);
132                                     }
133                                 } else if (!strncmp(param, "color=", 6)) {
134                                     param += 6 + (param[6] == '"');
135                                     snprintf(stack[sptr].param[PARAM_COLOR],
136                                          sizeof(stack[0].param[PARAM_COLOR]),
137                                          "{\\c&H%X&}",
138                                          html_color_parse(avctx, param));
139                                 } else if (!strncmp(param, "face=", 5)) {
140                                     param += 5 + (param[5] == '"');
141                                     len = strcspn(param,
142                                                   param[-1] == '"' ? "\"" :" ");
143                                     av_strlcpy(tmp, param,
144                                                FFMIN(sizeof(tmp), len+1));
145                                     param += len;
146                                     snprintf(stack[sptr].param[PARAM_FACE],
147                                              sizeof(stack[0].param[PARAM_FACE]),
148                                              "{\\fn%s}", tmp);
149                                 }
150                                 if ((param = strchr(param, ' ')))
151                                     param++;
152                             }
153                             for (i=0; i<PARAM_NUMBER; i++)
154                                 if (stack[sptr].param[i][0])
155                                     av_bprintf(dst, "%s", stack[sptr].param[i]);
156                         }
157                     } else if (!buffer[1] && strspn(buffer, "bisu") == 1) {
158                         av_bprintf(dst, "{\\%c%d}", buffer[0], !tag_close);
159                     } else {
160                         unknown = 1;
161                         snprintf(tmp, sizeof(tmp), "</%s>", buffer);
162                     }
163                     if (tag_close) {
164                         sptr--;
165                     } else if (unknown && !strstr(in, tmp)) {
166                         in -= len + tag_close;
167                         av_bprint_chars(dst, *in, 1);
168                     } else
169                         av_strlcpy(stack[sptr++].tag, buffer,
170                                    sizeof(stack[0].tag));
171                     break;
172                 }
173             }
174         default:
175             av_bprint_chars(dst, *in, 1);
176             break;
177         }
178         if (*in != ' ' && *in != '\r' && *in != '\n')
179             line_start = 0;
180     }
181
182     while (dst->len >= 2 && !strncmp(&dst->str[dst->len - 2], "\\N", 2))
183         dst->len -= 2;
184     dst->str[dst->len] = 0;
185     rstrip_spaces_buf(dst);
186 }
187
188 static int srt_decode_frame(AVCodecContext *avctx,
189                             void *data, int *got_sub_ptr, AVPacket *avpkt)
190 {
191     AVSubtitle *sub = data;
192     AVBPrint buffer;
193     int ts_start, ts_end, x1 = -1, y1 = -1, x2 = -1, y2 = -1;
194     int size, ret;
195     const uint8_t *p = av_packet_get_side_data(avpkt, AV_PKT_DATA_SUBTITLE_POSITION, &size);
196
197     if (p && size == 16) {
198         x1 = AV_RL32(p     );
199         y1 = AV_RL32(p +  4);
200         x2 = AV_RL32(p +  8);
201         y2 = AV_RL32(p + 12);
202     }
203
204     if (avpkt->size <= 0)
205         return avpkt->size;
206
207     av_bprint_init(&buffer, 0, AV_BPRINT_SIZE_UNLIMITED);
208
209         // TODO: reindent
210             // Do final divide-by-10 outside rescale to force rounding down.
211             ts_start = av_rescale_q(avpkt->pts,
212                                     avctx->time_base,
213                                     (AVRational){1,100});
214             ts_end   = av_rescale_q(avpkt->pts + avpkt->duration,
215                                     avctx->time_base,
216                                     (AVRational){1,100});
217
218     srt_to_ass(avctx, &buffer, avpkt->data, x1, y1, x2, y2);
219     ret = ff_ass_add_rect_bprint(sub, &buffer, ts_start, ts_end-ts_start);
220     av_bprint_finalize(&buffer, NULL);
221     if (ret < 0)
222         return ret;
223
224     *got_sub_ptr = sub->num_rects > 0;
225     return avpkt->size;
226 }
227
228 #if CONFIG_SRT_DECODER
229 /* deprecated decoder */
230 AVCodec ff_srt_decoder = {
231     .name         = "srt",
232     .long_name    = NULL_IF_CONFIG_SMALL("SubRip subtitle"),
233     .type         = AVMEDIA_TYPE_SUBTITLE,
234     .id           = AV_CODEC_ID_SUBRIP,
235     .init         = ff_ass_subtitle_header_default,
236     .decode       = srt_decode_frame,
237 };
238 #endif
239
240 #if CONFIG_SUBRIP_DECODER
241 AVCodec ff_subrip_decoder = {
242     .name         = "subrip",
243     .long_name    = NULL_IF_CONFIG_SMALL("SubRip subtitle"),
244     .type         = AVMEDIA_TYPE_SUBTITLE,
245     .id           = AV_CODEC_ID_SUBRIP,
246     .init         = ff_ass_subtitle_header_default,
247     .decode       = srt_decode_frame,
248 };
249 #endif