?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
demux_lavf: fix initial "-vid"-based video selection
[mplayer/greg.git] / sub / subreader.c
blob1f9294ec47b44bd60c54793aded2a8ea5c6c3b41
1 /*
2 * Subtitle reader with format autodetection
3 *
4 * Copyright (c) 2001 laaz
5 * Some code cleanup & realloc() by A'rpi/ESP-team
6 *
7 * This file is part of MPlayer.
8 *
9 * MPlayer is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU General Public License as published by
11 * the Free Software Foundation; either version 2 of the License, or
12 * (at your option) any later version.
13 *
14 * MPlayer is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 * GNU General Public License for more details.
18 *
19 * You should have received a copy of the GNU General Public License along
20 * with MPlayer; if not, write to the Free Software Foundation, Inc.,
21 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
22 */
23
24 #include <stdio.h>
25 #include <stdlib.h>
26 #include <string.h>
27 #include <ctype.h>
28
29 #include <sys/types.h>
30 #include <dirent.h>
31
32 #include "config.h"
33 #include "mp_msg.h"
34 #include "subreader.h"
35 #include "mpcommon.h"
36 #include "subassconvert.h"
37 #include "options.h"
38 #include "stream/stream.h"
39 #include "libavutil/common.h"
40 #include "libavutil/avstring.h"
41
42 #ifdef CONFIG_ENCA
43 #include <enca.h>
44 #endif
45
46 #define ERR ((void *) -1)
47
48 #ifdef CONFIG_ICONV
49 #include <iconv.h>
50 char *sub_cp=NULL;
51 #endif
52 #ifdef CONFIG_FRIBIDI
53 #include <fribidi/fribidi.h>
54 char *fribidi_charset = NULL; ///character set that will be passed to FriBiDi
55 int flip_hebrew = 1; ///flip subtitles using fribidi
56 int fribidi_flip_commas = 0; ///flip comma when fribidi is used
57 #endif
58
59 // Parameter struct for the format-specific readline functions
60 struct readline_args {
61 int utf16;
62 struct MPOpts *opts;
63 };
64
65 /* Maximal length of line of a subtitle */
66 #define LINE_LEN 1000
67 static float mpsub_position=0;
68 static float mpsub_multiplier=1.;
69 static int sub_slacktime = 20000; //20 sec
70
71 int sub_no_text_pp=0; // 1 => do not apply text post-processing
72 // like {\...} elimination in SSA format.
73
74 int sub_match_fuzziness=0; // level of sub name matching fuzziness
75
76 /* Use the SUB_* constant defined in the header file */
77 int sub_format=SUB_INVALID;
78 #ifdef CONFIG_SORTSUB
79 /*
80 Some subtitling formats, namely AQT and Subrip09, define the end of a
81 subtitle as the beginning of the following. Since currently we read one
82 subtitle at time, for these format we keep two global *subtitle,
83 previous_aqt_sub and previous_subrip09_sub, pointing to previous subtitle,
84 so we can change its end when we read current subtitle starting time.
85 When CONFIG_SORTSUB is defined, we use a single global unsigned long,
86 previous_sub_end, for both (and even future) formats, to store the end of
87 the previous sub: it is initialized to 0 in sub_read_file and eventually
88 modified by sub_read_aqt_line or sub_read_subrip09_line.
89 */
90 unsigned long previous_sub_end;
91 #endif
92
93 static int eol(char p) {
94 return p=='\r' || p=='\n' || p=='\0';
95 }
96
97 /* Remove leading and trailing space */
98 static void trail_space(char *s) {
99 int i = 0;
100 while (isspace(s[i])) ++i;
101 if (i) strcpy(s, s + i);
102 i = strlen(s) - 1;
103 while (i > 0 && isspace(s[i])) s[i--] = '\0';
104 }
105
106 static char *stristr(const char *haystack, const char *needle) {
107 int len = 0;
108 const char *p = haystack;
109
110 if (!(haystack && needle)) return NULL;
111
112 len=strlen(needle);
113 while (*p != '\0') {
114 if (strncasecmp(p, needle, len) == 0) return (char*)p;
115 p++;
116 }
117
118 return NULL;
119 }
120
121 static void sami_add_line(subtitle *current, char *buffer, char **pos) {
122 char *p = *pos;
123 *p = 0;
124 trail_space(buffer);
125 if (*buffer && current->lines < SUB_MAX_TEXT)
126 current->text[current->lines++] = strdup(buffer);
127 *pos = buffer;
128 }
129
130 static subtitle *sub_read_line_sami(stream_t* st, subtitle *current,
131 struct readline_args *args)
132 {
133 int utf16 = args->utf16;
134 static char line[LINE_LEN+1];
135 static char *s = NULL, *slacktime_s;
136 char text[LINE_LEN+1], *p=NULL, *q;
137 int state;
138
139 current->lines = current->start = current->end = 0;
140 current->alignment = SUB_ALIGNMENT_BOTTOMCENTER;
141 state = 0;
142
143 /* read the first line */
144 if (!s)
145 if (!(s = stream_read_line(st, line, LINE_LEN, utf16))) return 0;
146
147 do {
148 switch (state) {
149
150 case 0: /* find "START=" or "Slacktime:" */
151 slacktime_s = stristr (s, "Slacktime:");
152 if (slacktime_s)
153 sub_slacktime = strtol (slacktime_s+10, NULL, 0) / 10;
154
155 s = stristr (s, "Start=");
156 if (s) {
157 current->start = strtol (s + 6, &s, 0) / 10;
158 /* eat '>' */
159 for (; *s != '>' && *s != '\0'; s++);
160 s++;
161 state = 1; continue;
162 }
163 break;
164
165 case 1: /* find (optional) "<P", skip other TAGs */
166 for (; *s == ' ' || *s == '\t'; s++); /* strip blanks, if any */
167 if (*s == '\0') break;
168 if (*s != '<') { state = 3; p = text; continue; } /* not a TAG */
169 s++;
170 if (*s == 'P' || *s == 'p') { s++; state = 2; continue; } /* found '<P' */
171 for (; *s != '>' && *s != '\0'; s++); /* skip remains of non-<P> TAG */
172 if (s == '\0')
173 break;
174 s++;
175 continue;
176
177 case 2: /* find ">" */
178 if ((s = strchr (s, '>'))) { s++; state = 3; p = text; continue; }
179 break;
180
181 case 3: /* get all text until '<' appears */
182 if (*s == '\0') break;
183 else if (!strncasecmp (s, "<br>", 4)) {
184 sami_add_line(current, text, &p);
185 s += 4;
186 }
187 else if ((*s == '{') && !sub_no_text_pp) { state = 5; ++s; continue; }
188 else if (*s == '<') { state = 4; }
189 else if (!strncasecmp (s, "&nbsp;", 6)) { *p++ = ' '; s += 6; }
190 else if (*s == '\t') { *p++ = ' '; s++; }
191 else if (*s == '\r' || *s == '\n') { s++; }
192 else *p++ = *s++;
193
194 /* skip duplicated space */
195 if (p > text + 2) if (*(p-1) == ' ' && *(p-2) == ' ') p--;
196
197 continue;
198
199 case 4: /* get current->end or skip <TAG> */
200 q = stristr (s, "Start=");
201 if (q) {
202 current->end = strtol (q + 6, &q, 0) / 10 - 1;
203 *p = '\0'; trail_space (text);
204 if (text[0] != '\0')
205 current->text[current->lines++] = strdup (text);
206 if (current->lines > 0) { state = 99; break; }
207 state = 0; continue;
208 }
209 s = strchr (s, '>');
210 if (s) { s++; state = 3; continue; }
211 break;
212 case 5: /* get rid of {...} text, but read the alignment code */
213 if ((*s == '\\') && (*(s + 1) == 'a') && !sub_no_text_pp) {
214 if (stristr(s, "\\a1") != NULL) {
215 current->alignment = SUB_ALIGNMENT_BOTTOMLEFT;
216 s = s + 3;
217 }
218 if (stristr(s, "\\a2") != NULL) {
219 current->alignment = SUB_ALIGNMENT_BOTTOMCENTER;
220 s = s + 3;
221 } else if (stristr(s, "\\a3") != NULL) {
222 current->alignment = SUB_ALIGNMENT_BOTTOMRIGHT;
223 s = s + 3;
224 } else if ((stristr(s, "\\a4") != NULL) || (stristr(s, "\\a5") != NULL) || (stristr(s, "\\a8") != NULL)) {
225 current->alignment = SUB_ALIGNMENT_TOPLEFT;
226 s = s + 3;
227 } else if (stristr(s, "\\a6") != NULL) {
228 current->alignment = SUB_ALIGNMENT_TOPCENTER;
229 s = s + 3;
230 } else if (stristr(s, "\\a7") != NULL) {
231 current->alignment = SUB_ALIGNMENT_TOPRIGHT;
232 s = s + 3;
233 } else if (stristr(s, "\\a9") != NULL) {
234 current->alignment = SUB_ALIGNMENT_MIDDLELEFT;
235 s = s + 3;
236 } else if (stristr(s, "\\a10") != NULL) {
237 current->alignment = SUB_ALIGNMENT_MIDDLECENTER;
238 s = s + 4;
239 } else if (stristr(s, "\\a11") != NULL) {
240 current->alignment = SUB_ALIGNMENT_MIDDLERIGHT;
241 s = s + 4;
242 }
243 }
244 if (*s == '}') state = 3;
245 ++s;
246 continue;
247 }
248
249 /* read next line */
250 if (state != 99 && !(s = stream_read_line (st, line, LINE_LEN, utf16))) {
251 if (current->start > 0) {
252 break; // if it is the last subtitle
253 } else {
254 return 0;
255 }
256 }
257
258 } while (state != 99);
259
260 // For the last subtitle
261 if (current->end <= 0) {
262 current->end = current->start + sub_slacktime;
263 sami_add_line(current, text, &p);
264 }
265
266 return current;
267 }
268
269
270 static char *sub_readtext(char *source, char **dest) {
271 int len=0;
272 char *p=source;
273
274 // printf("src=%p dest=%p \n",source,dest);
275
276 while ( !eol(*p) && *p!= '|' ) {
277 p++,len++;
278 }
279
280 *dest= malloc (len+1);
281 if (!dest) {return ERR;}
282
283 strncpy(*dest, source, len);
284 (*dest)[len]=0;
285
286 while (*p=='\r' || *p=='\n' || *p=='|') p++;
287
288 if (*p) return p; // not-last text field
289 else return NULL; // last text field
290 }
291
292 static subtitle *sub_read_line_microdvd(stream_t *st,subtitle *current,
293 struct readline_args *args)
294 {
295 int utf16 = args->utf16;
296 char line[LINE_LEN+1];
297 char line2[LINE_LEN+1];
298 char *p, *next;
299 int i;
300
301 do {
302 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
303 } while ((sscanf (line,
304 "{%ld}{}%[^\r\n]",
305 &(current->start), line2) < 2) &&
306 (sscanf (line,
307 "{%ld}{%ld}%[^\r\n]",
308 &(current->start), &(current->end), line2) < 3));
309
310 if (args->opts->ass_enabled) {
311 subassconvert_microdvd(line2, line, LINE_LEN + 1);
312 p = line;
313 } else
314 p = line2;
315
316 next=p, i=0;
317 while ((next =sub_readtext (next, &(current->text[i])))) {
318 if (current->text[i]==ERR) {return ERR;}
319 i++;
320 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
321 }
