Fix vf_tcdump's compilation
[mplayer/kovensky.git] / subreader.c
bloba972cdb874f557edee533d16872a03c3ffa6e3c6
1 /*
2 * Subtitle reader with format autodetection
4 * Copyright (c) 2001 laaz
5 * Some code cleanup & realloc() by A'rpi/ESP-team
7 * This file is part of MPlayer.
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.
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.
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.
24 #include <stdio.h>
25 #include <stdlib.h>
26 #include <string.h>
27 #include <ctype.h>
29 #include <sys/types.h>
30 #include <dirent.h>
32 #include "config.h"
33 #include "mp_msg.h"
34 #include "subreader.h"
35 #include "stream/stream.h"
36 #include "libavutil/common.h"
37 #include "libavutil/avstring.h"
39 #ifdef CONFIG_ENCA
40 #include <enca.h>
41 #endif
43 #define ERR ((void *) -1)
45 #ifdef CONFIG_ICONV
46 #include <iconv.h>
47 char *sub_cp=NULL;
48 #endif
49 #ifdef CONFIG_FRIBIDI
50 #include <fribidi/fribidi.h>
51 char *fribidi_charset = NULL; ///character set that will be passed to FriBiDi
52 int flip_hebrew = 1; ///flip subtitles using fribidi
53 int fribidi_flip_commas = 0; ///flip comma when fribidi is used
54 #endif
56 extern char* dvdsub_lang;
58 /* Maximal length of line of a subtitle */
59 #define LINE_LEN 1000
60 static float mpsub_position=0;
61 static float mpsub_multiplier=1.;
62 static int sub_slacktime = 20000; //20 sec
64 int sub_no_text_pp=0; // 1 => do not apply text post-processing
65 // like {\...} elimination in SSA format.
67 int sub_match_fuzziness=0; // level of sub name matching fuzziness
69 /* Use the SUB_* constant defined in the header file */
70 int sub_format=SUB_INVALID;
71 #ifdef CONFIG_SORTSUB
73 Some subtitling formats, namely AQT and Subrip09, define the end of a
74 subtitle as the beginning of the following. Since currently we read one
75 subtitle at time, for these format we keep two global *subtitle,
76 previous_aqt_sub and previous_subrip09_sub, pointing to previous subtitle,
77 so we can change its end when we read current subtitle starting time.
78 When CONFIG_SORTSUB is defined, we use a single global unsigned long,
79 previous_sub_end, for both (and even future) formats, to store the end of
80 the previous sub: it is initialized to 0 in sub_read_file and eventually
81 modified by sub_read_aqt_line or sub_read_subrip09_line.
83 unsigned long previous_sub_end;
84 #endif
86 static int eol(char p) {
87 return p=='\r' || p=='\n' || p=='\0';
90 /* Remove leading and trailing space */
91 static void trail_space(char *s) {
92 int i = 0;
93 while (isspace(s[i])) ++i;
94 if (i) strcpy(s, s + i);
95 i = strlen(s) - 1;
96 while (i > 0 && isspace(s[i])) s[i--] = '\0';
99 static char *stristr(const char *haystack, const char *needle) {
100 int len = 0;
101 const char *p = haystack;
103 if (!(haystack && needle)) return NULL;
105 len=strlen(needle);
106 while (*p != '\0') {
107 if (strncasecmp(p, needle, len) == 0) return (char*)p;
108 p++;
111 return NULL;
114 static subtitle *sub_read_line_sami(stream_t* st, subtitle *current, int utf16) {
115 static char line[LINE_LEN+1];
116 static char *s = NULL, *slacktime_s;
117 char text[LINE_LEN+1], *p=NULL, *q;
118 int state;
120 current->lines = current->start = current->end = 0;
121 current->alignment = SUB_ALIGNMENT_BOTTOMCENTER;
122 state = 0;
124 /* read the first line */
125 if (!s)
126 if (!(s = stream_read_line(st, line, LINE_LEN, utf16))) return 0;
128 do {
129 switch (state) {
131 case 0: /* find "START=" or "Slacktime:" */
132 slacktime_s = stristr (s, "Slacktime:");
133 if (slacktime_s)
134 sub_slacktime = strtol (slacktime_s+10, NULL, 0) / 10;
136 s = stristr (s, "Start=");
137 if (s) {
138 current->start = strtol (s + 6, &s, 0) / 10;
139 /* eat '>' */
140 for (; *s != '>' && *s != '\0'; s++);
141 s++;
142 state = 1; continue;
144 break;
146 case 1: /* find (optionnal) "<P", skip other TAGs */
147 for (; *s == ' ' || *s == '\t'; s++); /* strip blanks, if any */
148 if (*s == '\0') break;
149 if (*s != '<') { state = 3; p = text; continue; } /* not a TAG */
150 s++;
151 if (*s == 'P' || *s == 'p') { s++; state = 2; continue; } /* found '<P' */
152 for (; *s != '>' && *s != '\0'; s++); /* skip remains of non-<P> TAG */
153 if (s == '\0')
154 break;
155 s++;
156 continue;
158 case 2: /* find ">" */
159 if ((s = strchr (s, '>'))) { s++; state = 3; p = text; continue; }
160 break;
162 case 3: /* get all text until '<' appears */
163 if (*s == '\0') break;
164 else if (!strncasecmp (s, "<br>", 4)) {
165 *p = '\0'; p = text; trail_space (text);
166 if (text[0] != '\0')
167 current->text[current->lines++] = strdup (text);
168 s += 4;
170 else if ((*s == '{') && !sub_no_text_pp) { state = 5; ++s; continue; }
171 else if (*s == '<') { state = 4; }
172 else if (!strncasecmp (s, "&nbsp;", 6)) { *p++ = ' '; s += 6; }
173 else if (*s == '\t') { *p++ = ' '; s++; }
174 else if (*s == '\r' || *s == '\n') { s++; }
175 else *p++ = *s++;
177 /* skip duplicated space */
178 if (p > text + 2) if (*(p-1) == ' ' && *(p-2) == ' ') p--;
180 continue;
182 case 4: /* get current->end or skip <TAG> */
183 q = stristr (s, "Start=");
184 if (q) {
185 current->end = strtol (q + 6, &q, 0) / 10 - 1;
186 *p = '\0'; trail_space (text);
187 if (text[0] != '\0')
188 current->text[current->lines++] = strdup (text);
189 if (current->lines > 0) { state = 99; break; }
190 state = 0; continue;
192 s = strchr (s, '>');
193 if (s) { s++; state = 3; continue; }
194 break;
195 case 5: /* get rid of {...} text, but read the alignment code */
196 if ((*s == '\\') && (*(s + 1) == 'a') && !sub_no_text_pp) {
197 if (stristr(s, "\\a1") != NULL) {
198 current->alignment = SUB_ALIGNMENT_BOTTOMLEFT;
199 s = s + 3;
201 if (stristr(s, "\\a2") != NULL) {
202 current->alignment = SUB_ALIGNMENT_BOTTOMCENTER;
203 s = s + 3;
204 } else if (stristr(s, "\\a3") != NULL) {
205 current->alignment = SUB_ALIGNMENT_BOTTOMRIGHT;
206 s = s + 3;
207 } else if ((stristr(s, "\\a4") != NULL) || (stristr(s, "\\a5") != NULL) || (stristr(s, "\\a8") != NULL)) {
208 current->alignment = SUB_ALIGNMENT_TOPLEFT;
209 s = s + 3;
210 } else if (stristr(s, "\\a6") != NULL) {
211 current->alignment = SUB_ALIGNMENT_TOPCENTER;
212 s = s + 3;
213 } else if (stristr(s, "\\a7") != NULL) {
214 current->alignment = SUB_ALIGNMENT_TOPRIGHT;
215 s = s + 3;
216 } else if (stristr(s, "\\a9") != NULL) {
217 current->alignment = SUB_ALIGNMENT_MIDDLELEFT;
218 s = s + 3;
219 } else if (stristr(s, "\\a10") != NULL) {
220 current->alignment = SUB_ALIGNMENT_MIDDLECENTER;
221 s = s + 4;
222 } else if (stristr(s, "\\a11") != NULL) {
223 current->alignment = SUB_ALIGNMENT_MIDDLERIGHT;
224 s = s + 4;
227 if (*s == '}') state = 3;
228 ++s;
229 continue;
232 /* read next line */
233 if (state != 99 && !(s = stream_read_line (st, line, LINE_LEN, utf16))) {
234 if (current->start > 0) {
235 break; // if it is the last subtitle
236 } else {
237 return 0;
241 } while (state != 99);
243 // For the last subtitle
244 if (current->end <= 0) {
245 current->end = current->start + sub_slacktime;
246 *p = '\0'; trail_space (text);
247 if (text[0] != '\0')
248 current->text[current->lines++] = strdup (text);
251 return current;
255 static char *sub_readtext(char *source, char **dest) {
256 int len=0;
257 char *p=source;
259 // printf("src=%p dest=%p \n",source,dest);
261 while ( !eol(*p) && *p!= '|' ) {
262 p++,len++;
265 *dest= malloc (len+1);
266 if (!dest) {return ERR;}
268 strncpy(*dest, source, len);
269 (*dest)[len]=0;
271 while (*p=='\r' || *p=='\n' || *p=='|') p++;
273 if (*p) return p; // not-last text field
274 else return NULL; // last text field
277 static subtitle *sub_read_line_microdvd(stream_t *st,subtitle *current, int utf16) {
278 char line[LINE_LEN+1];
279 char line2[LINE_LEN+1];
280 char *p, *next;
281 int i;
283 do {
284 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
285 } while ((sscanf (line,
286 "{%ld}{}%[^\r\n]",
287 &(current->start), line2) < 2) &&
288 (sscanf (line,
289 "{%ld}{%ld}%[^\r\n]",
290 &(current->start), &(current->end), line2) < 3));
292 p=line2;
294 next=p, i=0;
295 while ((next =sub_readtext (next, &(current->text[i])))) {
296 if (current->text[i]==ERR) {return ERR;}
297 i++;
298 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
300 current->lines= ++i;
302 return current;
305 static subtitle *sub_read_line_mpl2(stream_t *st,subtitle *current, int utf16) {
306 char line[LINE_LEN+1];
307 char line2[LINE_LEN+1];
308 char *p, *next;
309 int i;
311 do {
312 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
313 } while ((sscanf (line,
314 "[%ld][%ld]%[^\r\n]",
315 &(current->start), &(current->end), line2) < 3));
316 current->start *= 10;
317 current->end *= 10;
318 p=line2;
320 next=p, i=0;
321 while ((next =sub_readtext (next, &(current->text[i])))) {
322 if (current->text[i]==ERR) {return ERR;}
323 i++;
324 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
326 current->lines= ++i;
328 return current;
331 static subtitle *sub_read_line_subrip(stream_t* st, subtitle *current, int utf16) {
332 char line[LINE_LEN+1];
333 int a1,a2,a3,a4,b1,b2,b3,b4;
334 char *p=NULL, *q=NULL;
335 int len;
337 while (1) {
338 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
339 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4) < 8) continue;
340 current->start = a1*360000+a2*6000+a3*100+a4;
341 current->end = b1*360000+b2*6000+b3*100+b4;
343 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
345 p=q=line;
346 for (current->lines=1; current->lines < SUB_MAX_TEXT; current->lines++) {
347 for (q=p,len=0; *p && *p!='\r' && *p!='\n' && *p!='|' && strncmp(p,"[br]",4); p++,len++);
348 current->text[current->lines-1]=malloc (len+1);
349 if (!current->text[current->lines-1]) return ERR;
350 strncpy (current->text[current->lines-1], q, len);
351 current->text[current->lines-1][len]='\0';
352 if (!*p || *p=='\r' || *p=='\n') break;
353 if (*p=='|') p++;
354 else while (*p++!=']');
356 break;
358 return current;
361 static subtitle *sub_read_line_subviewer(stream_t *st,subtitle *current, int utf16) {
362 char line[LINE_LEN+1];
363 int a1,a2,a3,a4,b1,b2,b3,b4;
364 char *p=NULL;
365 int i,len;
367 while (!current->text[0]) {
368 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
369 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)
370 continue;
371 current->start = a1*360000+a2*6000+a3*100+a4/10;
372 current->end = b1*360000+b2*6000+b3*100+b4/10;
373 for (i=0; i<SUB_MAX_TEXT;) {
374 int blank = 1;
375 if (!stream_read_line (st, line, LINE_LEN, utf16)) break;
376 len=0;
377 for (p=line; *p!='\n' && *p!='\r' && *p; p++,len++)
378 if (*p != ' ' && *p != '\t')
379 blank = 0;
380 if (len && !blank) {
381 int j=0,skip=0;
382 char *curptr=current->text[i]=malloc (len+1);
383 if (!current->text[i]) return ERR;
384 //strncpy (current->text[i], line, len); current->text[i][len]='\0';
385 for(; j<len; j++) {
386 /* let's filter html tags ::atmos */
387 if(line[j]=='>') {
388 skip=0;
389 continue;
391 if(line[j]=='<') {
392 skip=1;
393 continue;
395 if(skip) {
396 continue;
398 *curptr=line[j];
399 curptr++;
401 *curptr='\0';
403 i++;
404 } else {
405 break;
408 current->lines=i;
410 return current;
413 static subtitle *sub_read_line_subviewer2(stream_t *st,subtitle *current, int utf16) {
414 char line[LINE_LEN+1];
415 int a1,a2,a3,a4;
416 char *p=NULL;
417 int i,len;
419 while (!current->text[0]) {
420 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
421 if (line[0]!='{')
422 continue;
423 if ((len=sscanf (line, "{T %d:%d:%d:%d",&a1,&a2,&a3,&a4)) < 4)
424 continue;
425 current->start = a1*360000+a2*6000+a3*100+a4/10;
426 for (i=0; i<SUB_MAX_TEXT;) {
427 if (!stream_read_line (st, line, LINE_LEN, utf16)) break;
428 if (line[0]=='}') break;
429 len=0;
430 for (p=line; *p!='\n' && *p!='\r' && *p; ++p,++len);
431 if (len) {
432 current->text[i]=malloc (len+1);
433 if (!current->text[i]) return ERR;
434 strncpy (current->text[i], line, len); current->text[i][len]='\0';
435 ++i;
436 } else {
437 break;
440 current->lines=i;
442 return current;
446 static subtitle *sub_read_line_vplayer(stream_t *st,subtitle *current, int utf16) {
447 char line[LINE_LEN+1];
448 int a1,a2,a3;
449 char *p=NULL, *next,separator;
450 int i,len,plen;
452 while (!current->text[0]) {
453 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
454 if ((len=sscanf (line, "%d:%d:%d%c%n",&a1,&a2,&a3,&separator,&plen)) < 4)
455 continue;
457 if (!(current->start = a1*360000+a2*6000+a3*100))
458 continue;
459 /* removed by wodzu
460 p=line;
461 // finds the body of the subtitle
462 for (i=0; i<3; i++){
463 p=strchr(p,':');
464 if (p==NULL) break;
465 ++p;
467 if (p==NULL) {
468 printf("SUB: Skipping incorrect subtitle line!\n");
469 continue;
472 // by wodzu: hey! this time we know what length it has! what is
473 // that magic for? it can't deal with space instead of third
474 // colon! look, what simple it can be:
475 p = &line[ plen ];
477 i=0;
478 if (*p!='|') {
480 next = p,i=0;
481 while ((next =sub_readtext (next, &(current->text[i])))) {
482 if (current->text[i]==ERR) {return ERR;}
483 i++;
484 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
486 current->lines=i+1;
489 return current;
492 static subtitle *sub_read_line_rt(stream_t *st,subtitle *current, int utf16) {
493 //TODO: This format uses quite rich (sub/super)set of xhtml
494 // I couldn't check it since DTD is not included.
495 // WARNING: full XML parses can be required for proper parsing
496 char line[LINE_LEN+1];
497 int a1,a2,a3,a4,b1,b2,b3,b4;
498 char *p=NULL,*next=NULL;
499 int i,len,plen;
501 while (!current->text[0]) {
502 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
503 //TODO: it seems that format of time is not easily determined, it may be 1:12, 1:12.0 or 0:1:12.0
504 //to describe the same moment in time. Maybe there are even more formats in use.
505 //if ((len=sscanf (line, "<Time Begin=\"%d:%d:%d.%d\" End=\"%d:%d:%d.%d\"",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8)
506 plen=a1=a2=a3=a4=b1=b2=b3=b4=0;
507 if (
508 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d.%d\" %*[Ee]nd=\"%d.%d\"%*[^<]<clear/>%n",&a3,&a4,&b3,&b4,&plen)) < 4) &&
509 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d.%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a3,&a4,&b2,&b3,&b4,&plen)) < 5) &&
510 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&plen)) < 4) &&
511 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&b4,&plen)) < 5) &&
512 // ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&b2,&b3,&plen)) < 5) &&
513 ((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) &&
514 ((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) &&
515 //now try it without end time
516 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d.%d\"%*[^<]<clear/>%n",&a3,&a4,&plen)) < 2) &&
517 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&plen)) < 2) &&
518 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&plen)) < 3) &&
519 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d:%d.%d\"%*[^<]<clear/>%n",&a1,&a2,&a3,&a4,&plen)) < 4)
521 continue;
522 current->start = a1*360000+a2*6000+a3*100+a4/10;
523 current->end = b1*360000+b2*6000+b3*100+b4/10;
524 if (b1 == 0 && b2 == 0 && b3 == 0 && b4 == 0)
525 current->end = current->start+200;
526 p=line; p+=plen;i=0;
527 // TODO: I don't know what kind of convention is here for marking multiline subs, maybe <br/> like in xml?
528 next = strstr(line,"<clear/>");
529 if(next && strlen(next)>8){
530 next+=8;i=0;
531 while ((next =sub_readtext (next, &(current->text[i])))) {
532 if (current->text[i]==ERR) {return ERR;}
533 i++;
534 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
537 current->lines=i+1;
539 return current;
542 static subtitle *sub_read_line_ssa(stream_t *st,subtitle *current, int utf16) {
544 * Sub Station Alpha v4 (and v2?) scripts have 9 commas before subtitle
545 * other Sub Station Alpha scripts have only 8 commas before subtitle
546 * Reading the "ScriptType:" field is not reliable since many scripts appear
547 * w/o it
549 * http://www.scriptclub.org is a good place to find more examples
550 * http://www.eswat.demon.co.uk is where the SSA specs can be found
552 int comma;
553 static int max_comma = 32; /* let's use 32 for the case that the */
554 /* amount of commas increase with newer SSA versions */
556 int hour1, min1, sec1, hunsec1,
557 hour2, min2, sec2, hunsec2, nothing;
558 int num;
560 char line[LINE_LEN+1],
561 line3[LINE_LEN+1],
562 *line2;
563 char *tmp;
565 do {
566 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
567 } while (sscanf (line, "Dialogue: Marked=%d,%d:%d:%d.%d,%d:%d:%d.%d,"
568 "%[^\n\r]", &nothing,
569 &hour1, &min1, &sec1, &hunsec1,
570 &hour2, &min2, &sec2, &hunsec2,
571 line3) < 9
573 sscanf (line, "Dialogue: %d,%d:%d:%d.%d,%d:%d:%d.%d,"
574 "%[^\n\r]", &nothing,
575 &hour1, &min1, &sec1, &hunsec1,
576 &hour2, &min2, &sec2, &hunsec2,
577 line3) < 9 );
579 line2=strchr(line3, ',');
581 for (comma = 4; comma < max_comma; comma ++)
583 tmp = line2;
584 if(!(tmp=strchr(++tmp, ','))) break;
585 if(*(++tmp) == ' ') break;
586 /* a space after a comma means we're already in a sentence */
587 line2 = tmp;
590 if(comma < max_comma)max_comma = comma;
591 /* eliminate the trailing comma */
592 if(*line2 == ',') line2++;
594 current->lines=0;num=0;
595 current->start = 360000*hour1 + 6000*min1 + 100*sec1 + hunsec1;
596 current->end = 360000*hour2 + 6000*min2 + 100*sec2 + hunsec2;
598 while (((tmp=strstr(line2, "\\n")) != NULL) || ((tmp=strstr(line2, "\\N")) != NULL) ){
599 current->text[num]=malloc(tmp-line2+1);
600 strncpy (current->text[num], line2, tmp-line2);
601 current->text[num][tmp-line2]='\0';
602 line2=tmp+2;
603 num++;
604 current->lines++;
605 if (current->lines >= SUB_MAX_TEXT) return current;
608 current->text[num]=strdup(line2);
609 current->lines++;
611 return current;
614 static void sub_pp_ssa(subtitle *sub) {
615 int l=sub->lines;
616 char *so,*de,*start;
618 while (l){
619 /* eliminate any text enclosed with {}, they are font and color settings */
620 so=de=sub->text[--l];
621 while (*so) {
622 if(*so == '{' && so[1]=='\\') {
623 for (start=so; *so && *so!='}'; so++);
624 if(*so) so++; else so=start;
626 if(*so) {
627 *de=*so;
628 so++; de++;
631 *de=*so;
636 * PJS subtitles reader.
637 * That's the "Phoenix Japanimation Society" format.