322 current->lines= ++i;
323
324 return current;
325 }
326
327 static subtitle *sub_read_line_mpl2(stream_t *st,subtitle *current,
328 struct readline_args *args)
329 {
330 int utf16 = args->utf16;
331 char line[LINE_LEN+1];
332 char line2[LINE_LEN+1];
333 char *p, *next;
334 int i;
335
336 do {
337 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
338 } while ((sscanf (line,
339 "[%ld][%ld]%[^\r\n]",
340 &(current->start), &(current->end), line2) < 3));
341 current->start *= 10;
342 current->end *= 10;
343 p=line2;
344
345 next=p, i=0;
346 while ((next =sub_readtext (next, &(current->text[i])))) {
347 if (current->text[i]==ERR) {return ERR;}
348 i++;
349 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
350 }
351 current->lines= ++i;
352
353 return current;
354 }
355
356 static subtitle *sub_read_line_subrip(stream_t* st, subtitle *current,
357 struct readline_args *args)
358 {
359 int utf16 = args->utf16;
360 char line[LINE_LEN+1];
361 int a1,a2,a3,a4,b1,b2,b3,b4;
362 char *p=NULL, *q=NULL;
363 int len;
364
365 while (1) {
366 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
367 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4) < 8) continue;
368 current->start = a1*360000+a2*6000+a3*100+a4;
369 current->end = b1*360000+b2*6000+b3*100+b4;
370
371 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
372
373 p=q=line;
374 for (current->lines=1; current->lines < SUB_MAX_TEXT; current->lines++) {
375 for (q=p,len=0; *p && *p!='\r' && *p!='\n' && *p!='|' && strncmp(p,"[br]",4); p++,len++);
376 current->text[current->lines-1]=malloc (len+1);
377 if (!current->text[current->lines-1]) return ERR;
378 strncpy (current->text[current->lines-1], q, len);
379 current->text[current->lines-1][len]='\0';
380 if (!*p || *p=='\r' || *p=='\n') break;
381 if (*p=='|') p++;
382 else while (*p++!=']');
383 }
384 break;
385 }
386 return current;
387 }
388
389 static subtitle *sub_ass_read_line_subviewer(stream_t *st, subtitle *current,
390 struct readline_args *args)
391 {
392 int utf16 = args->utf16;
393 int a1, a2, a3, a4, b1, b2, b3, b4, j = 0;
394
395 while (!current->text[0]) {
396 char line[LINE_LEN + 1], full_line[LINE_LEN + 1], sep;
397 int i;
398
399 /* Parse SubRip header */
400 if (!stream_read_line(st, line, LINE_LEN, utf16))
401 return NULL;
402 if (sscanf(line, "%d:%d:%d%[,.:]%d --> %d:%d:%d%[,.:]%d",
403 &a1, &a2, &a3, &sep, &a4, &b1, &b2, &b3, &sep, &b4) < 10)
404 continue;
405
406 current->start = a1 * 360000 + a2 * 6000 + a3 * 100 + a4 / 10;
407 current->end = b1 * 360000 + b2 * 6000 + b3 * 100 + b4 / 10;
408
409 /* Concat lines */
410 full_line[0] = 0;
411 for (i = 0; i < SUB_MAX_TEXT; i++) {
412 int blank = 1, len = 0;
413 char *p;
414
415 if (!stream_read_line(st, line, LINE_LEN, utf16))
416 break;
417
418 for (p = line; *p != '\n' && *p != '\r' && *p; p++, len++)
419 if (*p != ' ' && *p != '\t')
420 blank = 0;
421
422 if (blank)
423 break;
424
425 *p = 0;
426
427 if (!(j + 1 + len < sizeof(full_line) - 1))
428 break;
429
430 if (j != 0)
431 full_line[j++] = '\n';
432 strcpy(&full_line[j], line);
433 j += len;
434 }
435
436 /* Use the ASS/SSA converter to transform the whole lines */
437 if (full_line[0]) {
438 char converted_line[LINE_LEN + 1];
439 subassconvert_subrip(full_line, converted_line, LINE_LEN + 1);
440 current->text[0] = strdup(converted_line);
441 current->lines = 1;
442 }
443 }
444 return current;
445 }
446
447 static subtitle *sub_read_line_subviewer(stream_t *st,subtitle *current,
448 struct readline_args *args)
449 {
450 int utf16 = args->utf16;
451 char line[LINE_LEN+1];
452 int a1,a2,a3,a4,b1,b2,b3,b4;
453 char *p=NULL;
454 int i,len;
455
456 if (args->opts->ass_enabled)
457 return sub_ass_read_line_subviewer(st, current, args);
458 while (!current->text[0]) {
459 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
460 if ((len=sscanf (line, "%d:%d:%d%[,.:]%d --> %d:%d:%d%[,.:]%d",&a1,&a2,&a3,(char *)&i,&a4,&b1,&b2,&b3,(char *)&i,&b4)) < 10)
461 continue;
462 current->start = a1*360000+a2*6000+a3*100+a4/10;
463 current->end = b1*360000+b2*6000+b3*100+b4/10;
464 for (i=0; i<SUB_MAX_TEXT;) {
465 int blank = 1;
466 if (!stream_read_line (st, line, LINE_LEN, utf16)) break;
467 len=0;
468 for (p=line; *p!='\n' && *p!='\r' && *p; p++,len++)
469 if (*p != ' ' && *p != '\t')
470 blank = 0;
471 if (len && !blank) {
472 int j=0,skip=0;
473 char *curptr=current->text[i]=malloc (len+1);
474 if (!current->text[i]) return ERR;
475 //strncpy (current->text[i], line, len); current->text[i][len]='\0';
476 for(; j<len; j++) {
477 /* let's filter html tags ::atmos */
478 if(line[j]=='>') {
479 skip=0;
480 continue;
481 }
482 if(line[j]=='<') {
483 skip=1;
484 continue;
485 }
486 if(skip) {
487 continue;
488 }
489 *curptr=line[j];
490 curptr++;
491 }
492 *curptr='\0';
493
494 i++;
495 } else {
496 break;
497 }
498 }
499 current->lines=i;
500 }
501 return current;
502 }
503
504 static subtitle *sub_read_line_subviewer2(stream_t *st,subtitle *current,
505 struct readline_args *args)
506 {
507 int utf16 = args->utf16;
508 char line[LINE_LEN+1];
509 int a1,a2,a3,a4;
510 char *p=NULL;
511 int i,len;
512
513 while (!current->text[0]) {
514 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
515 if (line[0]!='{')
516 continue;
517 if ((len=sscanf (line, "{T %d:%d:%d:%d",&a1,&a2,&a3,&a4)) < 4)
518 continue;
519 current->start = a1*360000+a2*6000+a3*100+a4/10;
520 for (i=0; i<SUB_MAX_TEXT;) {
521 if (!stream_read_line (st, line, LINE_LEN, utf16)) break;
522 if (line[0]=='}') break;
523 len=0;
524 for (p=line; *p!='\n' && *p!='\r' && *p; ++p,++len);
525 if (len) {
526 current->text[i]=malloc (len+1);
527 if (!current->text[i]) return ERR;
528 strncpy (current->text[i], line, len); current->text[i][len]='\0';
529 ++i;
530 } else {
531 break;
532 }
533 }
534 current->lines=i;
535 }
536 return current;
537 }
538
539
540 static subtitle *sub_read_line_vplayer(stream_t *st,subtitle *current,
541 struct readline_args *args)
542 {
543 int utf16 = args->utf16;
544 char line[LINE_LEN+1];
545 int a1,a2,a3;
546 char *p=NULL, *next,separator;
547 int i,len,plen;
548
549 while (!current->text[0]) {
550 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
551 if ((len=sscanf (line, "%d:%d:%d%c%n",&a1,&a2,&a3,&separator,&plen)) < 4)
552 continue;
553
554 if (!(current->start = a1*360000+a2*6000+a3*100))
555 continue;
556 /* removed by wodzu
557 p=line;
558 // finds the body of the subtitle
559 for (i=0; i<3; i++){
560 p=strchr(p,':');
561 if (p==NULL) break;
562 ++p;
563 }
564 if (p==NULL) {
565 printf("SUB: Skipping incorrect subtitle line!\n");
566 continue;
567 }
568 */
569 // by wodzu: hey! this time we know what length it has! what is
570 // that magic for? it can't deal with space instead of third
571 // colon! look, what simple it can be:
572 p = &line[ plen ];
573
574 i=0;
575 if (*p!='|') {
576 //
577 next = p,i=0;
578 while ((next =sub_readtext (next, &(current->text[i])))) {
579 if (current->text[i]==ERR) {return ERR;}
580 i++;
581 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
582 }
583 current->lines=i+1;
584 }
585 }
586 return current;
587 }
588
589 static subtitle *sub_read_line_rt(stream_t *st,subtitle *current,
590 struct readline_args *args)
591 {
592 int utf16 = args->utf16;
593
594 //TODO: This format uses quite rich (sub/super)set of xhtml
595 // I couldn't check it since DTD is not included.
596 // WARNING: full XML parses can be required for proper parsing
597 char line[LINE_LEN+1];
598 int a1,a2,a3,a4,b1,b2,b3,b4;
599 char *p=NULL,*next=NULL;
600 int i,len,plen;
601
602 while (!current->text[0]) {
603 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
604 //TODO: it seems that format of time is not easily determined, it may be 1:12, 1:12.0 or 0:1:12.0
605 //to describe the same moment in time. Maybe there are even more formats in use.
606 //if ((len=sscanf (line, "<Time Begin=\"%d:%d:%d.%d\" End=\"%d:%d:%d.%d\"",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8)
607 plen=a1=a2=a3=a4=b1=b2=b3=b4=0;
608 if (
609 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d.%d\" %*[Ee]nd=\"%d.%d\"%*[^<]<clear/>%n",&a3,&a4,&b3,&b4,&plen)) < 4) &&
610 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d.%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a3,&a4,&b2,&b3,&b4,&plen)) < 5) &&
611 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&plen)) < 4) &&
612 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&b4,&plen)) < 5) &&
613 // ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&b2,&b3,&plen)) < 5) &&
614 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&b2,&b3,&b4,&plen)) < 6) &&
615 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d:%d.%d\" %*[Ee]nd=\"%d:%d:%d.%d\"%*[^<]<clear/>%n",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4,&plen)) < 8) &&
616 //now try it without end time
617 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d.%d\"%*[^<]<clear/>%n",&a3,&a4,&plen)) < 2) &&
618 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&plen)) < 2) &&
619 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&plen)) < 3) &&
620 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d:%d.%d\"%*[^<]<clear/>%n",&a1,&a2,&a3,&a4,&plen)) < 4)
621 )
622 continue;
623 current->start = a1*360000+a2*6000+a3*100+a4/10;
624 current->end = b1*360000+b2*6000+b3*100+b4/10;
625 if (b1 == 0 && b2 == 0 && b3 == 0 && b4 == 0)
626 current->end = current->start+200;
627 p=line; p+=plen;i=0;
628 // TODO: I don't know what kind of convention is here for marking multiline subs, maybe <br/> like in xml?
629 next = strstr(line,"<clear/>");
630 if(next && strlen(next)>8){
631 next+=8;i=0;
632 while ((next =sub_readtext (next, &(current->text[i])))) {
633 if (current->text[i]==ERR) {return ERR;}
634 i++;
635 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
636 }
637 }
638 current->lines=i+1;
639 }
640 return current;
641 }
642
643 static subtitle *sub_read_line_ssa(stream_t *st,subtitle *current,
644 struct readline_args *args)
645 {
646 /*
647 * Sub Station Alpha v4 (and v2?) scripts have 9 commas before subtitle
648 * other Sub Station Alpha scripts have only 8 commas before subtitle
649 * Reading the "ScriptType:" field is not reliable since many scripts appear
650 * w/o it
651 *
652 * http://www.scriptclub.org is a good place to find more examples
653 * http://www.eswat.demon.co.uk is where the SSA specs can be found
654 */
655 int utf16 = args->utf16;
656 int comma;
657 static int max_comma = 32; /* let's use 32 for the case that the */
658 /* amount of commas increase with newer SSA versions */
659
660 int hour1, min1, sec1, hunsec1,
661 hour2, min2, sec2, hunsec2, nothing;
662 int num;
663
664 char line[LINE_LEN+1],
665 line3[LINE_LEN+1],
666 *line2;
667 char *tmp;
668
669 do {
670 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
671 } while (sscanf (line, "Dialogue: Marked=%d,%d:%d:%d.%d,%d:%d:%d.%d"
672 "%[^\n\r]", &nothing,
673 &hour1, &min1, &sec1, &hunsec1,
674 &hour2, &min2, &sec2, &hunsec2,
675 line3) < 9
676 &&
677 sscanf (line, "Dialogue: %d,%d:%d:%d.%d,%d:%d:%d.%d"
678 "%[^\n\r]", &nothing,
679 &hour1, &min1, &sec1, &hunsec1,
680 &hour2, &min2, &sec2, &hunsec2,
681 line3) < 9 );
682
683 line2=strchr(line3, ',');
684 if (!line2) return NULL;
685
686 for (comma = 4; comma < max_comma; comma ++)
687 {
688 tmp = line2;
689 if(!(tmp=strchr(++tmp, ','))) break;
690 if(*(++tmp) == ' ') break;
691 /* a space after a comma means we're already in a sentence */
692 line2 = tmp;
693 }
694
695 if(comma < max_comma)max_comma = comma;
696 /* eliminate the trailing comma */
697 if(*line2 == ',') line2++;
698
699 current->lines=0;num=0;
700 current->start = 360000*hour1 + 6000*min1 + 100*sec1 + hunsec1;
701 current->end = 360000*hour2 + 6000*min2 + 100*sec2 + hunsec2;
702
703 while (((tmp=strstr(line2, "\\n")) != NULL) || ((tmp=strstr(line2, "\\N")) != NULL) ){
704 current->text[num]=malloc(tmp-line2+1);
705 strncpy (current->text[num], line2, tmp-line2);
706 current->text[num][tmp-line2]='\0';
707 line2=tmp+2;
708 num++;
709 current->lines++;
710 if (current->lines >= SUB_MAX_TEXT) return current;
711 }
712
713 current->text[num]=strdup(line2);
714 current->lines++;
715
716 return current;
717 }
718
719 static void sub_pp_ssa(subtitle *sub) {
720 int l=sub->lines;
721 char *so,*de,*start;
722
723 while (l){
724 /* eliminate any text enclosed with {}, they are font and color settings */
725 so=de=sub->text[--l];
726 while (*so) {
727 if(*so == '{' && so[1]=='\\') {
728 for (start=so; *so && *so!='}'; so++);
729 if(*so) so++; else so=start;
730 }
731 if(*so) {
732 *de=*so;
733 so++; de++;
734 }
735 }
736 *de=*so;
737 }
738 }
739
740 /*
741 * PJS subtitles reader.
742 * That's the "Phoenix Japanimation Society" format.
743 * I found some of them in http://www.scriptsclub.org/ (used for anime).
744 * The time is in tenths of second.
745 *
746 * by set, based on code by szabi (dunnowhat sub format ;-)
747 */
748 static subtitle *sub_read_line_pjs(stream_t *st,subtitle *current,
749 struct readline_args *args)
750 {
751 int utf16 = args->utf16;
752 char line[LINE_LEN+1];
753 char text[LINE_LEN+1], *s, *d;
754
755 if (!stream_read_line (st, line, LINE_LEN, utf16))
756 return NULL;
757 /* skip spaces */
758 for (s=line; *s && isspace(*s); s++);
759 /* allow empty lines at the end of the file */
760 if (*s==0)
761 return NULL;
762 /* get the time */
763 if (sscanf (s, "%ld,%ld,", &(current->start),
764 &(current->end)) <2) {
765 return ERR;
766 }
767 /* the files I have are in tenths of second */
768 current->start *= 10;
769 current->end *= 10;
770 /* walk to the beggining of the string */
771 for (; *s; s++) if (*s==',') break;
772 if (*s) {
773 for (s++; *s; s++) if (*s==',') break;
774 if (*s) s++;
775 }
776 if (*s!='"') {
777 return ERR;
778 }
779 /* copy the string to the text buffer */
780 for (s++, d=text; *s && *s!='"'; s++, d++)
781 *d=*s;
782 *d=0;
783 current->text[0] = strdup(text);
784 current->lines = 1;
785
786 return current;
787 }
788
789 static subtitle *sub_read_line_mpsub(stream_t *st, subtitle *current,
790 struct readline_args *args)
791 {
792 int utf16 = args->utf16;
793 char line[LINE_LEN+1];
794 float a,b;
795 int num=0;
796 char *p, *q;
797
798 do
799 {
800 if (!stream_read_line(st, line, LINE_LEN, utf16)) return NULL;
801 } while (sscanf (line, "%f %f", &a, &b) !=2);
802
803 mpsub_position += a*mpsub_multiplier;
804 current->start=(int) mpsub_position;
805 mpsub_position += b*mpsub_multiplier;
806 current->end=(int) mpsub_position;
807
808 while (num < SUB_MAX_TEXT) {
809 if (!stream_read_line (st, line, LINE_LEN, utf16)) {
810 if (num == 0) return NULL;
811 else return current;
812 }
813 p=line;
814 while (isspace(*p)) p++;
815 if (eol(*p) && num > 0) return current;
816 if (eol(*p)) return NULL;
817
818 for (q=p; !eol(*q); q++);
819 *q='\0';
820 if (strlen(p)) {
821 current->text[num]=strdup(p);
822 // printf (">%s<\n",p);
823 current->lines = ++num;
824 } else {
825 if (num) return current;
826 else return NULL;
827 }
828 }
829 return NULL; // we should have returned before if it's OK
830 }
831
832 #ifndef CONFIG_SORTSUB
833 //we don't need this if we use previous_sub_end
834 subtitle *previous_aqt_sub = NULL;
835 #endif
836
837 static subtitle *sub_read_line_aqt(stream_t *st,subtitle *current,
838 struct readline_args *args)
839 {
840 int utf16 = args->utf16;
841 char line[LINE_LEN+1];
842 char *next;
843 int i;
844
845 while (1) {
846 // try to locate next subtitle
847 if (!stream_read_line (st, line, LINE_LEN, utf16))
848 return NULL;
849 if (!(sscanf (line, "-->> %ld", &(current->start)) <1))
850 break;
851 }
852
853 #ifdef CONFIG_SORTSUB
854 previous_sub_end = (current->start) ? current->start - 1 : 0;
855 #else
856 if (previous_aqt_sub != NULL)
857 previous_aqt_sub->end = current->start-1;
858
859 previous_aqt_sub = current;
860 #endif
861
862 if (!stream_read_line (st, line, LINE_LEN, utf16))
863 return NULL;
864
865 sub_readtext((char *) &line,&current->text[0]);
866 current->lines = 1;
867 current->end = current->start; // will be corrected by next subtitle
868
869 if (!stream_read_line (st, line, LINE_LEN, utf16))
870 return current;
871
872 next = line,i=1;
873 while ((next =sub_readtext (next, &(current->text[i])))) {
874 if (current->text[i]==ERR) {return ERR;}
875 i++;
876 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
877 }
878 current->lines=i+1;
879
880 if (!strlen(current->text[0]) && !strlen(current->text[1])) {
881 #ifdef CONFIG_SORTSUB
882 previous_sub_end = 0;
883 #else
884 // void subtitle -> end of previous marked and exit
885 previous_aqt_sub = NULL;
886 #endif
887 return NULL;
888 }
889
890 return current;
891 }
892
893 #ifndef CONFIG_SORTSUB
894 subtitle *previous_subrip09_sub = NULL;
895 #endif
896
897 static subtitle *sub_read_line_subrip09(stream_t *st,subtitle *current,
898 struct readline_args *args)
899 {
900 int utf16 = args->utf16;
901 char line[LINE_LEN+1];
902 int a1,a2,a3;
903 char * next=NULL;
904 int i,len;
905
906 while (1) {
907 // try to locate next subtitle
908 if (!stream_read_line (st, line, LINE_LEN, utf16))
909 return NULL;
910 if (!((len=sscanf (line, "[%d:%d:%d]",&a1,&a2,&a3)) < 3))
911 break;
912 }
913
914 current->start = a1*360000+a2*6000+a3*100;
915
916 #ifdef CONFIG_SORTSUB
917 previous_sub_end = (current->start) ? current->start - 1 : 0;
918 #else
919 if (previous_subrip09_sub != NULL)
920 previous_subrip09_sub->end = current->start-1;
921
922 previous_subrip09_sub = current;
923 #endif
924
925 if (!stream_read_line (st, line, LINE_LEN, utf16))
926 return NULL;
927
928 next = line,i=0;
929
930 current->text[0]=""; // just to be sure that string is clear
931
932 while ((next =sub_readtext (next, &(current->text[i])))) {
933 if (current->text[i]==ERR) {return ERR;}
934 i++;
935 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
936 }
937 current->lines=i+1;
938
939 if (!strlen(current->text[0]) && (i==0)) {
940 #ifdef CONFIG_SORTSUB
941 previous_sub_end = 0;
942 #else
943 // void subtitle -> end of previous marked and exit
944 previous_subrip09_sub = NULL;
945 #endif
946 return NULL;
947 }
948
949 return current;
950 }
951
952 static subtitle *sub_read_line_jacosub(stream_t* st, subtitle * current,
953 struct readline_args *args)
954 {
955 int utf16 = args->utf16;
956 char line1[LINE_LEN], line2[LINE_LEN], directive[LINE_LEN], *p, *q;
957 unsigned a1, a2, a3, a4, b1, b2, b3, b4, comment = 0;
958 static unsigned jacoTimeres = 30;
959 static int jacoShift = 0;
960
961 memset(current, 0, sizeof(subtitle));
962 memset(line1, 0, LINE_LEN);
963 memset(line2, 0, LINE_LEN);
964 memset(directive, 0, LINE_LEN);
965 while (!current->text[0]) {