638 * I found some of them in http://www.scriptsclub.org/ (used for anime).
639 * The time is in tenths of second.
641 * by set, based on code by szabi (dunnowhat sub format ;-)
643 static subtitle *sub_read_line_pjs(stream_t *st,subtitle *current, int utf16) {
644 char line[LINE_LEN+1];
645 char text[LINE_LEN+1], *s, *d;
647 if (!stream_read_line (st, line, LINE_LEN, utf16))
648 return NULL;
649 /* skip spaces */
650 for (s=line; *s && isspace(*s); s++);
651 /* allow empty lines at the end of the file */
652 if (*s==0)
653 return NULL;
654 /* get the time */
655 if (sscanf (s, "%ld,%ld,", &(current->start),
656 &(current->end)) <2) {
657 return ERR;
659 /* the files I have are in tenths of second */
660 current->start *= 10;
661 current->end *= 10;
662 /* walk to the beggining of the string */
663 for (; *s; s++) if (*s==',') break;
664 if (*s) {
665 for (s++; *s; s++) if (*s==',') break;
666 if (*s) s++;
668 if (*s!='"') {
669 return ERR;
671 /* copy the string to the text buffer */
672 for (s++, d=text; *s && *s!='"'; s++, d++)
673 *d=*s;
674 *d=0;
675 current->text[0] = strdup(text);
676 current->lines = 1;
678 return current;
681 static subtitle *sub_read_line_mpsub(stream_t *st, subtitle *current, int utf16) {
682 char line[LINE_LEN+1];
683 float a,b;
684 int num=0;
685 char *p, *q;
689 if (!stream_read_line(st, line, LINE_LEN, utf16)) return NULL;
690 } while (sscanf (line, "%f %f", &a, &b) !=2);
692 mpsub_position += a*mpsub_multiplier;
693 current->start=(int) mpsub_position;
694 mpsub_position += b*mpsub_multiplier;
695 current->end=(int) mpsub_position;
697 while (num < SUB_MAX_TEXT) {
698 if (!stream_read_line (st, line, LINE_LEN, utf16)) {
699 if (num == 0) return NULL;
700 else return current;
702 p=line;
703 while (isspace(*p)) p++;
704 if (eol(*p) && num > 0) return current;
705 if (eol(*p)) return NULL;
707 for (q=p; !eol(*q); q++);
708 *q='\0';
709 if (strlen(p)) {
710 current->text[num]=strdup(p);
711 // printf (">%s<\n",p);
712 current->lines = ++num;
713 } else {
714 if (num) return current;
715 else return NULL;
718 return NULL; // we should have returned before if it's OK
721 #ifndef CONFIG_SORTSUB
722 //we don't need this if we use previous_sub_end
723 subtitle *previous_aqt_sub = NULL;
724 #endif
726 static subtitle *sub_read_line_aqt(stream_t *st,subtitle *current, int utf16) {
727 char line[LINE_LEN+1];
728 char *next;
729 int i;
731 while (1) {
732 // try to locate next subtitle
733 if (!stream_read_line (st, line, LINE_LEN, utf16))
734 return NULL;
735 if (!(sscanf (line, "-->> %ld", &(current->start)) <1))
736 break;
739 #ifdef CONFIG_SORTSUB
740 previous_sub_end = (current->start) ? current->start - 1 : 0;
741 #else
742 if (previous_aqt_sub != NULL)
743 previous_aqt_sub->end = current->start-1;
745 previous_aqt_sub = current;
746 #endif
748 if (!stream_read_line (st, line, LINE_LEN, utf16))
749 return NULL;
751 sub_readtext((char *) &line,&current->text[0]);
752 current->lines = 1;
753 current->end = current->start; // will be corrected by next subtitle
755 if (!stream_read_line (st, line, LINE_LEN, utf16))
756 return current;
758 next = line,i=1;
759 while ((next =sub_readtext (next, &(current->text[i])))) {
760 if (current->text[i]==ERR) {return ERR;}
761 i++;
762 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
764 current->lines=i+1;
766 if (!strlen(current->text[0]) && !strlen(current->text[1])) {
767 #ifdef CONFIG_SORTSUB
768 previous_sub_end = 0;
769 #else
770 // void subtitle -> end of previous marked and exit
771 previous_aqt_sub = NULL;
772 #endif
773 return NULL;
776 return current;
779 #ifndef CONFIG_SORTSUB
780 subtitle *previous_subrip09_sub = NULL;
781 #endif
783 static subtitle *sub_read_line_subrip09(stream_t *st,subtitle *current, int utf16) {
784 char line[LINE_LEN+1];
785 int a1,a2,a3;
786 char * next=NULL;
787 int i,len;
789 while (1) {
790 // try to locate next subtitle
791 if (!stream_read_line (st, line, LINE_LEN, utf16))
792 return NULL;
793 if (!((len=sscanf (line, "[%d:%d:%d]",&a1,&a2,&a3)) < 3))
794 break;
797 current->start = a1*360000+a2*6000+a3*100;
799 #ifdef CONFIG_SORTSUB
800 previous_sub_end = (current->start) ? current->start - 1 : 0;
801 #else
802 if (previous_subrip09_sub != NULL)
803 previous_subrip09_sub->end = current->start-1;
805 previous_subrip09_sub = current;
806 #endif
808 if (!stream_read_line (st, line, LINE_LEN, utf16))
809 return NULL;
811 next = line,i=0;
813 current->text[0]=""; // just to be sure that string is clear
815 while ((next =sub_readtext (next, &(current->text[i])))) {
816 if (current->text[i]==ERR) {return ERR;}
817 i++;
818 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
820 current->lines=i+1;
822 if (!strlen(current->text[0]) && (i==0)) {
823 #ifdef CONFIG_SORTSUB
824 previous_sub_end = 0;
825 #else
826 // void subtitle -> end of previous marked and exit
827 previous_subrip09_sub = NULL;
828 #endif
829 return NULL;
832 return current;
835 static subtitle *sub_read_line_jacosub(stream_t* st, subtitle * current, int utf16)
837 char line1[LINE_LEN], line2[LINE_LEN], directive[LINE_LEN], *p, *q;
838 unsigned a1, a2, a3, a4, b1, b2, b3, b4, comment = 0;
839 static unsigned jacoTimeres = 30;
840 static int jacoShift = 0;
842 memset(current, 0, sizeof(subtitle));
843 memset(line1, 0, LINE_LEN);
844 memset(line2, 0, LINE_LEN);
845 memset(directive, 0, LINE_LEN);
846 while (!current->text[0]) {
847 if (!stream_read_line(st, line1, LINE_LEN, utf16)) {
848 return NULL;
850 if (sscanf
851 (line1, "%u:%u:%u.%u %u:%u:%u.%u %[^\n\r]", &a1, &a2, &a3, &a4,
852 &b1, &b2, &b3, &b4, line2) < 9) {
853 if (sscanf(line1, "@%u @%u %[^\n\r]", &a4, &b4, line2) < 3) {
854 if (line1[0] == '#') {
855 int hours = 0, minutes = 0, seconds, delta, inverter =
857 unsigned units = jacoShift;
858 switch (toupper(line1[1])) {
859 case 'S':
860 if (isalpha(line1[2])) {
861 delta = 6;
862 } else {
863 delta = 2;
865 if (sscanf(&line1[delta], "%d", &hours)) {
866 if (hours < 0) {
867 hours *= -1;
868 inverter = -1;
870 if (sscanf(&line1[delta], "%*d:%d", &minutes)) {
871 if (sscanf
872 (&line1[delta], "%*d:%*d:%d",
873 &seconds)) {
874 sscanf(&line1[delta], "%*d:%*d:%*d.%d",
875 &units);
876 } else {
877 hours = 0;
878 sscanf(&line1[delta], "%d:%d.%d",
879 &minutes, &seconds, &units);
880 minutes *= inverter;
882 } else {
883 hours = minutes = 0;
884 sscanf(&line1[delta], "%d.%d", &seconds,
885 &units);
886 seconds *= inverter;
888 jacoShift =
889 ((hours * 3600 + minutes * 60 +
890 seconds) * jacoTimeres +
891 units) * inverter;
893 break;
894 case 'T':
895 if (isalpha(line1[2])) {
896 delta = 8;
897 } else {
898 delta = 2;
900 sscanf(&line1[delta], "%u", &jacoTimeres);
901 break;
904 continue;
905 } else {
906 current->start =
907 (unsigned long) ((a4 + jacoShift) * 100.0 /
908 jacoTimeres);
909 current->end =
910 (unsigned long) ((b4 + jacoShift) * 100.0 /
911 jacoTimeres);
913 } else {
914 current->start =
915 (unsigned
916 long) (((a1 * 3600 + a2 * 60 + a3) * jacoTimeres + a4 +
917 jacoShift) * 100.0 / jacoTimeres);
918 current->end =
919 (unsigned
920 long) (((b1 * 3600 + b2 * 60 + b3) * jacoTimeres + b4 +
921 jacoShift) * 100.0 / jacoTimeres);
923 current->lines = 0;
924 p = line2;
925 while ((*p == ' ') || (*p == '\t')) {
926 ++p;
928 if (isalpha(*p)||*p == '[') {
929 int cont, jLength;
931 if (sscanf(p, "%s %[^\n\r]", directive, line1) < 2)
932 return (subtitle *) ERR;
933 jLength = strlen(directive);
934 for (cont = 0; cont < jLength; ++cont) {
935 if (isalpha(*(directive + cont)))
936 *(directive + cont) = toupper(*(directive + cont));
938 if ((strstr(directive, "RDB") != NULL)
939 || (strstr(directive, "RDC") != NULL)
940 || (strstr(directive, "RLB") != NULL)
941 || (strstr(directive, "RLG") != NULL)) {
942 continue;
944 if (strstr(directive, "JL") != NULL) {
945 current->alignment = SUB_ALIGNMENT_BOTTOMLEFT;
946 } else if (strstr(directive, "JR") != NULL) {
947 current->alignment = SUB_ALIGNMENT_BOTTOMRIGHT;
948 } else {
949 current->alignment = SUB_ALIGNMENT_BOTTOMCENTER;
951 strcpy(line2, line1);
952 p = line2;
954 for (q = line1; (!eol(*p)) && (current->lines < SUB_MAX_TEXT); ++p) {
955 switch (*p) {
956 case '{':
957 comment++;
958 break;
959 case '}':
960 if (comment) {
961 --comment;
962 //the next line to get rid of a blank after the comment
963 if ((*(p + 1)) == ' ')
964 p++;
966 break;
967 case '~':
968 if (!comment) {
969 *q = ' ';
970 ++q;
972 break;
973 case ' ':
974 case '\t':
975 if ((*(p + 1) == ' ') || (*(p + 1) == '\t'))
976 break;
977 if (!comment) {
978 *q = ' ';
979 ++q;
981 break;
982 case '\\':
983 if (*(p + 1) == 'n') {
984 *q = '\0';
985 q = line1;
986 current->text[current->lines++] = strdup(line1);
987 ++p;
988 break;
990 if ((toupper(*(p + 1)) == 'C')
991 || (toupper(*(p + 1)) == 'F')) {
992 ++p,++p;