966 if (!stream_read_line(st, line1, LINE_LEN, utf16)) {
967 return NULL;
968 }
969 if (sscanf
970 (line1, "%u:%u:%u.%u %u:%u:%u.%u %[^\n\r]", &a1, &a2, &a3, &a4,
971 &b1, &b2, &b3, &b4, line2) < 9) {
972 if (sscanf(line1, "@%u @%u %[^\n\r]", &a4, &b4, line2) < 3) {
973 if (line1[0] == '#') {
974 int hours = 0, minutes = 0, seconds, delta, inverter =
975 1;
976 unsigned units = jacoShift;
977 switch (toupper(line1[1])) {
978 case 'S':
979 if (isalpha(line1[2])) {
980 delta = 6;
981 } else {
982 delta = 2;
983 }
984 if (sscanf(&line1[delta], "%d", &hours)) {
985 if (hours < 0) {
986 hours *= -1;
987 inverter = -1;
988 }
989 if (sscanf(&line1[delta], "%*d:%d", &minutes)) {
990 if (sscanf
991 (&line1[delta], "%*d:%*d:%d",
992 &seconds)) {
993 sscanf(&line1[delta], "%*d:%*d:%*d.%d",
994 &units);
995 } else {
996 hours = 0;
997 sscanf(&line1[delta], "%d:%d.%d",
998 &minutes, &seconds, &units);
999 minutes *= inverter;
1000 }
1001 } else {
1002 hours = minutes = 0;
1003 sscanf(&line1[delta], "%d.%d", &seconds,
1004 &units);
1005 seconds *= inverter;
1006 }
1007 jacoShift =
1008 ((hours * 3600 + minutes * 60 +
1009 seconds) * jacoTimeres +
1010 units) * inverter;
1011 }
1012 break;
1013 case 'T':
1014 if (isalpha(line1[2])) {
1015 delta = 8;
1016 } else {
1017 delta = 2;
1018 }
1019 sscanf(&line1[delta], "%u", &jacoTimeres);
1020 break;
1021 }
1022 }
1023 continue;
1024 } else {
1025 current->start =
1026 (unsigned long) ((a4 + jacoShift) * 100.0 /
1027 jacoTimeres);
1028 current->end =
1029 (unsigned long) ((b4 + jacoShift) * 100.0 /
1030 jacoTimeres);
1031 }
1032 } else {
1033 current->start =
1034 (unsigned
1035 long) (((a1 * 3600 + a2 * 60 + a3) * jacoTimeres + a4 +
1036 jacoShift) * 100.0 / jacoTimeres);
1037 current->end =
1038 (unsigned
1039 long) (((b1 * 3600 + b2 * 60 + b3) * jacoTimeres + b4 +
1040 jacoShift) * 100.0 / jacoTimeres);
1041 }
1042 current->lines = 0;
1043 p = line2;
1044 while ((*p == ' ') || (*p == '\t')) {
1045 ++p;
1046 }
1047 if (isalpha(*p)||*p == '[') {
1048 int cont, jLength;
1049
1050 if (sscanf(p, "%s %[^\n\r]", directive, line1) < 2)
1051 return (subtitle *) ERR;
1052 jLength = strlen(directive);
1053 for (cont = 0; cont < jLength; ++cont) {
1054 if (isalpha(*(directive + cont)))
1055 *(directive + cont) = toupper(*(directive + cont));
1056 }
1057 if ((strstr(directive, "RDB") != NULL)
1058 || (strstr(directive, "RDC") != NULL)
1059 || (strstr(directive, "RLB") != NULL)
1060 || (strstr(directive, "RLG") != NULL)) {
1061 continue;
1062 }
1063 if (strstr(directive, "JL") != NULL) {
1064 current->alignment = SUB_ALIGNMENT_BOTTOMLEFT;
1065 } else if (strstr(directive, "JR") != NULL) {
1066 current->alignment = SUB_ALIGNMENT_BOTTOMRIGHT;
1067 } else {
1068 current->alignment = SUB_ALIGNMENT_BOTTOMCENTER;
1069 }
1070 strcpy(line2, line1);
1071 p = line2;
1072 }
1073 for (q = line1; (!eol(*p)) && (current->lines < SUB_MAX_TEXT); ++p) {
1074 switch (*p) {
1075 case '{':
1076 comment++;
1077 break;
1078 case '}':
1079 if (comment) {
1080 --comment;
1081 //the next line to get rid of a blank after the comment
1082 if ((*(p + 1)) == ' ')
1083 p++;
1084 }
1085 break;
1086 case '~':
1087 if (!comment) {
1088 *q = ' ';
1089 ++q;
1090 }
1091 break;
1092 case ' ':
1093 case '\t':
1094 if ((*(p + 1) == ' ') || (*(p + 1) == '\t'))
1095 break;
1096 if (!comment) {
1097 *q = ' ';
1098 ++q;
1099 }
1100 break;
1101 case '\\':
1102 if (*(p + 1) == 'n') {
1103 *q = '\0';
1104 q = line1;
1105 current->text[current->lines++] = strdup(line1);
1106 ++p;
1107 break;
1108 }
1109 if ((toupper(*(p + 1)) == 'C')
1110 || (toupper(*(p + 1)) == 'F')) {
1111 ++p,++p;
1112 break;
1113 }
1114 if ((*(p + 1) == 'B') || (*(p + 1) == 'b') || (*(p + 1) == 'D') || //actually this means "insert current date here"
1115 (*(p + 1) == 'I') || (*(p + 1) == 'i') || (*(p + 1) == 'N') || (*(p + 1) == 'T') || //actually this means "insert current time here"
1116 (*(p + 1) == 'U') || (*(p + 1) == 'u')) {
1117 ++p;
1118 break;
1119 }
1120 if ((*(p + 1) == '\\') ||
1121 (*(p + 1) == '~') || (*(p + 1) == '{')) {
1122 ++p;
1123 } else if (eol(*(p + 1))) {
1124 if (!stream_read_line(st, directive, LINE_LEN, utf16))
1125 return NULL;
1126 trail_space(directive);
1127 av_strlcat(line2, directive, LINE_LEN);
1128 break;
1129 }
1130 default:
1131 if (!comment) {
1132 *q = *p;
1133 ++q;
1134 }
1135 } //-- switch
1136 } //-- for
1137 *q = '\0';
1138 current->text[current->lines] = strdup(line1);
1139 } //-- while
1140 current->lines++;
1141 return current;
1142 }
1143
1144 static int sub_autodetect (stream_t* st, int *uses_time, int utf16) {
1145 char line[LINE_LEN+1];
1146 int i,j=0;
1147
1148 while (j < 100) {
1149 j++;
1150 if (!stream_read_line (st, line, LINE_LEN, utf16))
1151 return SUB_INVALID;
1152
1153 if (sscanf (line, "{%d}{%d}", &i, &i)==2)
1154 {*uses_time=0;return SUB_MICRODVD;}
1155 if (sscanf (line, "{%d}{}", &i)==1)
1156 {*uses_time=0;return SUB_MICRODVD;}
1157 if (sscanf (line, "[%d][%d]", &i, &i)==2)
1158 {*uses_time=1;return SUB_MPL2;}
1159 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d", &i, &i, &i, &i, &i, &i, &i, &i)==8)
1160 {*uses_time=1;return SUB_SUBRIP;}
1161 if (sscanf (line, "%d:%d:%d%[,.:]%d --> %d:%d:%d%[,.:]%d", &i, &i, &i, (char *)&i, &i, &i, &i, &i, (char *)&i, &i)==10)
1162 {*uses_time=1;return SUB_SUBVIEWER;}
1163 if (sscanf (line, "{T %d:%d:%d:%d",&i, &i, &i, &i)==4)
1164 {*uses_time=1;return SUB_SUBVIEWER2;}
1165 if (strstr (line, "<SAMI>"))
1166 {*uses_time=1; return SUB_SAMI;}
1167 if (sscanf(line, "%d:%d:%d.%d %d:%d:%d.%d", &i, &i, &i, &i, &i, &i, &i, &i) == 8)
1168 {*uses_time = 1; return SUB_JACOSUB;}
1169 if (sscanf(line, "@%d @%d", &i, &i) == 2)
1170 {*uses_time = 1; return SUB_JACOSUB;}
1171 if (sscanf (line, "%d:%d:%d:", &i, &i, &i )==3)
1172 {*uses_time=1;return SUB_VPLAYER;}
1173 if (sscanf (line, "%d:%d:%d ", &i, &i, &i )==3)
1174 {*uses_time=1;return SUB_VPLAYER;}
1175 if (!strncasecmp(line, "<window", 7))
1176 {*uses_time=1;return SUB_RT;}
1177 if (!memcmp(line, "Dialogue: Marked", 16))
1178 {*uses_time=1; return SUB_SSA;}
1179 if (!memcmp(line, "Dialogue: ", 10))
1180 {*uses_time=1; return SUB_SSA;}
1181 if (sscanf (line, "%d,%d,\"%c", &i, &i, (char *) &i) == 3)
1182 {*uses_time=1;return SUB_PJS;}
1183 if (sscanf (line, "FORMAT=%d", &i) == 1)
1184 {*uses_time=0; return SUB_MPSUB;}
1185 if (!memcmp(line, "FORMAT=TIME", 11))
1186 {*uses_time=1; return SUB_MPSUB;}
1187 if (strstr (line, "-->>"))
1188 {*uses_time=0; return SUB_AQTITLE;}
1189 if (sscanf (line, "[%d:%d:%d]", &i, &i, &i)==3)
1190 {*uses_time=1;return SUB_SUBRIP09;}
1191 }
1192
1193 return SUB_INVALID; // too many bad lines
1194 }
1195
1196 extern int sub_utf8;
1197 int sub_utf8_prev=0;
1198
1199 extern float sub_delay;
1200 extern float sub_fps;
1201
1202 #ifdef CONFIG_ICONV
1203 static iconv_t icdsc = (iconv_t)(-1);
1204
1205 void subcp_open (stream_t *st)
1206 {
1207 char *tocp = "UTF-8";
1208
1209 if (sub_cp){
1210 const char *cp_tmp = sub_cp;
1211 #ifdef CONFIG_ENCA
1212 char enca_lang[3], enca_fallback[100];
1213 if (sscanf(sub_cp, "enca:%2s:%99s", enca_lang, enca_fallback) == 2
1214 || sscanf(sub_cp, "ENCA:%2s:%99s", enca_lang, enca_fallback) == 2) {
1215 if (st && st->flags & MP_STREAM_SEEK ) {
1216 cp_tmp = guess_cp(st, enca_lang, enca_fallback);
1217 } else {
1218 cp_tmp = enca_fallback;
1219 if (st)
1220 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: enca failed, stream must be seekable.\n");
1221 }
1222 }
1223 #endif
1224 if ((icdsc = iconv_open (tocp, cp_tmp)) != (iconv_t)(-1)){
1225 mp_msg(MSGT_SUBREADER,MSGL_V,"SUB: opened iconv descriptor.\n");
1226 sub_utf8 = 2;
1227 } else
1228 mp_msg(MSGT_SUBREADER,MSGL_ERR,"SUB: error opening iconv descriptor.\n");
1229 }
1230 }
1231
1232 void subcp_close (void)
1233 {
1234 if (icdsc != (iconv_t)(-1)){
1235 (void) iconv_close (icdsc);
1236 icdsc = (iconv_t)(-1);
1237 mp_msg(MSGT_SUBREADER,MSGL_V,"SUB: closed iconv descriptor.\n");
1238 }
1239 }
1240
1241 subtitle* subcp_recode (subtitle *sub)
1242 {
1243 int l=sub->lines;
1244 size_t ileft, oleft;
1245 char *op, *ip, *ot;
1246 if(icdsc == (iconv_t)(-1)) return sub;
1247
1248 while (l){
1249 ip = sub->text[--l];
1250 ileft = strlen(ip);
1251 oleft = 4 * ileft;
1252
1253 if (!(ot = malloc(oleft + 1))){
1254 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: error allocating mem.\n");
1255 continue;
1256 }
1257 op = ot;
1258 if (iconv(icdsc, &ip, &ileft,
1259 &op, &oleft) == (size_t)(-1)) {
1260 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: error recoding line.\n");
1261 free(ot);
1262 continue;
1263 }
1264 // In some stateful encodings, we must clear the state to handle the last character
1265 if (iconv(icdsc, NULL, NULL,
1266 &op, &oleft) == (size_t)(-1)) {
1267 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: error recoding line, can't clear encoding state.\n");
1268 }
1269 *op='\0' ;
1270 free (sub->text[l]);
1271 sub->text[l] = ot;
1272 }
1273 return sub;
1274 }
1275 #endif
1276
1277 #ifdef CONFIG_FRIBIDI
1278 /**
1279 * Do conversion necessary for right-to-left language support via fribidi.
1280 * @param sub subtitle to convert
1281 * @param sub_utf8 whether the subtitle is encoded in UTF-8
1282 * @param from first new subtitle, all lines before this are assumed to be already converted
1283 */
1284 static subtitle* sub_fribidi (subtitle *sub, int sub_utf8, int from)
1285 {
1286 FriBidiChar logical[LINE_LEN+1], visual[LINE_LEN+1]; // Hopefully these two won't smash the stack
1287 char *ip = NULL, *op = NULL;
1288 size_t len,orig_len;
1289 int l=sub->lines;
1290 int char_set_num;
1291 fribidi_boolean log2vis;
1292 if (!flip_hebrew)
1293 return sub;
1294 fribidi_set_mirroring(1);
1295 fribidi_set_reorder_nsm(0);
1296
1297 if( sub_utf8 == 0 ) {
1298 char_set_num = fribidi_parse_charset (fribidi_charset?fribidi_charset:"ISO8859-8");
1299 }else {
1300 char_set_num = fribidi_parse_charset ("UTF-8");
1301 }
1302 while (l > from) {
1303 ip = sub->text[--l];
1304 orig_len = len = strlen( ip ); // We assume that we don't use full unicode, only UTF-8 or ISO8859-x
1305 if(len > LINE_LEN) {
1306 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: sub->text is longer than LINE_LEN.\n");
1307 l++;
1308 break;
1309 }
1310 len = fribidi_charset_to_unicode (char_set_num, ip, len, logical);
1311 #if FRIBIDI_INTERFACE_VERSION < 3
1312 FriBidiCharType base = fribidi_flip_commas?FRIBIDI_TYPE_ON:FRIBIDI_TYPE_L;
1313 #else
1314 FriBidiParType base = fribidi_flip_commas?FRIBIDI_TYPE_ON:FRIBIDI_TYPE_L;
1315 #endif
1316 log2vis = fribidi_log2vis (logical, len, &base,
1317 /* output */
1318 visual, NULL, NULL, NULL);
1319 if(log2vis) {
1320 len = fribidi_remove_bidi_marks (visual, len, NULL, NULL,
1321 NULL);
1322 if((op = malloc((FFMAX(2*orig_len,2*len) + 1))) == NULL) {
1323 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: error allocating mem.\n");
1324 l++;
1325 break;
1326 }
1327 fribidi_unicode_to_charset ( char_set_num, visual, len,op);
1328 free (ip);
1329 sub->text[l] = op;
1330 }
1331 }
1332 if (!from && l){
1333 for (l = sub->lines; l;)
1334 free (sub->text[--l]);
1335 return ERR;
1336 }
1337 return sub;
1338 }
1339
1340 #endif
1341
1342 static void adjust_subs_time(subtitle* sub, float subtime, float fps, int block,
1343 int sub_num, int sub_uses_time) {
1344 int n,m;
1345 subtitle* nextsub;
1346 int i = sub_num;
1347 unsigned long subfms = (sub_uses_time ? 100 : fps) * subtime;
1348 unsigned long overlap = (sub_uses_time ? 100 : fps) / 5; // 0.2s
1349
1350 n=m=0;
1351 if (i) for (;;){
1352 if (sub->end <= sub->start){
1353 sub->end = sub->start + subfms;
1354 m++;
1355 n++;
1356 }
1357 if (!--i) break;
1358 nextsub = sub + 1;
1359 if(block){
1360 if ((sub->end > nextsub->start) && (sub->end <= nextsub->start + overlap)) {
1361 // these subtitles overlap for less than 0.2 seconds
1362 // and would result in very short overlapping subtitle
1363 // so let's fix the problem here, before overlapping code
1364 // get its hands on them
1365 unsigned delta = sub->end - nextsub->start, half = delta / 2;
1366 sub->end -= half + 1;
1367 nextsub->start += delta - half;
1368 }
1369 if (sub->end >= nextsub->start){
1370 sub->end = nextsub->start - 1;
1371 if (sub->end - sub->start > subfms)
1372 sub->end = sub->start + subfms;
1373 if (!m)
1374 n++;
1375 }
1376 }
1377
1378 /* Theory:
1379 * Movies are often converted from FILM (24 fps)
1380 * to PAL (25) by simply speeding it up, so we
1381 * to multiply the original timestmaps by
1382 * (Movie's FPS / Subtitle's (guessed) FPS)
1383 * so eg. for 23.98 fps movie and PAL time based
1384 * subtitles we say -subfps 25 and we're fine!
1385 */
1386
1387 /* timed sub fps correction ::atmos */
1388 /* the frame-based case is handled in mpcommon.c
1389 * where find_sub is called */
1390 if(sub_uses_time && sub_fps) {
1391 sub->start *= sub_fps/fps;
1392 sub->end *= sub_fps/fps;
1393 }
1394
1395 sub = nextsub;
1396 m = 0;
1397 }
1398 if (n) mp_msg(MSGT_SUBREADER,MSGL_V,"SUB: Adjusted %d subtitle(s).\n", n);
1399 }
1400
1401 struct subreader {
1402 subtitle * (*read)(stream_t *st, subtitle *dest,
1403 struct readline_args *args);
1404 void (*post)(subtitle *dest);
1405 const char *name;
1406 };
1407
1408 #ifdef CONFIG_ENCA
1409 const char* guess_buffer_cp(unsigned char* buffer, int buflen, const char *preferred_language, const char *fallback)
1410 {
1411 const char **languages;
1412 size_t langcnt;
1413 EncaAnalyser analyser;
1414 EncaEncoding encoding;
1415 const char *detected_sub_cp = NULL;
1416 int i;
1417
1418 languages = enca_get_languages(&langcnt);
1419 mp_msg(MSGT_SUBREADER, MSGL_V, "ENCA supported languages: ");
1420 for (i = 0; i < langcnt; i++) {
1421 mp_msg(MSGT_SUBREADER, MSGL_V, "%s ", languages[i]);
1422 }
1423 mp_msg(MSGT_SUBREADER, MSGL_V, "\n");
1424
1425 for (i = 0; i < langcnt; i++) {
1426 if (strcasecmp(languages[i], preferred_language) != 0) continue;
1427 analyser = enca_analyser_alloc(languages[i]);
1428 encoding = enca_analyse_const(analyser, buffer, buflen);
1429 enca_analyser_free(analyser);
1430 if (encoding.charset != ENCA_CS_UNKNOWN) {
1431 detected_sub_cp = enca_charset_name(encoding.charset, ENCA_NAME_STYLE_ICONV);
1432 break;
1433 }
1434 }
1435
1436 free(languages);
1437
1438 if (!detected_sub_cp) {
1439 detected_sub_cp = fallback;
1440 mp_msg(MSGT_SUBREADER, MSGL_INFO, "ENCA detection failed: fallback to %s\n", fallback);
1441 }else{
1442 mp_msg(MSGT_SUBREADER, MSGL_INFO, "ENCA detected charset: %s\n", detected_sub_cp);
1443 }
1444
1445 return detected_sub_cp;
1446 }
1447
1448 #define MAX_GUESS_BUFFER_SIZE (256*1024)
1449 const char* guess_cp(stream_t *st, const char *preferred_language, const char *fallback)
1450 {
1451 size_t buflen;
1452 unsigned char *buffer;
1453 const char *detected_sub_cp = NULL;
1454
1455 buffer = malloc(MAX_GUESS_BUFFER_SIZE);
1456 buflen = stream_read(st,buffer, MAX_GUESS_BUFFER_SIZE);
1457
1458 detected_sub_cp = guess_buffer_cp(buffer, buflen, preferred_language, fallback);
1459
1460 free(buffer);
1461 stream_reset(st);
1462 stream_seek(st,0);
1463
1464 return detected_sub_cp;
1465 }
1466 #undef MAX_GUESS_BUFFER_SIZE
1467 #endif
1468
1469 sub_data* sub_read_file(char *filename, float fps, struct MPOpts *opts)
1470 {
1471 int utf16;
1472 stream_t* fd;
1473 int n_max, n_first, i, j, sub_first, sub_orig;
1474 subtitle *first, *second, *sub, *return_sub, *alloced_sub = NULL;
1475 sub_data *subt_data;
1476 int uses_time = 0, sub_num = 0, sub_errs = 0;
1477 static const struct subreader sr[]=
1478 {
1479 { sub_read_line_microdvd, NULL, "microdvd" },
1480 { sub_read_line_subrip, NULL, "subrip" },
1481 { sub_read_line_subviewer, NULL, "subviewer" },
1482 { sub_read_line_sami, NULL, "sami" },
1483 { sub_read_line_vplayer, NULL, "vplayer" },
1484 { sub_read_line_rt, NULL, "rt" },
1485 { sub_read_line_ssa, sub_pp_ssa, "ssa" },
1486 { sub_read_line_pjs, NULL, "pjs" },
1487 { sub_read_line_mpsub, NULL, "mpsub" },
1488 { sub_read_line_aqt, NULL, "aqt" },
1489 { sub_read_line_subviewer2, NULL, "subviewer 2.0" },
1490 { sub_read_line_subrip09, NULL, "subrip 0.9" },
1491 { sub_read_line_jacosub, NULL, "jacosub" },
1492 { sub_read_line_mpl2, NULL, "mpl2" }
1493 };
1494 const struct subreader *srp;
1495
1496 if(filename==NULL) return NULL; //qnx segfault
1497 fd=open_stream (filename, NULL, NULL); if (!fd) return NULL;
1498
1499 sub_format = SUB_INVALID;
1500 for (utf16 = 0; sub_format == SUB_INVALID && utf16 < 3; utf16++) {
1501 sub_format=sub_autodetect (fd, &uses_time, utf16);
1502 stream_reset(fd);
1503 stream_seek(fd,0);
1504 }
1505 utf16--;
1506
1507 mpsub_multiplier = (uses_time ? 100.0 : 1.0);
1508 if (sub_format==SUB_INVALID) {mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: Could not determine file format\n");return NULL;}
1509 srp=sr+sub_format;
1510 mp_msg(MSGT_SUBREADER, MSGL_V, "SUB: Detected subtitle file format: %s\n", srp->name);
1511
1512 #ifdef CONFIG_ICONV
1513 sub_utf8_prev=sub_utf8;
1514 {
1515 int l,k;
1516 k = -1;
1517 if ((l=strlen(filename))>4){
1518 char *exts[] = {".utf", ".utf8", ".utf-8" };
1519 for (k=3;--k>=0;)
1520 if (l >= strlen(exts[k]) && !strcasecmp(filename+(l - strlen(exts[k])), exts[k])){
1521 sub_utf8 = 1;
1522 break;
1523 }
1524 }
1525 if (k<0) subcp_open(fd);
1526 }
1527 #endif
1528
1529 sub_num=0;n_max=32;
1530 first=malloc(n_max*sizeof(subtitle));
1531 if(!first){
1532 #ifdef CONFIG_ICONV
1533 subcp_close();
1534 sub_utf8=sub_utf8_prev;
1535 #endif
1536 return NULL;
1537 }
1538
1539 #ifdef CONFIG_SORTSUB
1540 alloced_sub =
1541 sub = malloc(sizeof(subtitle));
1542 //This is to deal with those formats (AQT & Subrip) which define the end of a subtitle
1543 //as the beginning of the following
1544 previous_sub_end = 0;
1545 #endif
1546 while(1){
1547 if(sub_num>=n_max){
1548 n_max+=16;
1549 first=realloc(first,n_max*sizeof(subtitle));
1550 }
1551 #ifndef CONFIG_SORTSUB
1552 sub = &first[sub_num];
1553 #endif
1554 memset(sub, '\0', sizeof(subtitle));
1555 sub=srp->read(fd, sub, &(struct readline_args){utf16, opts});
1556 if(!sub) break; // EOF
1557 #ifdef CONFIG_ICONV
1558 if ((sub!=ERR) && sub_utf8 == 2) sub=subcp_recode(sub);
1559 #endif
1560 #ifdef CONFIG_FRIBIDI
1561 if (sub!=ERR) sub=sub_fribidi(sub,sub_utf8,0);
1562 #endif
1563 if ( sub == ERR )
1564 {
1565 #ifdef CONFIG_ICONV
1566 subcp_close();
1567 #endif
1568 free(first);
1569 free(alloced_sub);
1570 return NULL;
1571 }
1572 // Apply any post processing that needs recoding first
1573 if ((sub!=ERR) && !sub_no_text_pp && srp->post) srp->post(sub);
1574 #ifdef CONFIG_SORTSUB
1575 if(!sub_num || (first[sub_num - 1].start <= sub->start)){
1576 first[sub_num].start = sub->start;
1577 first[sub_num].end = sub->end;
1578 first[sub_num].lines = sub->lines;
1579 first[sub_num].alignment = sub->alignment;
1580 for(i = 0; i < sub->lines; ++i){
1581 first[sub_num].text[i] = sub->text[i];
1582 }
1583 if (previous_sub_end){
1584 first[sub_num - 1].end = previous_sub_end;
1585 previous_sub_end = 0;
1586 }
1587 } else {
1588 for(j = sub_num - 1; j >= 0; --j){
1589 first[j + 1].start = first[j].start;
1590 first[j + 1].end = first[j].end;
1591 first[j + 1].lines = first[j].lines;
1592 first[j + 1].alignment = first[j].alignment;
1593 for(i = 0; i < first[j].lines; ++i){
1594 first[j + 1].text[i] = first[j].text[i];
1595 }
1596 if(!j || (first[j - 1].start <= sub->start)){
1597 first[j].start = sub->start;
1598 first[j].end = sub->end;
1599 first[j].lines = sub->lines;
1600 first[j].alignment = sub->alignment;
1601 for(i = 0; i < SUB_MAX_TEXT; ++i){
1602 first[j].text[i] = sub->text[i];
1603 }
1604 if (previous_sub_end){
1605 first[j].end = first[j - 1].end;
1606 first[j - 1].end = previous_sub_end;
1607 previous_sub_end = 0;
1608 }
1609 break;
1610 }
1611 }
1612 }
1613 #endif
1614 if(sub==ERR) ++sub_errs; else ++sub_num; // Error vs. Valid
1615 }
1616
1617 free_stream(fd);
1618
1619 #ifdef CONFIG_ICONV
1620 subcp_close();
1621 #endif
1622 free(alloced_sub);
1623
1624 // printf ("SUB: Subtitle format %s time.\n", uses_time?"uses":"doesn't use");
1625 mp_msg(MSGT_SUBREADER, MSGL_V,"SUB: Read %i subtitles, %i bad line(s).\n",
1626 sub_num, sub_errs);
1627
1628 if(sub_num<=0){
1629 free(first);
1630 return NULL;
1631 }
1632
1633 // we do overlap if the user forced it (suboverlap_enable == 2) or
1634 // the user didn't forced no-overlapsub and the format is Jacosub or Ssa.