993 break;
995 if ((*(p + 1) == 'B') || (*(p + 1) == 'b') || (*(p + 1) == 'D') || //actually this means "insert current date here"
996 (*(p + 1) == 'I') || (*(p + 1) == 'i') || (*(p + 1) == 'N') || (*(p + 1) == 'T') || //actually this means "insert current time here"
997 (*(p + 1) == 'U') || (*(p + 1) == 'u')) {
998 ++p;
999 break;
1001 if ((*(p + 1) == '\\') ||
1002 (*(p + 1) == '~') || (*(p + 1) == '{')) {
1003 ++p;
1004 } else if (eol(*(p + 1))) {
1005 if (!stream_read_line(st, directive, LINE_LEN, utf16))
1006 return NULL;
1007 trail_space(directive);
1008 av_strlcat(line2, directive, LINE_LEN);
1009 break;
1011 default:
1012 if (!comment) {
1013 *q = *p;
1014 ++q;
1016 } //-- switch
1017 } //-- for
1018 *q = '\0';
1019 current->text[current->lines] = strdup(line1);
1020 } //-- while
1021 current->lines++;
1022 return current;
1025 static int sub_autodetect (stream_t* st, int *uses_time, int utf16) {
1026 char line[LINE_LEN+1];
1027 int i,j=0;
1028 char p;
1030 while (j < 100) {
1031 j++;
1032 if (!stream_read_line (st, line, LINE_LEN, utf16))
1033 return SUB_INVALID;
1035 if (sscanf (line, "{%d}{%d}", &i, &i)==2)
1036 {*uses_time=0;return SUB_MICRODVD;}
1037 if (sscanf (line, "{%d}{}", &i)==1)
1038 {*uses_time=0;return SUB_MICRODVD;}
1039 if (sscanf (line, "[%d][%d]", &i, &i)==2)
1040 {*uses_time=1;return SUB_MPL2;}
1041 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d", &i, &i, &i, &i, &i, &i, &i, &i)==8)
1042 {*uses_time=1;return SUB_SUBRIP;}
1043 if (sscanf (line, "%d:%d:%d%[,.:]%d --> %d:%d:%d%[,.:]%d", &i, &i, &i, (char *)&i, &i, &i, &i, &i, (char *)&i, &i)==10)
1044 {*uses_time=1;return SUB_SUBVIEWER;}
1045 if (sscanf (line, "{T %d:%d:%d:%d",&i, &i, &i, &i)==4)
1046 {*uses_time=1;return SUB_SUBVIEWER2;}
1047 if (strstr (line, "<SAMI>"))
1048 {*uses_time=1; return SUB_SAMI;}
1049 if (sscanf(line, "%d:%d:%d.%d %d:%d:%d.%d", &i, &i, &i, &i, &i, &i, &i, &i) == 8)
1050 {*uses_time = 1; return SUB_JACOSUB;}
1051 if (sscanf(line, "@%d @%d", &i, &i) == 2)
1052 {*uses_time = 1; return SUB_JACOSUB;}
1053 if (sscanf (line, "%d:%d:%d:", &i, &i, &i )==3)
1054 {*uses_time=1;return SUB_VPLAYER;}
1055 if (sscanf (line, "%d:%d:%d ", &i, &i, &i )==3)
1056 {*uses_time=1;return SUB_VPLAYER;}
1057 //TODO: just checking if first line of sub starts with "<" is WAY
1058 // too weak test for RT
1059 // Please someone who knows the format of RT... FIX IT!!!
1060 // It may conflict with other sub formats in the future (actually it doesn't)
1061 if ( *line == '<' )
1062 {*uses_time=1;return SUB_RT;}
1064 if (!memcmp(line, "Dialogue: Marked", 16))
1065 {*uses_time=1; return SUB_SSA;}
1066 if (!memcmp(line, "Dialogue: ", 10))
1067 {*uses_time=1; return SUB_SSA;}
1068 if (sscanf (line, "%d,%d,\"%c", &i, &i, (char *) &i) == 3)
1069 {*uses_time=1;return SUB_PJS;}
1070 if (sscanf (line, "FORMAT=%d", &i) == 1)
1071 {*uses_time=0; return SUB_MPSUB;}
1072 if (sscanf (line, "FORMAT=TIM%c", &p)==1 && p=='E')
1073 {*uses_time=1; return SUB_MPSUB;}
1074 if (strstr (line, "-->>"))
1075 {*uses_time=0; return SUB_AQTITLE;}
1076 if (sscanf (line, "[%d:%d:%d]", &i, &i, &i)==3)
1077 {*uses_time=1;return SUB_SUBRIP09;}
1080 return SUB_INVALID; // too many bad lines
1083 extern int sub_utf8;
1084 int sub_utf8_prev=0;
1086 extern float sub_delay;
1087 extern float sub_fps;
1089 #ifdef CONFIG_ICONV
1090 static iconv_t icdsc = (iconv_t)(-1);
1092 void subcp_open (stream_t *st)
1094 char *tocp = "UTF-8";
1096 if (sub_cp){
1097 const char *cp_tmp = sub_cp;
1098 #ifdef CONFIG_ENCA
1099 char enca_lang[3], enca_fallback[100];
1100 if (sscanf(sub_cp, "enca:%2s:%99s", enca_lang, enca_fallback) == 2
1101 || sscanf(sub_cp, "ENCA:%2s:%99s", enca_lang, enca_fallback) == 2) {
1102 if (st && st->flags & MP_STREAM_SEEK ) {
1103 cp_tmp = guess_cp(st, enca_lang, enca_fallback);
1104 } else {
1105 cp_tmp = enca_fallback;
1106 if (st)
1107 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: enca failed, stream must be seekable.\n");
1110 #endif
1111 if ((icdsc = iconv_open (tocp, cp_tmp)) != (iconv_t)(-1)){
1112 mp_msg(MSGT_SUBREADER,MSGL_V,"SUB: opened iconv descriptor.\n");
1113 sub_utf8 = 2;
1114 } else
1115 mp_msg(MSGT_SUBREADER,MSGL_ERR,"SUB: error opening iconv descriptor.\n");
1119 void subcp_close (void)
1121 if (icdsc != (iconv_t)(-1)){
1122 (void) iconv_close (icdsc);
1123 icdsc = (iconv_t)(-1);
1124 mp_msg(MSGT_SUBREADER,MSGL_V,"SUB: closed iconv descriptor.\n");
1128 subtitle* subcp_recode (subtitle *sub)
1130 int l=sub->lines;
1131 size_t ileft, oleft;
1132 char *op, *ip, *ot;
1133 if(icdsc == (iconv_t)(-1)) return sub;
1135 while (l){
1136 ip = sub->text[--l];
1137 ileft = strlen(ip);
1138 oleft = 4 * ileft;
1140 if (!(ot = malloc(oleft + 1))){
1141 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: error allocating mem.\n");
1142 continue;
1144 op = ot;
1145 if (iconv(icdsc, &ip, &ileft,
1146 &op, &oleft) == (size_t)(-1)) {
1147 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: error recoding line.\n");
1148 free(ot);
1149 continue;
1151 // In some stateful encodings, we must clear the state to handle the last character
1152 if (iconv(icdsc, NULL, NULL,
1153 &op, &oleft) == (size_t)(-1)) {
1154 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: error recoding line, can't clear encoding state.\n");
1156 *op='\0' ;
1157 free (sub->text[l]);
1158 sub->text[l] = ot;
1160 return sub;
1162 #endif
1164 #ifdef CONFIG_FRIBIDI
1166 * Do conversion necessary for right-to-left language support via fribidi.
1167 * @param sub subtitle to convert
1168 * @param sub_utf8 whether the subtitle is encoded in UTF-8
1169 * @param from first new subtitle, all lines before this are assumed to be already converted
1171 static subtitle* sub_fribidi (subtitle *sub, int sub_utf8, int from)
1173 FriBidiChar logical[LINE_LEN+1], visual[LINE_LEN+1]; // Hopefully these two won't smash the stack
1174 char *ip = NULL, *op = NULL;
1175 FriBidiCharType base;
1176 size_t len,orig_len;
1177 int l=sub->lines;
1178 int char_set_num;
1179 fribidi_boolean log2vis;
1180 if (!flip_hebrew)
1181 return sub;
1182 fribidi_set_mirroring(1);
1183 fribidi_set_reorder_nsm(0);
1185 if( sub_utf8 == 0 ) {
1186 char_set_num = fribidi_parse_charset (fribidi_charset?fribidi_charset:"ISO8859-8");
1187 }else {
1188 char_set_num = fribidi_parse_charset ("UTF-8");
1190 while (l > from) {
1191 ip = sub->text[--l];
1192 orig_len = len = strlen( ip ); // We assume that we don't use full unicode, only UTF-8 or ISO8859-x
1193 if(len > LINE_LEN) {
1194 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: sub->text is longer than LINE_LEN.\n");
1195 l++;
1196 break;
1198 len = fribidi_charset_to_unicode (char_set_num, ip, len, logical);
1199 base = fribidi_flip_commas?FRIBIDI_TYPE_ON:FRIBIDI_TYPE_L;
1200 log2vis = fribidi_log2vis (logical, len, &base,
1201 /* output */
1202 visual, NULL, NULL, NULL);
1203 if(log2vis) {
1204 len = fribidi_remove_bidi_marks (visual, len, NULL, NULL,
1205 NULL);
1206 if((op = malloc((FFMAX(2*orig_len,2*len) + 1))) == NULL) {
1207 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: error allocating mem.\n");
1208 l++;
1209 break;
1211 fribidi_unicode_to_charset ( char_set_num, visual, len,op);
1212 free (ip);
1213 sub->text[l] = op;
1216 if (!from && l){
1217 for (l = sub->lines; l;)
1218 free (sub->text[--l]);
1219 return ERR;
1221 return sub;
1224 #endif
1226 static void adjust_subs_time(subtitle* sub, float subtime, float fps, int block,
1227 int sub_num, int sub_uses_time) {
1228 int n,m;
1229 subtitle* nextsub;
1230 int i = sub_num;
1231 unsigned long subfms = (sub_uses_time ? 100 : fps) * subtime;
1232 unsigned long overlap = (sub_uses_time ? 100 : fps) / 5; // 0.2s
1234 n=m=0;
1235 if (i) for (;;){
1236 if (sub->end <= sub->start){
1237 sub->end = sub->start + subfms;
1238 m++;
1239 n++;
1241 if (!--i) break;
1242 nextsub = sub + 1;
1243 if(block){
1244 if ((sub->end > nextsub->start) && (sub->end <= nextsub->start + overlap)) {
1245 // these subtitles overlap for less than 0.2 seconds
1246 // and would result in very short overlapping subtitle
1247 // so let's fix the problem here, before overlapping code
1248 // get its hands on them
1249 unsigned delta = sub->end - nextsub->start, half = delta / 2;
1250 sub->end -= half + 1;
1251 nextsub->start += delta - half;
1253 if (sub->end >= nextsub->start){
1254 sub->end = nextsub->start - 1;
1255 if (sub->end - sub->start > subfms)
1256 sub->end = sub->start + subfms;
1257 if (!m)
1258 n++;
1262 /* Theory:
1263 * Movies are often converted from FILM (24 fps)
1264 * to PAL (25) by simply speeding it up, so we
1265 * to multiply the original timestmaps by
1266 * (Movie's FPS / Subtitle's (guessed) FPS)
1267 * so eg. for 23.98 fps movie and PAL time based
1268 * subtitles we say -subfps 25 and we're fine!