1635 // this is because usually overlapping subtitles are found in these formats,
1636 // while in others they are probably result of bad timing
1637 if ((suboverlap_enabled == 2) ||
1638 ((suboverlap_enabled) && ((sub_format == SUB_JACOSUB) || (sub_format == SUB_SSA)))) {
1639 adjust_subs_time(first, 6.0, fps, 0, sub_num, uses_time);/*~6 secs AST*/
1640 // here we manage overlapping subtitles
1641 sub_orig = sub_num;
1642 n_first = sub_num;
1643 sub_num = 0;
1644 second = NULL;
1645 // for each subtitle in first[] we deal with its 'block' of
1646 // bonded subtitles
1647 for (sub_first = 0; sub_first < n_first; ++sub_first) {
1648 unsigned long global_start = first[sub_first].start,
1649 global_end = first[sub_first].end, local_start, local_end;
1650 int lines_to_add = first[sub_first].lines, sub_to_add = 0,
1651 **placeholder = NULL, higher_line = 0, counter, start_block_sub = sub_num;
1652 char real_block = 1;
1653
1654 // here we find the number of subtitles inside the 'block'
1655 // and its span interval. this works well only with sorted
1656 // subtitles
1657 while ((sub_first + sub_to_add + 1 < n_first) && (first[sub_first + sub_to_add + 1].start < global_end)) {
1658 ++sub_to_add;
1659 lines_to_add += first[sub_first + sub_to_add].lines;
1660 if (first[sub_first + sub_to_add].start < global_start) {
1661 global_start = first[sub_first + sub_to_add].start;
1662 }
1663 if (first[sub_first + sub_to_add].end > global_end) {
1664 global_end = first[sub_first + sub_to_add].end;
1665 }
1666 }
1667
1668 /* Avoid n^2 memory use for the "placeholder" data structure
1669 * below with subtitles that have a huge number of
1670 * consecutive overlapping lines. */
1671 lines_to_add = FFMIN(lines_to_add, SUB_MAX_TEXT);
1672
1673 // we need a structure to keep trace of the screen lines
1674 // used by the subs, a 'placeholder'
1675 counter = 2 * sub_to_add + 1; // the maximum number of subs derived
1676 // from a block of sub_to_add+1 subs
1677 placeholder = malloc(sizeof(int *) * counter);
1678 for (i = 0; i < counter; ++i) {
1679 placeholder[i] = malloc(sizeof(int) * lines_to_add);
1680 for (j = 0; j < lines_to_add; ++j) {
1681 placeholder[i][j] = -1;
1682 }
1683 }
1684
1685 counter = 0;
1686 local_end = global_start - 1;
1687 do {
1688 int ls;
1689
1690 // here we find the beginning and the end of a new
1691 // subtitle in the block
1692 local_start = local_end + 1;
1693 local_end = global_end;
1694 for (j = 0; j <= sub_to_add; ++j) {
1695 if ((first[sub_first + j].start - 1 > local_start) && (first[sub_first + j].start - 1 < local_end)) {
1696 local_end = first[sub_first + j].start - 1;
1697 } else if ((first[sub_first + j].end > local_start) && (first[sub_first + j].end < local_end)) {
1698 local_end = first[sub_first + j].end;
1699 }
1700 }
1701 // here we allocate the screen lines to subs we must
1702 // display in current local_start-local_end interval.
1703 // if the subs were yet presents in the previous interval
1704 // they keep the same lines, otherside they get unused lines
1705 for (j = 0; j <= sub_to_add; ++j) {
1706 if ((first[sub_first + j].start <= local_end) && (first[sub_first + j].end > local_start)) {
1707 unsigned long sub_lines = first[sub_first + j].lines, fragment_length = lines_to_add + 1,
1708 tmp = 0;
1709 char boolean = 0;
1710 int fragment_position = -1;
1711
1712 // if this is not the first new sub of the block
1713 // we find if this sub was present in the previous
1714 // new sub
1715 if (counter)
1716 for (i = 0; i < lines_to_add; ++i) {
1717 if (placeholder[counter - 1][i] == sub_first + j) {
1718 placeholder[counter][i] = sub_first + j;
1719 boolean = 1;
1720 }
1721 }
1722 if (boolean)
1723 continue;
1724
1725 // we are looking for the shortest among all groups of
1726 // sequential blank lines whose length is greater than or
1727 // equal to sub_lines. we store in fragment_position the
1728 // position of the shortest group, in fragment_length its
1729 // length, and in tmp the length of the group currently
1730 // examinated
1731 for (i = 0; i < lines_to_add; ++i) {
1732 if (placeholder[counter][i] == -1) {
1733 // placeholder[counter][i] is part of the current group
1734 // of blank lines
1735 ++tmp;
1736 } else {
1737 if (tmp == sub_lines) {
1738 // current group's size fits exactly the one we
1739 // need, so we stop looking
1740 fragment_position = i - tmp;
1741 tmp = 0;
1742 break;
1743 }
1744 if ((tmp) && (tmp > sub_lines) && (tmp < fragment_length)) {
1745 // current group is the best we found till here,
1746 // but is still bigger than the one we are looking
1747 // for, so we keep on looking
1748 fragment_length = tmp;
1749 fragment_position = i - tmp;
1750 tmp = 0;
1751 } else {
1752 // current group doesn't fit at all, so we forget it
1753 tmp = 0;
1754 }
1755 }
1756 }
1757 if (tmp) {
1758 // last screen line is blank, a group ends with it
1759 if ((tmp >= sub_lines) && (tmp < fragment_length)) {
1760 fragment_position = i - tmp;
1761 }
1762 }
1763 if (fragment_position == -1) {
1764 // it was not possible to find free screen line(s) for a subtitle,
1765 // usually this means a bug in the code; however we do not overlap
1766 mp_msg(MSGT_SUBREADER, MSGL_WARN, "SUB: we could not find a suitable position for an overlapping subtitle\n");
1767 higher_line = SUB_MAX_TEXT + 1;
1768 break;
1769 } else {
1770 for (tmp = 0; tmp < sub_lines; ++tmp) {
1771 placeholder[counter][fragment_position + tmp] = sub_first + j;
1772 }
1773 }
1774 }
1775 }
1776 for (j = higher_line + 1; j < lines_to_add; ++j) {
1777 if (placeholder[counter][j] != -1)
1778 higher_line = j;
1779 else
1780 break;
1781 }
1782 if (higher_line >= SUB_MAX_TEXT) {
1783 // the 'block' has too much lines, so we don't overlap the
1784 // subtitles
1785 second = realloc(second, (sub_num + sub_to_add + 1) * sizeof(subtitle));
1786 for (j = 0; j <= sub_to_add; ++j) {
1787 int ls;
1788 memset(&second[sub_num + j], '\0', sizeof(subtitle));
1789 second[sub_num + j].start = first[sub_first + j].start;
1790 second[sub_num + j].end = first[sub_first + j].end;
1791 second[sub_num + j].lines = first[sub_first + j].lines;
1792 second[sub_num + j].alignment = first[sub_first + j].alignment;
1793 for (ls = 0; ls < second[sub_num + j].lines; ls++) {
1794 second[sub_num + j].text[ls] = strdup(first[sub_first + j].text[ls]);
1795 }
1796 }
1797 sub_num += sub_to_add + 1;
1798 sub_first += sub_to_add;
1799 real_block = 0;
1800 break;
1801 }
1802
1803 // we read the placeholder structure and create the new
1804 // subs.
1805 second = realloc(second, (sub_num + 1) * sizeof(subtitle));
1806 memset(&second[sub_num], '\0', sizeof(subtitle));
1807 second[sub_num].start = local_start;
1808 second[sub_num].end = local_end;
1809 second[sub_num].alignment = first[sub_first].alignment;
1810 n_max = (lines_to_add < SUB_MAX_TEXT) ? lines_to_add : SUB_MAX_TEXT;
1811 for (i = 0, j = 0; j < n_max; ++j) {
1812 if (placeholder[counter][j] != -1) {
1813 int lines = first[placeholder[counter][j]].lines;
1814 for (ls = 0; ls < lines; ++ls) {
1815 second[sub_num].text[i++] = strdup(first[placeholder[counter][j]].text[ls]);
1816 }
1817 j += lines - 1;
1818 } else {
1819 second[sub_num].text[i++] = strdup(" ");
1820 }
1821 }
1822 ++sub_num;
1823 ++counter;
1824 } while (local_end < global_end);
1825 if (real_block)
1826 for (i = 0; i < counter; ++i)
1827 second[start_block_sub + i].lines = higher_line + 1;
1828
1829 counter = 2 * sub_to_add + 1;
1830 for (i = 0; i < counter; ++i) {
1831 free(placeholder[i]);
1832 }
1833 free(placeholder);
1834 sub_first += sub_to_add;
1835 }
1836
1837 for (j = sub_orig - 1; j >= 0; --j) {
1838 for (i = first[j].lines - 1; i >= 0; --i) {
1839 free(first[j].text[i]);
1840 }
1841 }
1842 free(first);
1843
1844 return_sub = second;
1845 } else { //if(suboverlap_enabled)
1846 adjust_subs_time(first, 6.0, fps, 1, sub_num, uses_time);/*~6 secs AST*/
1847 return_sub = first;
1848 }
1849 if (return_sub == NULL) return NULL;
1850 subt_data = malloc(sizeof(sub_data));
1851 subt_data->filename = strdup(filename);
1852 subt_data->sub_uses_time = uses_time;
1853 subt_data->sub_num = sub_num;
1854 subt_data->sub_errs = sub_errs;
1855 subt_data->subtitles = return_sub;
1856 return subt_data;
1857 }
1858
1859 #if 0
1860 char * strreplace( char * in,char * what,char * whereof )
1861 {
1862 int i;
1863 char * tmp;
1864
1865 if ( ( in == NULL )||( what == NULL )||( whereof == NULL )||( ( tmp=strstr( in,what ) ) == NULL ) ) return NULL;
1866 for( i=0;i<strlen( whereof );i++ ) tmp[i]=whereof[i];
1867 if ( strlen( what ) > strlen( whereof ) ) tmp[i]=0;
1868 return in;
1869 }
1870 #endif
1871
1872
1873 static void strcpy_trim(char *d, char *s)
1874 {
1875 // skip leading whitespace
1876 while (*s && isspace(*s)) {
1877 s++;
1878 }
1879 for (;;) {
1880 // copy word
1881 while (*s && !isspace(*s)) {
1882 *d = tolower(*s);
1883 s++; d++;
1884 }
1885 if (*s == 0) break;
1886 // trim excess whitespace
1887 while (*s && isspace(*s)) {
1888 s++;
1889 }
1890 if (*s == 0) break;
1891 *d++ = ' ';
1892 }
1893 *d = 0;
1894 }
1895
1896 static void strcpy_strip_ext(char *d, char *s)
1897 {
1898 char *tmp = strrchr(s,'.');
1899 if (!tmp) {
1900 strcpy(d, s);
1901 return;
1902 } else {
1903 strncpy(d, s, tmp-s);
1904 d[tmp-s] = 0;
1905 }
1906 while (*d) {
1907 *d = tolower(*d);
1908 d++;
1909 }
1910 }
1911
1912 static void strcpy_get_ext(char *d, char *s)
1913 {
1914 char *tmp = strrchr(s,'.');
1915 if (!tmp) {
1916 strcpy(d, "");
1917 return;
1918 } else {
1919 strcpy(d, tmp+1);
1920 }
1921 }
1922
1923 static int whiteonly(char *s)
1924 {
1925 while (*s) {
1926 if (!isspace(*s)) return 0;
1927 s++;
1928 }
1929 return 1;
1930 }
1931
1932 typedef struct subfn
1933 {
1934 int priority;
1935 char *fname;
1936 } subfn;
1937
1938 static int compare_sub_priority(const void *a, const void *b)
1939 {
1940 if (((const subfn*)a)->priority > ((const subfn*)b)->priority) {
1941 return -1;
1942 } else if (((const subfn*)a)->priority < ((const subfn*)b)->priority) {
1943 return 1;
1944 } else {
1945 return strcoll(((const subfn*)a)->fname, ((const subfn*)b)->fname);
1946 }
1947 }
1948
1949 char** sub_filenames(const char* path, char *fname)
1950 {
1951 char *f_dir, *f_fname, *f_fname_noext, *f_fname_trim, *tmp, *tmp_sub_id;
1952 char *tmp_fname_noext, *tmp_fname_trim, *tmp_fname_ext, *tmpresult;
1953
1954 int len, pos, found, i, j;
1955 char * sub_exts[] = { "utf", "utf8", "utf-8", "sub", "srt", "smi", "rt", "txt", "ssa", "aqt", "jss", "js", "ass", NULL};
1956 subfn *result;
1957 char **result2;
1958
1959 int subcnt;
1960
1961 FILE *f;
1962
1963 DIR *d;
1964 struct dirent *de;
1965
1966 len = (strlen(fname) > 256 ? strlen(fname) : 256)
1967 +(strlen(path) > 256 ? strlen(path) : 256)+2;
1968
1969 f_dir = malloc(len);
1970 f_fname = malloc(len);
1971 f_fname_noext = malloc(len);
1972 f_fname_trim = malloc(len);
1973
1974 tmp_fname_noext = malloc(len);
1975 tmp_fname_trim = malloc(len);
1976 tmp_fname_ext = malloc(len);
1977
1978 tmpresult = malloc(len);
1979
1980 result = calloc(MAX_SUBTITLE_FILES, sizeof(*result));
1981
1982 subcnt = 0;
1983
1984 tmp = strrchr(fname,'/');
1985 #if HAVE_DOS_PATHS
1986 if(!tmp)tmp = strrchr(fname,'\\');
1987 if(!tmp)tmp = strrchr(fname,':');
1988 #endif
1989
1990 // extract filename & dirname from fname
1991 if (tmp) {
1992 strcpy(f_fname, tmp+1);
1993 pos = tmp - fname;
1994 strncpy(f_dir, fname, pos+1);
1995 f_dir[pos+1] = 0;
1996 } else {
1997 strcpy(f_fname, fname);
1998 strcpy(f_dir, "./");
1999 }
2000
2001 strcpy_strip_ext(f_fname_noext, f_fname);
2002 strcpy_trim(f_fname_trim, f_fname_noext);
2003
2004 /* The code using sub language here is broken - it assumes strict
2005 * "videoname languagename" syntax for the subtitle file, which is
2006 * very unlikely to match especially if language name uses "en,de"
2007 * syntax... */
2008 tmp_sub_id = NULL;
2009 #if 0
2010 if (dvdsub_lang && !whiteonly(dvdsub_lang)) {
2011 tmp_sub_id = malloc(strlen(dvdsub_lang)+1);
2012 strcpy_trim(tmp_sub_id, dvdsub_lang);
2013 }
2014 #endif
2015
2016 // 0 = nothing
2017 // 1 = any subtitle file
2018 // 2 = any sub file containing movie name
2019 // 3 = sub file containing movie name and the lang extension
2020 for (j = 0; j <= 1; j++) {
2021 d = opendir(j == 0 ? f_dir : path);
2022 if (d) {
2023 while ((de = readdir(d))) {
2024 // retrieve various parts of the filename
2025 strcpy_strip_ext(tmp_fname_noext, de->d_name);
2026 strcpy_get_ext(tmp_fname_ext, de->d_name);
2027 strcpy_trim(tmp_fname_trim, tmp_fname_noext);
2028
2029 // does it end with a subtitle extension?
2030 found = 0;
2031 #ifdef CONFIG_ICONV
2032 #ifdef CONFIG_ENCA
2033 for (i = ((sub_cp && strncasecmp(sub_cp, "enca", 4) != 0) ? 3 : 0); sub_exts[i]; i++) {
2034 #else
2035 for (i = (sub_cp ? 3 : 0); sub_exts[i]; i++) {
2036 #endif
2037 #else
2038 for (i = 0; sub_exts[i]; i++) {
2039 #endif
2040 if (strcasecmp(sub_exts[i], tmp_fname_ext) == 0) {
2041 found = 1;
2042 break;
2043 }
2044 }
2045
2046 // we have a (likely) subtitle file
2047 if (found) {
2048 int prio = 0;
2049 if (!prio && tmp_sub_id)
2050 {
2051 sprintf(tmpresult, "%s %s", f_fname_trim, tmp_sub_id);
2052 if (strcmp(tmp_fname_trim, tmpresult) == 0 && sub_match_fuzziness >= 1) {
2053 // matches the movie name + lang extension
2054 prio = 5;
2055 }
2056 }
2057 if (!prio && strcmp(tmp_fname_trim, f_fname_trim) == 0) {
2058 // matches the movie name
2059 prio = 4;
2060 }
2061 if (!prio && (tmp = strstr(tmp_fname_trim, f_fname_trim)) && (sub_match_fuzziness >= 1)) {
2062 // contains the movie name
2063 tmp += strlen(f_fname_trim);
2064 if (tmp_sub_id && strstr(tmp, tmp_sub_id)) {
2065 // with sub_id specified prefer localized subtitles
2066 prio = 3;
2067 } else if ((tmp_sub_id == NULL) && whiteonly(tmp)) {
2068 // without sub_id prefer "plain" name
2069 prio = 3;
2070 } else {
2071 // with no localized subs found, try any else instead
2072 prio = 2;
2073 }
2074 }
2075 if (!prio) {
2076 // doesn't contain the movie name
2077 // don't try in the mplayer subtitle directory
2078 if ((j == 0) && (sub_match_fuzziness >= 2)) {
2079 prio = 1;
2080 }
2081 }
2082
2083 mp_msg(MSGT_SUBREADER, MSGL_DBG2, "Potential sub file: "
2084 "\"%s\" Priority: %d\n", de->d_name, prio);
2085 if (prio) {
2086 prio += prio;
2087 #ifdef CONFIG_ICONV
2088 if (i<3){ // prefer UTF-8 coded
2089 prio++;
2090 }
2091 #endif
2092 sprintf(tmpresult, "%s%s", j == 0 ? f_dir : path, de->d_name);
2093 // fprintf(stderr, "%s priority %d\n", tmpresult, prio);
2094 if ((f = fopen(tmpresult, "rt"))) {
2095 fclose(f);
2096 result[subcnt].priority = prio;
2097 result[subcnt].fname = strdup(tmpresult);
2098 subcnt++;
2099 }
2100 }
2101
2102 }
2103 if (subcnt >= MAX_SUBTITLE_FILES) break;
2104 }
2105 closedir(d);
2106 }
2107
2108 }
2109
2110 free(tmp_sub_id);
2111
2112 free(f_dir);
2113 free(f_fname);
2114 free(f_fname_noext);
2115 free(f_fname_trim);
2116
2117 free(tmp_fname_noext);
2118 free(tmp_fname_trim);
2119 free(tmp_fname_ext);
2120
2121 free(tmpresult);
2122
2123 qsort(result, subcnt, sizeof(subfn), compare_sub_priority);
2124
2125 result2 = calloc(subcnt + 1, sizeof(*result2));
2126
2127 for (i = 0; i < subcnt; i++) {
2128 result2[i] = result[i].fname;
2129 }
2130 result2[subcnt] = NULL;
2131
2132 free(result);
2133
2134 return result2;
2135 }
2136
2137 void list_sub_file(sub_data* subd){
2138 int i,j;
2139 subtitle *subs = subd->subtitles;
2140
2141 for(j=0; j < subd->sub_num; j++){
2142 subtitle* egysub=&subs[j];
2143 mp_msg(MSGT_SUBREADER,MSGL_INFO,"%i line%c (%li-%li)\n",
2144 egysub->lines,
2145 (1==egysub->lines)?' ':'s',
2146 egysub->start,
2147 egysub->end);
2148 for (i=0; i<egysub->lines; i++) {
2149 mp_msg(MSGT_SUBREADER,MSGL_INFO,"\t\t%d: %s%s", i,egysub->text[i], i==egysub->lines-1?"":" \n ");
2150 }
2151 mp_msg(MSGT_SUBREADER,MSGL_INFO,"\n");
2152 }
2153
2154 mp_msg(MSGT_SUBREADER,MSGL_INFO,"Subtitle format %s time.\n",
2155 subd->sub_uses_time ? "uses":"doesn't use");
2156 mp_msg(MSGT_SUBREADER,MSGL_INFO,"Read %i subtitles, %i errors.\n", subd->sub_num, subd->sub_errs);
2157 }
2158
2159 void dump_srt(sub_data* subd, float fps){