1271 /* timed sub fps correction ::atmos */
1272 /* the frame-based case is handled in mpcommon.c
1273 * where find_sub is called */
1274 if(sub_uses_time && sub_fps) {
1275 sub->start *= sub_fps/fps;
1276 sub->end *= sub_fps/fps;
1279 sub = nextsub;
1280 m = 0;
1282 if (n) mp_msg(MSGT_SUBREADER,MSGL_V,"SUB: Adjusted %d subtitle(s).\n", n);
1285 struct subreader {
1286 subtitle * (*read)(stream_t *st,subtitle *dest,int utf16);
1287 void (*post)(subtitle *dest);
1288 const char *name;
1291 #ifdef CONFIG_ENCA
1292 const char* guess_buffer_cp(unsigned char* buffer, int buflen, const char *preferred_language, const char *fallback)
1294 const char **languages;
1295 size_t langcnt;
1296 EncaAnalyser analyser;
1297 EncaEncoding encoding;
1298 const char *detected_sub_cp = NULL;
1299 int i;
1301 languages = enca_get_languages(&langcnt);
1302 mp_msg(MSGT_SUBREADER, MSGL_V, "ENCA supported languages: ");
1303 for (i = 0; i < langcnt; i++) {
1304 mp_msg(MSGT_SUBREADER, MSGL_V, "%s ", languages[i]);
1306 mp_msg(MSGT_SUBREADER, MSGL_V, "\n");
1308 for (i = 0; i < langcnt; i++) {
1309 if (strcasecmp(languages[i], preferred_language) != 0) continue;
1310 analyser = enca_analyser_alloc(languages[i]);
1311 encoding = enca_analyse_const(analyser, buffer, buflen);
1312 enca_analyser_free(analyser);
1313 if (encoding.charset != ENCA_CS_UNKNOWN) {
1314 detected_sub_cp = enca_charset_name(encoding.charset, ENCA_NAME_STYLE_ICONV);
1315 break;
1319 free(languages);
1321 if (!detected_sub_cp) {
1322 detected_sub_cp = fallback;
1323 mp_msg(MSGT_SUBREADER, MSGL_INFO, "ENCA detection failed: fallback to %s\n", fallback);
1324 }else{
1325 mp_msg(MSGT_SUBREADER, MSGL_INFO, "ENCA detected charset: %s\n", detected_sub_cp);
1328 return detected_sub_cp;
1331 #define MAX_GUESS_BUFFER_SIZE (256*1024)
1332 const char* guess_cp(stream_t *st, const char *preferred_language, const char *fallback)
1334 size_t buflen;
1335 unsigned char *buffer;
1336 const char *detected_sub_cp = NULL;
1338 buffer = malloc(MAX_GUESS_BUFFER_SIZE);
1339 buflen = stream_read(st,buffer, MAX_GUESS_BUFFER_SIZE);
1341 detected_sub_cp = guess_buffer_cp(buffer, buflen, preferred_language, fallback);
1343 free(buffer);
1344 stream_reset(st);
1345 stream_seek(st,0);
1347 return detected_sub_cp;
1349 #undef MAX_GUESS_BUFFER_SIZE
1350 #endif
1352 sub_data* sub_read_file (char *filename, float fps) {
1353 int utf16;
1354 stream_t* fd;
1355 int n_max, n_first, i, j, sub_first, sub_orig;
1356 subtitle *first, *second, *sub, *return_sub, *alloced_sub = NULL;
1357 sub_data *subt_data;
1358 int uses_time = 0, sub_num = 0, sub_errs = 0;
1359 static const struct subreader sr[]=
1361 { sub_read_line_microdvd, NULL, "microdvd" },
1362 { sub_read_line_subrip, NULL, "subrip" },
1363 { sub_read_line_subviewer, NULL, "subviewer" },
1364 { sub_read_line_sami, NULL, "sami" },
1365 { sub_read_line_vplayer, NULL, "vplayer" },
1366 { sub_read_line_rt, NULL, "rt" },
1367 { sub_read_line_ssa, sub_pp_ssa, "ssa" },
1368 { sub_read_line_pjs, NULL, "pjs" },
1369 { sub_read_line_mpsub, NULL, "mpsub" },
1370 { sub_read_line_aqt, NULL, "aqt" },
1371 { sub_read_line_subviewer2, NULL, "subviewer 2.0" },
1372 { sub_read_line_subrip09, NULL, "subrip 0.9" },
1373 { sub_read_line_jacosub, NULL, "jacosub" },
1374 { sub_read_line_mpl2, NULL, "mpl2" }
1376 const struct subreader *srp;
1378 if(filename==NULL) return NULL; //qnx segfault
1379 i = 0;
1380 fd=open_stream (filename, NULL, &i); if (!fd) return NULL;
1382 sub_format = SUB_INVALID;
1383 for (utf16 = 0; sub_format == SUB_INVALID && utf16 < 3; utf16++) {
1384 sub_format=sub_autodetect (fd, &uses_time, utf16);
1385 stream_reset(fd);
1386 stream_seek(fd,0);
1388 utf16--;
1390 mpsub_multiplier = (uses_time ? 100.0 : 1.0);
1391 if (sub_format==SUB_INVALID) {mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: Could not determine file format\n");return NULL;}
1392 srp=sr+sub_format;
1393 mp_msg(MSGT_SUBREADER, MSGL_V, "SUB: Detected subtitle file format: %s\n", srp->name);
1395 #ifdef CONFIG_ICONV
1396 sub_utf8_prev=sub_utf8;
1398 int l,k;
1399 k = -1;
1400 if ((l=strlen(filename))>4){
1401 char *exts[] = {".utf", ".utf8", ".utf-8" };
1402 for (k=3;--k>=0;)
1403 if (l >= strlen(exts[k]) && !strcasecmp(filename+(l - strlen(exts[k])), exts[k])){
1404 sub_utf8 = 1;
1405 break;
1408 if (k<0) subcp_open(fd);
1410 #endif
1412 sub_num=0;n_max=32;
1413 first=malloc(n_max*sizeof(subtitle));
1414 if(!first){
1415 #ifdef CONFIG_ICONV
1416 subcp_close();
1417 sub_utf8=sub_utf8_prev;
1418 #endif
1419 return NULL;
1422 #ifdef CONFIG_SORTSUB
1423 alloced_sub =
1424 sub = malloc(sizeof(subtitle));
1425 //This is to deal with those formats (AQT & Subrip) which define the end of a subtitle
1426 //as the beginning of the following
1427 previous_sub_end = 0;
1428 #endif
1429 while(1){
1430 if(sub_num>=n_max){
1431 n_max+=16;
1432 first=realloc(first,n_max*sizeof(subtitle));
1434 #ifndef CONFIG_SORTSUB
1435 sub = &first[sub_num];
1436 #endif
1437 memset(sub, '\0', sizeof(subtitle));
1438 sub=srp->read(fd,sub,utf16);
1439 if(!sub) break; // EOF
1440 #ifdef CONFIG_ICONV
1441 if ((sub!=ERR) && sub_utf8 == 2) sub=subcp_recode(sub);
1442 #endif
1443 #ifdef CONFIG_FRIBIDI
1444 if (sub!=ERR) sub=sub_fribidi(sub,sub_utf8,0);
1445 #endif
1446 if ( sub == ERR )
1448 #ifdef CONFIG_ICONV
1449 subcp_close();
1450 #endif
1451 if ( first ) free(first);
1452 free(alloced_sub);
1453 return NULL;
1455 // Apply any post processing that needs recoding first
1456 if ((sub!=ERR) && !sub_no_text_pp && srp->post) srp->post(sub);
1457 #ifdef CONFIG_SORTSUB
1458 if(!sub_num || (first[sub_num - 1].start <= sub->start)){
1459 first[sub_num].start = sub->start;
1460 first[sub_num].end = sub->end;
1461 first[sub_num].lines = sub->lines;
1462 first[sub_num].alignment = sub->alignment;
1463 for(i = 0; i < sub->lines; ++i){
1464 first[sub_num].text[i] = sub->text[i];
1466 if (previous_sub_end){
1467 first[sub_num - 1].end = previous_sub_end;
1468 previous_sub_end = 0;
1470 } else {
1471 for(j = sub_num - 1; j >= 0; --j){
1472 first[j + 1].start = first[j].start;
1473 first[j + 1].end = first[j].end;
1474 first[j + 1].lines = first[j].lines;
1475 first[j + 1].alignment = first[j].alignment;
1476 for(i = 0; i < first[j].lines; ++i){
1477 first[j + 1].text[i] = first[j].text[i];
1479 if(!j || (first[j - 1].start <= sub->start)){
1480 first[j].start = sub->start;
1481 first[j].end = sub->end;
1482 first[j].lines = sub->lines;
1483 first[j].alignment = sub->alignment;
1484 for(i = 0; i < SUB_MAX_TEXT; ++i){
1485 first[j].text[i] = sub->text[i];
1487 if (previous_sub_end){
1488 first[j].end = first[j - 1].end;
1489 first[j - 1].end = previous_sub_end;
1490 previous_sub_end = 0;
1492 break;
1496 #endif
1497 if(sub==ERR) ++sub_errs; else ++sub_num; // Error vs. Valid
1500 free_stream(fd);
1502 #ifdef CONFIG_ICONV
1503 subcp_close();
1504 #endif
1505 free(alloced_sub);
1507 // printf ("SUB: Subtitle format %s time.\n", uses_time?"uses":"doesn't use");
1508 mp_msg(MSGT_SUBREADER, MSGL_V,"SUB: Read %i subtitles, %i bad line(s).\n",
1509 sub_num, sub_errs);
1511 if(sub_num<=0){
1512 free(first);
1513 return NULL;
1516 // we do overlap if the user forced it (suboverlap_enable == 2) or
1517 // the user didn't forced no-overlapsub and the format is Jacosub or Ssa.