2160 int i,j;
2161 int h,m,s,ms;
2162 FILE * fd;
2163 subtitle * onesub;
2164 unsigned long temp;
2165 subtitle *subs = subd->subtitles;
2166
2167 if (!subd->sub_uses_time && sub_fps == 0)
2168 sub_fps = fps;
2169 fd=fopen("dumpsub.srt","w");
2170 if(!fd)
2171 {
2172 perror("dump_srt: fopen");
2173 return;
2174 }
2175 for(i=0; i < subd->sub_num; i++)
2176 {
2177 onesub=subs+i; //=&subs[i];
2178 fprintf(fd,"%d\n",i+1);//line number
2179
2180 temp=onesub->start;
2181 if (!subd->sub_uses_time)
2182 temp = temp * 100 / sub_fps;
2183 temp -= sub_delay * 100;
2184 h=temp/360000;temp%=360000; //h =1*100*60*60
2185 m=temp/6000; temp%=6000; //m =1*100*60
2186 s=temp/100; temp%=100; //s =1*100
2187 ms=temp*10; //ms=1*10
2188 fprintf(fd,"%02d:%02d:%02d,%03d --> ",h,m,s,ms);
2189
2190 temp=onesub->end;
2191 if (!subd->sub_uses_time)
2192 temp = temp * 100 / sub_fps;
2193 temp -= sub_delay * 100;
2194 h=temp/360000;temp%=360000;
2195 m=temp/6000; temp%=6000;
2196 s=temp/100; temp%=100;
2197 ms=temp*10;
2198 fprintf(fd,"%02d:%02d:%02d,%03d\n",h,m,s,ms);
2199
2200 for(j=0;j<onesub->lines;j++)
2201 fprintf(fd,"%s\n",onesub->text[j]);
2202
2203 fprintf(fd,"\n");
2204 }
2205 fclose(fd);
2206 mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.srt\'.\n");
2207 }
2208
2209 void dump_mpsub(sub_data* subd, float fps){
2210 int i,j;
2211 FILE *fd;
2212 float a,b;
2213 subtitle *subs = subd->subtitles;
2214
2215 mpsub_position = subd->sub_uses_time? (sub_delay*100) : (sub_delay*fps);
2216 if (sub_fps==0) sub_fps=fps;
2217
2218 fd=fopen ("dump.mpsub", "w");
2219 if (!fd) {
2220 perror ("dump_mpsub: fopen");
2221 return;
2222 }
2223
2224
2225 if (subd->sub_uses_time) fprintf (fd,"FORMAT=TIME\n\n");
2226 else fprintf (fd, "FORMAT=%5.2f\n\n", fps);
2227
2228 for(j=0; j < subd->sub_num; j++){
2229 subtitle* egysub=&subs[j];
2230 if (subd->sub_uses_time) {
2231 a=((egysub->start-mpsub_position)/100.0);
2232 b=((egysub->end-egysub->start)/100.0);
2233 if ( (float)((int)a) == a)
2234 fprintf (fd, "%.0f",a);
2235 else
2236 fprintf (fd, "%.2f",a);
2237
2238 if ( (float)((int)b) == b)
2239 fprintf (fd, " %.0f\n",b);
2240 else
2241 fprintf (fd, " %.2f\n",b);
2242 } else {
2243 fprintf (fd, "%ld %ld\n", (long)((egysub->start*(fps/sub_fps))-((mpsub_position*(fps/sub_fps)))),
2244 (long)(((egysub->end)-(egysub->start))*(fps/sub_fps)));
2245 }
2246
2247 mpsub_position = egysub->end;
2248 for (i=0; i<egysub->lines; i++) {
2249 fprintf (fd, "%s\n",egysub->text[i]);
2250 }
2251 fprintf (fd, "\n");
2252 }
2253 fclose (fd);
2254 mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dump.mpsub\'.\n");
2255 }
2256
2257 void dump_microdvd(sub_data* subd, float fps) {
2258 int i, delay;
2259 FILE *fd;
2260 subtitle *subs = subd->subtitles;
2261 if (sub_fps == 0)
2262 sub_fps = fps;
2263 fd = fopen("dumpsub.sub", "w");
2264 if (!fd) {
2265 perror("dumpsub.sub: fopen");
2266 return;
2267 }
2268 delay = sub_delay * sub_fps;
2269 for (i = 0; i < subd->sub_num; ++i) {
2270 int j, start, end;
2271 start = subs[i].start;
2272 end = subs[i].end;
2273 if (subd->sub_uses_time) {
2274 start = start * sub_fps / 100 ;
2275 end = end * sub_fps / 100;
2276 }
2277 else {
2278 start = start * sub_fps / fps;
2279 end = end * sub_fps / fps;
2280 }
2281 start -= delay;
2282 end -= delay;
2283 fprintf(fd, "{%d}{%d}", start, end);
2284 for (j = 0; j < subs[i].lines; ++j)
2285 fprintf(fd, "%s%s", j ? "|" : "", subs[i].text[j]);
2286 fprintf(fd, "\n");
2287 }
2288 fclose(fd);
2289 mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.sub\'.\n");
2290 }
2291
2292 void dump_jacosub(sub_data* subd, float fps) {
2293 int i,j;
2294 int h,m,s,cs;
2295 FILE * fd;
2296 subtitle * onesub;
2297 unsigned long temp;
2298 subtitle *subs = subd->subtitles;
2299
2300 if (!subd->sub_uses_time && sub_fps == 0)
2301 sub_fps = fps;
2302 fd=fopen("dumpsub.jss","w");
2303 if(!fd)
2304 {
2305 perror("dump_jacosub: fopen");
2306 return;
2307 }
2308 fprintf(fd, "#TIMERES %d\n", (subd->sub_uses_time) ? 100 : (int)sub_fps);
2309 for(i=0; i < subd->sub_num; i++)
2310 {
2311 onesub=subs+i; //=&subs[i];
2312
2313 temp=onesub->start;
2314 if (!subd->sub_uses_time)
2315 temp = temp * 100 / sub_fps;
2316 temp -= sub_delay * 100;
2317 h=temp/360000;temp%=360000; //h =1*100*60*60
2318 m=temp/6000; temp%=6000; //m =1*100*60
2319 s=temp/100; temp%=100; //s =1*100
2320 cs=temp; //cs=1*10
2321 fprintf(fd,"%02d:%02d:%02d.%02d ",h,m,s,cs);
2322
2323 temp=onesub->end;
2324 if (!subd->sub_uses_time)
2325 temp = temp * 100 / sub_fps;
2326 temp -= sub_delay * 100;
2327 h=temp/360000;temp%=360000;
2328 m=temp/6000; temp%=6000;
2329 s=temp/100; temp%=100;
2330 cs=temp;
2331 fprintf(fd,"%02d:%02d:%02d.%02d {~} ",h,m,s,cs);
2332
2333 for(j=0;j<onesub->lines;j++)
2334 fprintf(fd,"%s%s",j ? "\\n" : "", onesub->text[j]);
2335
2336 fprintf(fd,"\n");
2337 }
2338 fclose(fd);
2339 mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.js\'.\n");
2340 }
2341
2342 void dump_sami(sub_data* subd, float fps) {
2343 int i,j;
2344 FILE * fd;
2345 subtitle * onesub;
2346 unsigned long temp;
2347 subtitle *subs = subd->subtitles;
2348
2349 if (!subd->sub_uses_time && sub_fps == 0)
2350 sub_fps = fps;
2351 fd=fopen("dumpsub.smi","w");
2352 if(!fd)
2353 {
2354 perror("dump_jacosub: fopen");
2355 return;
2356 }
2357 fprintf(fd, "<SAMI>\n"
2358 "<HEAD>\n"
2359 " <STYLE TYPE=\"Text/css\">\n"
2360 " <!--\n"
2361 " P {margin-left: 29pt; margin-right: 29pt; font-size: 24pt; text-align: center; font-family: Tahoma; font-weight: bold; color: #FCDD03; background-color: #000000;}\n"
2362 " .SUBTTL {Name: 'Subtitles'; Lang: en-US; SAMIType: CC;}\n"
2363 " -->\n"
2364 " </STYLE>\n"
2365 "</HEAD>\n"
2366 "<BODY>\n");
2367 for(i=0; i < subd->sub_num; i++)
2368 {
2369 onesub=subs+i; //=&subs[i];
2370
2371 temp=onesub->start;
2372 if (!subd->sub_uses_time)
2373 temp = temp * 100 / sub_fps;
2374 temp -= sub_delay * 100;
2375 fprintf(fd,"\t<SYNC Start=%lu>\n"
2376 "\t <P>", temp * 10);
2377
2378 for(j=0;j<onesub->lines;j++)
2379 fprintf(fd,"%s%s",j ? "<br>" : "", onesub->text[j]);
2380
2381 fprintf(fd,"\n");
2382
2383 temp=onesub->end;
2384 if (!subd->sub_uses_time)
2385 temp = temp * 100 / sub_fps;
2386 temp -= sub_delay * 100;
2387 fprintf(fd,"\t<SYNC Start=%lu>\n"
2388 "\t <P>&nbsp;\n", temp * 10);
2389 }
2390 fprintf(fd, "</BODY>\n"
2391 "</SAMI>\n");
2392 fclose(fd);
2393 mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.smi\'.\n");
2394 }
2395
2396 void sub_free( sub_data * subd )
2397 {
2398 int i, j;
2399
2400 if ( !subd ) return;
2401
2402 for (i = 0; i < subd->sub_num; i++)
2403 for (j = 0; j < subd->subtitles[i].lines; j++)
2404 free( subd->subtitles[i].text[j] );
2405 free( subd->subtitles );
2406 free( subd->filename );
2407 free( subd );
2408 }
2409
2410 #define MAX_SUBLINE 512
2411 /**
2412 * \brief parse text and append it to subtitle in sub
2413 * \param sub subtitle struct to add text to
2414 * \param txt text to parse
2415 * \param len length of text in txt
2416 * \param endpts pts at which this subtitle text should be removed again
2417 *
2418 * <> and {} are interpreted as comment delimiters, "\n", "\N", '\n', '\r'
2419 * and '\0' are interpreted as newlines, duplicate, leading and trailing
2420 * newlines are ignored.
2421 */
2422 void sub_add_text(subtitle *sub, const char *txt, int len, double endpts) {
2423 int comment = 0;
2424 int double_newline = 1; // ignore newlines at the beginning
2425 int i, pos;
2426 char *buf;
2427 #ifdef CONFIG_FRIBIDI
2428 int orig_lines = sub->lines;
2429 #endif
2430 if (sub->lines >= SUB_MAX_TEXT) return;
2431 pos = 0;
2432 buf = malloc(MAX_SUBLINE + 1);
2433 sub->text[sub->lines] = buf;
2434 sub->endpts[sub->lines] = endpts;
2435 for (i = 0; i < len && pos < MAX_SUBLINE; i++) {
2436 char c = txt[i];
2437 if (c == '<') comment |= 1;
2438 if (c == '{') comment |= 2;
2439 if (comment) {
2440 if (c == '}') comment &= ~2;
2441 if (c == '>') comment &= ~1;
2442 continue;
2443 }
2444 if (pos == MAX_SUBLINE - 1) {
2445 i--;
2446 c = 0;
2447 }
2448 if (c == '\\' && i + 1 < len) {
2449 c = txt[++i];
2450 if (c == 'n' || c == 'N') c = 0;
2451 }
2452 if (c == '\n' || c == '\r') c = 0;
2453 if (c) {
2454 double_newline = 0;
2455 buf[pos++] = c;
2456 } else if (!double_newline) {
2457 if (sub->lines >= SUB_MAX_TEXT - 1) {
2458 mp_msg(MSGT_VO, MSGL_WARN, "Too many subtitle lines\n");
2459 break;
2460 }
2461 double_newline = 1;
2462 buf[pos] = 0;
2463 sub->lines++;
2464 pos = 0;
2465 buf = malloc(MAX_SUBLINE + 1);
2466 sub->text[sub->lines] = buf;
2467 sub->endpts[sub->lines] = endpts;
2468 }
2469 }
2470 buf[pos] = 0;
2471 if (sub->lines < SUB_MAX_TEXT &&
2472 strlen(sub->text[sub->lines]))
2473 sub->lines++;
2474 #ifdef CONFIG_FRIBIDI
2475 sub = sub_fribidi(sub, sub_utf8, orig_lines);
2476 #endif
2477 }
2478
2479 /**
2480 * \brief remove outdated subtitle lines.
2481 * \param sub subtitle struct to modify
2482 * \param pts current pts. All lines with endpts <= this will be removed.
2483 * Use MP_NOPTS_VALUE to remove all lines
2484 * \return 1 if sub was modified, 0 otherwise.
2485 */
2486 int sub_clear_text(subtitle *sub, double pts) {
2487 int i = 0;
2488 int changed = 0;
2489 while (i < sub->lines) {
2490 double endpts = sub->endpts[i];
2491 if (pts == MP_NOPTS_VALUE || (endpts != MP_NOPTS_VALUE && pts >= endpts)) {
2492 int j;
2493 free(sub->text[i]);
2494 for (j = i + 1; j < sub->lines; j++) {
2495 sub->text[j - 1] = sub->text[j];
2496 sub->endpts[j - 1] = sub->endpts[j];
2497 }
2498 sub->lines--;
2499 changed = 1;
2500 } else
2501 i++;
2502 }
2503 return changed;
2504 }