1518 // this is because usually overlapping subtitles are found in these formats,
1519 // while in others they are probably result of bad timing
1520 if ((suboverlap_enabled == 2) ||
1521 ((suboverlap_enabled) && ((sub_format == SUB_JACOSUB) || (sub_format == SUB_SSA)))) {
1522 adjust_subs_time(first, 6.0, fps, 0, sub_num, uses_time);/*~6 secs AST*/
1523 // here we manage overlapping subtitles
1524 sub_orig = sub_num;
1525 n_first = sub_num;
1526 sub_num = 0;
1527 second = NULL;
1528 // for each subtitle in first[] we deal with its 'block' of
1529 // bonded subtitles
1530 for (sub_first = 0; sub_first < n_first; ++sub_first) {
1531 unsigned long global_start = first[sub_first].start,
1532 global_end = first[sub_first].end, local_start, local_end;
1533 int lines_to_add = first[sub_first].lines, sub_to_add = 0,
1534 **placeholder = NULL, higher_line = 0, counter, start_block_sub = sub_num;
1535 char real_block = 1;
1537 // here we find the number of subtitles inside the 'block'
1538 // and its span interval. this works well only with sorted
1539 // subtitles
1540 while ((sub_first + sub_to_add + 1 < n_first) && (first[sub_first + sub_to_add + 1].start < global_end)) {
1541 ++sub_to_add;
1542 lines_to_add += first[sub_first + sub_to_add].lines;
1543 if (first[sub_first + sub_to_add].start < global_start) {
1544 global_start = first[sub_first + sub_to_add].start;
1546 if (first[sub_first + sub_to_add].end > global_end) {
1547 global_end = first[sub_first + sub_to_add].end;
1551 /* Avoid n^2 memory use for the "placeholder" data structure
1552 * below with subtitles that have a huge number of
1553 * consecutive overlapping lines. */
1554 lines_to_add = FFMIN(lines_to_add, SUB_MAX_TEXT);
1556 // we need a structure to keep trace of the screen lines
1557 // used by the subs, a 'placeholder'
1558 counter = 2 * sub_to_add + 1; // the maximum number of subs derived
1559 // from a block of sub_to_add+1 subs
1560 placeholder = malloc(sizeof(int *) * counter);
1561 for (i = 0; i < counter; ++i) {
1562 placeholder[i] = malloc(sizeof(int) * lines_to_add);
1563 for (j = 0; j < lines_to_add; ++j) {
1564 placeholder[i][j] = -1;
1568 counter = 0;
1569 local_end = global_start - 1;
1570 do {
1571 int ls;
1573 // here we find the beginning and the end of a new
1574 // subtitle in the block
1575 local_start = local_end + 1;
1576 local_end = global_end;
1577 for (j = 0; j <= sub_to_add; ++j) {
1578 if ((first[sub_first + j].start - 1 > local_start) && (first[sub_first + j].start - 1 < local_end)) {
1579 local_end = first[sub_first + j].start - 1;
1580 } else if ((first[sub_first + j].end > local_start) && (first[sub_first + j].end < local_end)) {
1581 local_end = first[sub_first + j].end;
1584 // here we allocate the screen lines to subs we must
1585 // display in current local_start-local_end interval.
1586 // if the subs were yet presents in the previous interval
1587 // they keep the same lines, otherside they get unused lines
1588 for (j = 0; j <= sub_to_add; ++j) {
1589 if ((first[sub_first + j].start <= local_end) && (first[sub_first + j].end > local_start)) {
1590 unsigned long sub_lines = first[sub_first + j].lines, fragment_length = lines_to_add + 1,
1591 tmp = 0;
1592 char boolean = 0;
1593 int fragment_position = -1;
1595 // if this is not the first new sub of the block
1596 // we find if this sub was present in the previous
1597 // new sub
1598 if (counter)
1599 for (i = 0; i < lines_to_add; ++i) {
1600 if (placeholder[counter - 1][i] == sub_first + j) {
1601 placeholder[counter][i] = sub_first + j;
1602 boolean = 1;
1605 if (boolean)
1606 continue;
1608 // we are looking for the shortest among all groups of
1609 // sequential blank lines whose length is greater than or
1610 // equal to sub_lines. we store in fragment_position the
1611 // position of the shortest group, in fragment_length its
1612 // length, and in tmp the length of the group currently
1613 // examinated
1614 for (i = 0; i < lines_to_add; ++i) {
1615 if (placeholder[counter][i] == -1) {
1616 // placeholder[counter][i] is part of the current group
1617 // of blank lines
1618 ++tmp;
1619 } else {
1620 if (tmp == sub_lines) {
1621 // current group's size fits exactly the one we
1622 // need, so we stop looking
1623 fragment_position = i - tmp;
1624 tmp = 0;
1625 break;
1627 if ((tmp) && (tmp > sub_lines) && (tmp < fragment_length)) {
1628 // current group is the best we found till here,
1629 // but is still bigger than the one we are looking
1630 // for, so we keep on looking
1631 fragment_length = tmp;
1632 fragment_position = i - tmp;
1633 tmp = 0;
1634 } else {
1635 // current group doesn't fit at all, so we forget it
1636 tmp = 0;
1640 if (tmp) {
1641 // last screen line is blank, a group ends with it
1642 if ((tmp >= sub_lines) && (tmp < fragment_length)) {
1643 fragment_position = i - tmp;
1646 if (fragment_position == -1) {
1647 // it was not possible to find free screen line(s) for a subtitle,
1648 // usually this means a bug in the code; however we do not overlap
1649 mp_msg(MSGT_SUBREADER, MSGL_WARN, "SUB: we could not find a suitable position for an overlapping subtitle\n");
1650 higher_line = SUB_MAX_TEXT + 1;
1651 break;
1652 } else {
1653 for (tmp = 0; tmp < sub_lines; ++tmp) {
1654 placeholder[counter][fragment_position + tmp] = sub_first + j;
1659 for (j = higher_line + 1; j < lines_to_add; ++j) {
1660 if (placeholder[counter][j] != -1)
1661 higher_line = j;
1662 else
1663 break;
1665 if (higher_line >= SUB_MAX_TEXT) {
1666 // the 'block' has too much lines, so we don't overlap the
1667 // subtitles
1668 second = realloc(second, (sub_num + sub_to_add + 1) * sizeof(subtitle));
1669 for (j = 0; j <= sub_to_add; ++j) {
1670 int ls;
1671 memset(&second[sub_num + j], '\0', sizeof(subtitle));
1672 second[sub_num + j].start = first[sub_first + j].start;
1673 second[sub_num + j].end = first[sub_first + j].end;
1674 second[sub_num + j].lines = first[sub_first + j].lines;
1675 second[sub_num + j].alignment = first[sub_first + j].alignment;
1676 for (ls = 0; ls < second[sub_num + j].lines; ls++) {
1677 second[sub_num + j].text[ls] = strdup(first[sub_first + j].text[ls]);
1680 sub_num += sub_to_add + 1;
1681 sub_first += sub_to_add;
1682 real_block = 0;
1683 break;
1686 // we read the placeholder structure and create the new
1687 // subs.
1688 second = realloc(second, (sub_num + 1) * sizeof(subtitle));
1689 memset(&second[sub_num], '\0', sizeof(subtitle));
1690 second[sub_num].start = local_start;
1691 second[sub_num].end = local_end;
1692 second[sub_num].alignment = first[sub_first].alignment;
1693 n_max = (lines_to_add < SUB_MAX_TEXT) ? lines_to_add : SUB_MAX_TEXT;
1694 for (i = 0, j = 0; j < n_max; ++j) {
1695 if (placeholder[counter][j] != -1) {
1696 int lines = first[placeholder[counter][j]].lines;
1697 for (ls = 0; ls < lines; ++ls) {
1698 second[sub_num].text[i++] = strdup(first[placeholder[counter][j]].text[ls]);
1700 j += lines - 1;
1701 } else {
1702 second[sub_num].text[i++] = strdup(" ");
1705 ++sub_num;
1706 ++counter;
1707 } while (local_end < global_end);
1708 if (real_block)
1709 for (i = 0; i < counter; ++i)
1710 second[start_block_sub + i].lines = higher_line + 1;
1712 counter = 2 * sub_to_add + 1;
1713 for (i = 0; i < counter; ++i) {
1714 free(placeholder[i]);
1716 free(placeholder);
1717 sub_first += sub_to_add;
1720 for (j = sub_orig - 1; j >= 0; --j) {
1721 for (i = first[j].lines - 1; i >= 0; --i) {
1722 free(first[j].text[i]);
1725 free(first);
1727 return_sub = second;
1728 } else { //if(suboverlap_enabled)
1729 adjust_subs_time(first, 6.0, fps, 1, sub_num, uses_time);/*~6 secs AST*/
1730 return_sub = first;
1732 if (return_sub == NULL) return NULL;
1733 subt_data = malloc(sizeof(sub_data));
1734 subt_data->filename = strdup(filename);
1735 subt_data->sub_uses_time = uses_time;
1736 subt_data->sub_num = sub_num;
1737 subt_data->sub_errs = sub_errs;
1738 subt_data->subtitles = return_sub;
1739 return subt_data;
1742 #if 0
1743 char * strreplace( char * in,char * what,char * whereof )
1745 int i;
1746 char * tmp;
1748 if ( ( in == NULL )||( what == NULL )||( whereof == NULL )||( ( tmp=strstr( in,what ) ) == NULL ) ) return NULL;
1749 for( i=0;i<strlen( whereof );i++ ) tmp[i]=whereof[i];
1750 if ( strlen( what ) > strlen( whereof ) ) tmp[i]=0;
1751 return in;
1753 #endif
1756 static void strcpy_trim(char *d, char *s)
1758 // skip leading whitespace
1759 while (*s && isspace(*s)) {
1760 s++;
1762 for (;;) {
1763 // copy word
1764 while (*s && !isspace(*s)) {
1765 *d = tolower(*s);
1766 s++; d++;
1768 if (*s == 0) break;
1769 // trim excess whitespace
1770 while (*s && isspace(*s)) {
1771 s++;
1773 if (*s == 0) break;
1774 *d++ = ' ';
1776 *d = 0;
1779 static void strcpy_strip_ext(char *d, char *s)
1781 char *tmp = strrchr(s,'.');
1782 if (!tmp) {
1783 strcpy(d, s);
1784 return;
1785 } else {
1786 strncpy(d, s, tmp-s);
1787 d[tmp-s] = 0;
1789 while (*d) {
1790 *d = tolower(*d);
1791 d++;
1795 static void strcpy_get_ext(char *d, char *s)
1797 char *tmp = strrchr(s,'.');
1798 if (!tmp) {
1799 strcpy(d, "");
1800 return;
1801 } else {
1802 strcpy(d, tmp+1);
1806 static int whiteonly(char *s)
1808 while (*s) {
1809 if (!isspace(*s)) return 0;
1810 s++;
1812 return 1;
1815 typedef struct subfn
1817 int priority;
1818 char *fname;
1819 } subfn;
1821 static int compare_sub_priority(const void *a, const void *b)
1823 if (((const subfn*)a)->priority > ((const subfn*)b)->priority) {
1824 return -1;
1825 } else if (((const subfn*)a)->priority < ((const subfn*)b)->priority) {
1826 return 1;
1827 } else {
1828 return strcoll(((const subfn*)a)->fname, ((const subfn*)b)->fname);
1832 char** sub_filenames(const char* path, char *fname)
1834 char *f_dir, *f_fname, *f_fname_noext, *f_fname_trim, *tmp, *tmp_sub_id;
1835 char *tmp_fname_noext, *tmp_fname_trim, *tmp_fname_ext, *tmpresult;
1837 int len, pos, found, i, j;
1838 char * sub_exts[] = { "utf", "utf8", "utf-8", "sub", "srt", "smi", "rt", "txt", "ssa", "aqt", "jss", "js", "ass", NULL};
1839 subfn *result;
1840 char **result2;
1842 int subcnt;
1844 FILE *f;
1846 DIR *d;
1847 struct dirent *de;
1849 len = (strlen(fname) > 256 ? strlen(fname) : 256)
1850 +(strlen(path) > 256 ? strlen(path) : 256)+2;
1852 f_dir = malloc(len);
1853 f_fname = malloc(len);
1854 f_fname_noext = malloc(len);
1855 f_fname_trim = malloc(len);
1857 tmp_fname_noext = malloc(len);
1858 tmp_fname_trim = malloc(len);
1859 tmp_fname_ext = malloc(len);
1861 tmpresult = malloc(len);
1863 result = malloc(sizeof(subfn)*MAX_SUBTITLE_FILES);
1864 memset(result, 0, sizeof(subfn)*MAX_SUBTITLE_FILES);
1866 subcnt = 0;
1868 tmp = strrchr(fname,'/');
1869 #if HAVE_DOS_PATHS
1870 if(!tmp)tmp = strrchr(fname,'\\');
1871 if(!tmp)tmp = strrchr(fname,':');
1872 #endif
1874 // extract filename & dirname from fname
1875 if (tmp) {
1876 strcpy(f_fname, tmp+1);
1877 pos = tmp - fname;
1878 strncpy(f_dir, fname, pos+1);
1879 f_dir[pos+1] = 0;
1880 } else {
1881 strcpy(f_fname, fname);
1882 strcpy(f_dir, "./");
1885 strcpy_strip_ext(f_fname_noext, f_fname);
1886 strcpy_trim(f_fname_trim, f_fname_noext);
1888 tmp_sub_id = NULL;
1889 if (dvdsub_lang && !whiteonly(dvdsub_lang)) {
1890 tmp_sub_id = malloc(strlen(dvdsub_lang)+1);
1891 strcpy_trim(tmp_sub_id, dvdsub_lang);
1894 // 0 = nothing
1895 // 1 = any subtitle file
1896 // 2 = any sub file containing movie name
1897 // 3 = sub file containing movie name and the lang extension
1898 for (j = 0; j <= 1; j++) {
1899 d = opendir(j == 0 ? f_dir : path);
1900 if (d) {
1901 while ((de = readdir(d))) {
1902 // retrieve various parts of the filename
1903 strcpy_strip_ext(tmp_fname_noext, de->d_name);
1904 strcpy_get_ext(tmp_fname_ext, de->d_name);
1905 strcpy_trim(tmp_fname_trim, tmp_fname_noext);
1907 // does it end with a subtitle extension?
1908 found = 0;
1909 #ifdef CONFIG_ICONV
1910 #ifdef CONFIG_ENCA
1911 for (i = ((sub_cp && strncasecmp(sub_cp, "enca", 4) != 0) ? 3 : 0); sub_exts[i]; i++) {
1912 #else
1913 for (i = (sub_cp ? 3 : 0); sub_exts[i]; i++) {
1914 #endif
1915 #else
1916 for (i = 0; sub_exts[i]; i++) {
1917 #endif
1918 if (strcasecmp(sub_exts[i], tmp_fname_ext) == 0) {
1919 found = 1;
1920 break;
1924 // we have a (likely) subtitle file
1925 if (found) {
1926 int prio = 0;
1927 if (!prio && tmp_sub_id)
1929 sprintf(tmpresult, "%s %s", f_fname_trim, tmp_sub_id);
1930 mp_msg(MSGT_SUBREADER,MSGL_INFO,"dvdsublang...%s\n", tmpresult);
1931 if (strcmp(tmp_fname_trim, tmpresult) == 0 && sub_match_fuzziness >= 1) {
1932 // matches the movie name + lang extension
1933 prio = 5;
1936 if (!prio && strcmp(tmp_fname_trim, f_fname_trim) == 0) {
1937 // matches the movie name
1938 prio = 4;
1940 if (!prio && (tmp = strstr(tmp_fname_trim, f_fname_trim)) && (sub_match_fuzziness >= 1)) {
1941 // contains the movie name
1942 tmp += strlen(f_fname_trim);
1943 if (tmp_sub_id && strstr(tmp, tmp_sub_id)) {
1944 // with sub_id specified prefer localized subtitles
1945 prio = 3;
1946 } else if ((tmp_sub_id == NULL) && whiteonly(tmp)) {
1947 // without sub_id prefer "plain" name
1948 prio = 3;
1949 } else {
1950 // with no localized subs found, try any else instead
1951 prio = 2;
1954 if (!prio) {
1955 // doesn't contain the movie name
1956 // don't try in the mplayer subtitle directory
1957 if ((j == 0) && (sub_match_fuzziness >= 2)) {
1958 prio = 1;
1962 if (prio) {
1963 prio += prio;
1964 #ifdef CONFIG_ICONV
1965 if (i<3){ // prefer UTF-8 coded
1966 prio++;
1968 #endif
1969 sprintf(tmpresult, "%s%s", j == 0 ? f_dir : path, de->d_name);
1970 // fprintf(stderr, "%s priority %d\n", tmpresult, prio);
1971 if ((f = fopen(tmpresult, "rt"))) {
1972 fclose(f);
1973 result[subcnt].priority = prio;
1974 result[subcnt].fname = strdup(tmpresult);
1975 subcnt++;
1980 if (subcnt >= MAX_SUBTITLE_FILES) break;
1982 closedir(d);
1987 if (tmp_sub_id) free(tmp_sub_id);
1989 free(f_dir);
1990 free(f_fname);
1991 free(f_fname_noext);
1992 free(f_fname_trim);
1994 free(tmp_fname_noext);
1995 free(tmp_fname_trim);
1996 free(tmp_fname_ext);
1998 free(tmpresult);
2000 qsort(result, subcnt, sizeof(subfn), compare_sub_priority);
2002 result2 = malloc(sizeof(char*)*(subcnt+1));
2003 memset(result2, 0, sizeof(char*)*(subcnt+1));
2005 for (i = 0; i < subcnt; i++) {
2006 result2[i] = result[i].fname;
2008 result2[subcnt] = NULL;
2010 free(result);
2012 return result2;
2015 void list_sub_file(sub_data* subd){
2016 int i,j;
2017 subtitle *subs = subd->subtitles;
2019 for(j=0; j < subd->sub_num; j++){
2020 subtitle* egysub=&subs[j];
2021 mp_msg(MSGT_SUBREADER,MSGL_INFO,"%i line%c (%li-%li)\n",
2022 egysub->lines,
2023 (1==egysub->lines)?' ':'s',
2024 egysub->start,
2025 egysub->end);
2026 for (i=0; i<egysub->lines; i++) {
2027 mp_msg(MSGT_SUBREADER,MSGL_INFO,"\t\t%d: %s%s", i,egysub->text[i], i==egysub->lines-1?"":" \n ");
2029 mp_msg(MSGT_SUBREADER,MSGL_INFO,"\n");
2032 mp_msg(MSGT_SUBREADER,MSGL_INFO,"Subtitle format %s time.\n",
2033 subd->sub_uses_time ? "uses":"doesn't use");
2034 mp_msg(MSGT_SUBREADER,MSGL_INFO,"Read %i subtitles, %i errors.\n", subd->sub_num, subd->sub_errs);
2037 void dump_srt(sub_data* subd, float fps){
2038 int i,j;
2039 int h,m,s,ms;
2040 FILE * fd;
2041 subtitle * onesub;
2042 unsigned long temp;
2043 subtitle *subs = subd->subtitles;
2045 if (!subd->sub_uses_time && sub_fps == 0)
2046 sub_fps = fps;
2047 fd=fopen("dumpsub.srt","w");
2048 if(!fd)
2050 perror("dump_srt: fopen");
2051 return;
2053 for(i=0; i < subd->sub_num; i++)
2055 onesub=subs+i; //=&subs[i];
2056 fprintf(fd,"%d\n",i+1);//line number
2058 temp=onesub->start;
2059 if (!subd->sub_uses_time)
2060 temp = temp * 100 / sub_fps;
2061 temp -= sub_delay * 100;
2062 h=temp/360000;temp%=360000; //h =1*100*60*60
2063 m=temp/6000; temp%=6000; //m =1*100*60
2064 s=temp/100; temp%=100; //s =1*100
2065 ms=temp*10; //ms=1*10
2066 fprintf(fd,"%02d:%02d:%02d,%03d --> ",h,m,s,ms);
2068 temp=onesub->end;
2069 if (!subd->sub_uses_time)
2070 temp = temp * 100 / sub_fps;
2071 temp -= sub_delay * 100;
2072 h=temp/360000;temp%=360000;
2073 m=temp/6000; temp%=6000;
2074 s=temp/100; temp%=100;
2075 ms=temp*10;
2076 fprintf(fd,"%02d:%02d:%02d,%03d\n",h,m,s,ms);
2078 for(j=0;j<onesub->lines;j++)
2079 fprintf(fd,"%s\n",onesub->text[j]);
2081 fprintf(fd,"\n");
2083 fclose(fd);
2084 mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.srt\'.\n");
2087 void dump_mpsub(sub_data* subd, float fps){
2088 int i,j;
2089 FILE *fd;
2090 float a,b;
2091 subtitle *subs = subd->subtitles;
2093 mpsub_position = subd->sub_uses_time? (sub_delay*100) : (sub_delay*fps);
2094 if (sub_fps==0) sub_fps=fps;
2096 fd=fopen ("dump.mpsub", "w");
2097 if (!fd) {
2098 perror ("dump_mpsub: fopen");
2099 return;
2103 if (subd->sub_uses_time) fprintf (fd,"FORMAT=TIME\n\n");
2104 else fprintf (fd, "FORMAT=%5.2f\n\n", fps);
2106 for(j=0; j < subd->sub_num; j++){
2107 subtitle* egysub=&subs[j];
2108 if (subd->sub_uses_time) {
2109 a=((egysub->start-mpsub_position)/100.0);
2110 b=((egysub->end-egysub->start)/100.0);
2111 if ( (float)((int)a) == a)
2112 fprintf (fd, "%.0f",a);
2113 else
2114 fprintf (fd, "%.2f",a);
2116 if ( (float)((int)b) == b)
2117 fprintf (fd, " %.0f\n",b);
2118 else
2119 fprintf (fd, " %.2f\n",b);
2120 } else {
2121 fprintf (fd, "%ld %ld\n", (long)((egysub->start*(fps/sub_fps))-((mpsub_position*(fps/sub_fps)))),
2122 (long)(((egysub->end)-(egysub->start))*(fps/sub_fps)));
2125 mpsub_position = egysub->end;
2126 for (i=0; i<egysub->lines; i++) {
2127 fprintf (fd, "%s\n",egysub->text[i]);
2129 fprintf (fd, "\n");
2131 fclose (fd);
2132 mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dump.mpsub\'.\n");
2135 void dump_microdvd(sub_data* subd, float fps) {
2136 int i, delay;
2137 FILE *fd;
2138 subtitle *subs = subd->subtitles;
2139 if (sub_fps == 0)
2140 sub_fps = fps;
2141 fd = fopen("dumpsub.sub", "w");
2142 if (!fd) {
2143 perror("dumpsub.sub: fopen");
2144 return;
2146 delay = sub_delay * sub_fps;
2147 for (i = 0; i < subd->sub_num; ++i) {
2148 int j, start, end;
2149 start = subs[i].start;
2150 end = subs[i].end;
2151 if (subd->sub_uses_time) {
2152 start = start * sub_fps / 100 ;
2153 end = end * sub_fps / 100;
2155 else {
2156 start = start * sub_fps / fps;
2157 end = end * sub_fps / fps;
2159 start -= delay;
2160 end -= delay;
2161 fprintf(fd, "{%d}{%d}", start, end);
2162 for (j = 0; j < subs[i].lines; ++j)
2163 fprintf(fd, "%s%s", j ? "|" : "", subs[i].text[j]);
2164 fprintf(fd, "\n");
2166 fclose(fd);
2167 mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.sub\'.\n");
2170 void dump_jacosub(sub_data* subd, float fps) {
2171 int i,j;
2172 int h,m,s,cs;
2173 FILE * fd;
2174 subtitle * onesub;
2175 unsigned long temp;
2176 subtitle *subs = subd->subtitles;
2178 if (!subd->sub_uses_time && sub_fps == 0)
2179 sub_fps = fps;
2180 fd=fopen("dumpsub.jss","w");
2181 if(!fd)
2183 perror("dump_jacosub: fopen");
2184 return;
2186 fprintf(fd, "#TIMERES %d\n", (subd->sub_uses_time) ? 100 : (int)sub_fps);
2187 for(i=0; i < subd->sub_num; i++)
2189 onesub=subs+i; //=&subs[i];
2191 temp=onesub->start;
2192 if (!subd->sub_uses_time)
2193 temp = temp * 100 / sub_fps;
2194 temp -= sub_delay * 100;
2195 h=temp/360000;temp%=360000; //h =1*100*60*60
2196 m=temp/6000; temp%=6000; //m =1*100*60
2197 s=temp/100; temp%=100; //s =1*100
2198 cs=temp; //cs=1*10
2199 fprintf(fd,"%02d:%02d:%02d.%02d ",h,m,s,cs);
2201 temp=onesub->end;
2202 if (!subd->sub_uses_time)
2203 temp = temp * 100 / sub_fps;
2204 temp -= sub_delay * 100;
2205 h=temp/360000;temp%=360000;
2206 m=temp/6000; temp%=6000;
2207 s=temp/100; temp%=100;
2208 cs=temp;
2209 fprintf(fd,"%02d:%02d:%02d.%02d {~} ",h,m,s,cs);
2211 for(j=0;j<onesub->lines;j++)
2212 fprintf(fd,"%s%s",j ? "\\n" : "", onesub->text[j]);
2214 fprintf(fd,"\n");
2216 fclose(fd);
2217 mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.js\'.\n");
2220 void dump_sami(sub_data* subd, float fps) {
2221 int i,j;
2222 FILE * fd;
2223 subtitle * onesub;
2224 unsigned long temp;
2225 subtitle *subs = subd->subtitles;
2227 if (!subd->sub_uses_time && sub_fps == 0)
2228 sub_fps = fps;
2229 fd=fopen("dumpsub.smi","w");
2230 if(!fd)
2232 perror("dump_jacosub: fopen");
2233 return;
2235 fprintf(fd, "<SAMI>\n"
2236 "<HEAD>\n"
2237 " <STYLE TYPE=\"Text/css\">\n"
2238 " <!--\n"
2239 " 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"
2240 " .SUBTTL {Name: 'Subtitles'; Lang: en-US; SAMIType: CC;}\n"
2241 " -->\n"
2242 " </STYLE>\n"
2243 "</HEAD>\n"
2244 "<BODY>\n");
2245 for(i=0; i < subd->sub_num; i++)
2247 onesub=subs+i; //=&subs[i];
2249 temp=onesub->start;
2250 if (!subd->sub_uses_time)
2251 temp = temp * 100 / sub_fps;
2252 temp -= sub_delay * 100;
2253 fprintf(fd,"\t<SYNC Start=%lu>\n"
2254 "\t <P>", temp * 10);
2256 for(j=0;j<onesub->lines;j++)
2257 fprintf(fd,"%s%s",j ? "<br>" : "", onesub->text[j]);
2259 fprintf(fd,"\n");
2261 temp=onesub->end;
2262 if (!subd->sub_uses_time)
2263 temp = temp * 100 / sub_fps;
2264 temp -= sub_delay * 100;
2265 fprintf(fd,"\t<SYNC Start=%lu>\n"
2266 "\t <P>&nbsp;\n", temp * 10);
2268 fprintf(fd, "</BODY>\n"
2269 "</SAMI>\n");
2270 fclose(fd);
2271 mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.smi\'.\n");
2274 void sub_free( sub_data * subd )
2276 int i, j;
2278 if ( !subd ) return;
2280 for (i = 0; i < subd->sub_num; i++)
2281 for (j = 0; j < subd->subtitles[i].lines; j++)
2282 free( subd->subtitles[i].text[j] );
2283 free( subd->subtitles );
2284 free( subd->filename );
2285 free( subd );
2288 #define MAX_SUBLINE 512
2290 * \brief parse text and append it to subtitle in sub
2291 * \param sub subtitle struct to add text to
2292 * \param txt text to parse
2293 * \param len length of text in txt
2294 * \param endpts pts at which this subtitle text should be removed again
2296 * <> and {} are interpreted as comment delimiters, "\n", "\N", '\n', '\r'
2297 * and '\0' are interpreted as newlines, duplicate, leading and trailing
2298 * newlines are ignored.
2300 void sub_add_text(subtitle *sub, const char *txt, int len, double endpts) {
2301 int comment = 0;
2302 int double_newline = 1; // ignore newlines at the beginning
2303 int i, pos;
2304 char *buf;
2305 int orig_lines = sub->lines;
2306 if (sub->lines >= SUB_MAX_TEXT) return;
2307 pos = 0;
2308 buf = malloc(MAX_SUBLINE + 1);
2309 sub->text[sub->lines] = buf;
2310 sub->endpts[sub->lines] = endpts;
2311 for (i = 0; i < len && pos < MAX_SUBLINE; i++) {
2312 char c = txt[i];
2313 if (c == '<') comment |= 1;
2314 if (c == '{') comment |= 2;
2315 if (comment) {
2316 if (c == '}') comment &= ~2;
2317 if (c == '>') comment &= ~1;
2318 continue;
2320 if (pos == MAX_SUBLINE - 1) {
2321 i--;
2322 c = 0;
2324 if (c == '\\' && i + 1 < len) {
2325 c = txt[++i];
2326 if (c == 'n' || c == 'N') c = 0;
2328 if (c == '\n' || c == '\r') c = 0;
2329 if (c) {
2330 double_newline = 0;
2331 buf[pos++] = c;
2332 } else if (!double_newline) {
2333 if (sub->lines >= SUB_MAX_TEXT - 1) {
2334 mp_msg(MSGT_VO, MSGL_WARN, "Too many subtitle lines\n");
2335 break;
2337 double_newline = 1;
2338 buf[pos] = 0;
2339 sub->lines++;
2340 pos = 0;
2341 buf = malloc(MAX_SUBLINE + 1);
2342 sub->text[sub->lines] = buf;
2343 sub->endpts[sub->lines] = endpts;
2346 buf[pos] = 0;
2347 if (sub->lines < SUB_MAX_TEXT &&
2348 strlen(sub->text[sub->lines]))
2349 sub->lines++;
2350 #ifdef CONFIG_FRIBIDI
2351 sub = sub_fribidi(sub, sub_utf8, orig_lines);
2352 #endif
2355 #define MP_NOPTS_VALUE (-1LL<<63)
2357 * \brief remove outdated subtitle lines.
2358 * \param sub subtitle struct to modify
2359 * \param pts current pts. All lines with endpts <= this will be removed.
2360 * Use MP_NOPTS_VALUE to remove all lines
2361 * \return 1 if sub was modified, 0 otherwise.
2363 int sub_clear_text(subtitle *sub, double pts) {
2364 int i = 0;
2365 int changed = 0;
2366 while (i < sub->lines) {
2367 double endpts = sub->endpts[i];
2368 if (pts == MP_NOPTS_VALUE || (endpts != MP_NOPTS_VALUE && pts >= endpts)) {
2369 int j;
2370 free(sub->text[i]);
2371 for (j = i + 1; j < sub->lines; j++) {
2372 sub->text[j - 1] = sub->text[j];
2373 sub->endpts[j - 1] = sub->endpts[j];
2375 sub->lines--;
2376 changed = 1;
2377 } else
2378 i++;
2380 return changed;