Merge svn changes up to r31303
[mplayer/glamo.git] / subreader.c
blobf2b52823c271082fac50adf7cf6c2b41e17af992
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 /* Maximal length of line of a subtitle */
57 #define LINE_LEN 1000
58 static float mpsub_position=0;
59 static float mpsub_multiplier=1.;
60 static int sub_slacktime = 20000; //20 sec
62 int sub_no_text_pp=0; // 1 => do not apply text post-processing
63 // like {\...} elimination in SSA format.
65 int sub_match_fuzziness=0; // level of sub name matching fuzziness
67 /* Use the SUB_* constant defined in the header file */
68 int sub_format=SUB_INVALID;
69 #ifdef CONFIG_SORTSUB
71 Some subtitling formats, namely AQT and Subrip09, define the end of a
72 subtitle as the beginning of the following. Since currently we read one
73 subtitle at time, for these format we keep two global *subtitle,
74 previous_aqt_sub and previous_subrip09_sub, pointing to previous subtitle,
75 so we can change its end when we read current subtitle starting time.
76 When CONFIG_SORTSUB is defined, we use a single global unsigned long,
77 previous_sub_end, for both (and even future) formats, to store the end of
78 the previous sub: it is initialized to 0 in sub_read_file and eventually
79 modified by sub_read_aqt_line or sub_read_subrip09_line.
81 unsigned long previous_sub_end;
82 #endif
84 static int eol(char p) {
85 return p=='\r' || p=='\n' || p=='\0';
88 /* Remove leading and trailing space */
89 static void trail_space(char *s) {
90 int i = 0;
91 while (isspace(s[i])) ++i;
92 if (i) strcpy(s, s + i);
93 i = strlen(s) - 1;
94 while (i > 0 && isspace(s[i])) s[i--] = '\0';
97 static char *stristr(const char *haystack, const char *needle) {
98 int len = 0;
99 const char *p = haystack;
101 if (!(haystack && needle)) return NULL;
103 len=strlen(needle);
104 while (*p != '\0') {
105 if (strncasecmp(p, needle, len) == 0) return (char*)p;
106 p++;
109 return NULL;
112 static subtitle *sub_read_line_sami(stream_t* st, subtitle *current, int utf16) {
113 static char line[LINE_LEN+1];
114 static char *s = NULL, *slacktime_s;
115 char text[LINE_LEN+1], *p=NULL, *q;
116 int state;
118 current->lines = current->start = current->end = 0;
119 current->alignment = SUB_ALIGNMENT_BOTTOMCENTER;
120 state = 0;
122 /* read the first line */
123 if (!s)
124 if (!(s = stream_read_line(st, line, LINE_LEN, utf16))) return 0;
126 do {
127 switch (state) {
129 case 0: /* find "START=" or "Slacktime:" */
130 slacktime_s = stristr (s, "Slacktime:");
131 if (slacktime_s)
132 sub_slacktime = strtol (slacktime_s+10, NULL, 0) / 10;
134 s = stristr (s, "Start=");
135 if (s) {
136 current->start = strtol (s + 6, &s, 0) / 10;
137 /* eat '>' */
138 for (; *s != '>' && *s != '\0'; s++);
139 s++;
140 state = 1; continue;
142 break;
144 case 1: /* find (optionnal) "<P", skip other TAGs */
145 for (; *s == ' ' || *s == '\t'; s++); /* strip blanks, if any */
146 if (*s == '\0') break;
147 if (*s != '<') { state = 3; p = text; continue; } /* not a TAG */
148 s++;
149 if (*s == 'P' || *s == 'p') { s++; state = 2; continue; } /* found '<P' */
150 for (; *s != '>' && *s != '\0'; s++); /* skip remains of non-<P> TAG */
151 if (s == '\0')
152 break;
153 s++;
154 continue;
156 case 2: /* find ">" */
157 if ((s = strchr (s, '>'))) { s++; state = 3; p = text; continue; }
158 break;
160 case 3: /* get all text until '<' appears */
161 if (*s == '\0') break;
162 else if (!strncasecmp (s, "<br>", 4)) {
163 *p = '\0'; p = text; trail_space (text);
164 if (text[0] != '\0')
165 current->text[current->lines++] = strdup (text);
166 s += 4;
168 else if ((*s == '{') && !sub_no_text_pp) { state = 5; ++s; continue; }
169 else if (*s == '<') { state = 4; }
170 else if (!strncasecmp (s, "&nbsp;", 6)) { *p++ = ' '; s += 6; }
171 else if (*s == '\t') { *p++ = ' '; s++; }
172 else if (*s == '\r' || *s == '\n') { s++; }
173 else *p++ = *s++;
175 /* skip duplicated space */
176 if (p > text + 2) if (*(p-1) == ' ' && *(p-2) == ' ') p--;
178 continue;
180 case 4: /* get current->end or skip <TAG> */
181 q = stristr (s, "Start=");
182 if (q) {
183 current->end = strtol (q + 6, &q, 0) / 10 - 1;
184 *p = '\0'; trail_space (text);
185 if (text[0] != '\0')
186 current->text[current->lines++] = strdup (text);
187 if (current->lines > 0) { state = 99; break; }
188 state = 0; continue;
190 s = strchr (s, '>');
191 if (s) { s++; state = 3; continue; }
192 break;
193 case 5: /* get rid of {...} text, but read the alignment code */
194 if ((*s == '\\') && (*(s + 1) == 'a') && !sub_no_text_pp) {
195 if (stristr(s, "\\a1") != NULL) {
196 current->alignment = SUB_ALIGNMENT_BOTTOMLEFT;
197 s = s + 3;
199 if (stristr(s, "\\a2") != NULL) {
200 current->alignment = SUB_ALIGNMENT_BOTTOMCENTER;
201 s = s + 3;
202 } else if (stristr(s, "\\a3") != NULL) {
203 current->alignment = SUB_ALIGNMENT_BOTTOMRIGHT;
204 s = s + 3;
205 } else if ((stristr(s, "\\a4") != NULL) || (stristr(s, "\\a5") != NULL) || (stristr(s, "\\a8") != NULL)) {
206 current->alignment = SUB_ALIGNMENT_TOPLEFT;
207 s = s + 3;
208 } else if (stristr(s, "\\a6") != NULL) {
209 current->alignment = SUB_ALIGNMENT_TOPCENTER;
210 s = s + 3;
211 } else if (stristr(s, "\\a7") != NULL) {
212 current->alignment = SUB_ALIGNMENT_TOPRIGHT;
213 s = s + 3;
214 } else if (stristr(s, "\\a9") != NULL) {
215 current->alignment = SUB_ALIGNMENT_MIDDLELEFT;
216 s = s + 3;
217 } else if (stristr(s, "\\a10") != NULL) {
218 current->alignment = SUB_ALIGNMENT_MIDDLECENTER;
219 s = s + 4;
220 } else if (stristr(s, "\\a11") != NULL) {
221 current->alignment = SUB_ALIGNMENT_MIDDLERIGHT;
222 s = s + 4;
225 if (*s == '}') state = 3;
226 ++s;
227 continue;
230 /* read next line */
231 if (state != 99 && !(s = stream_read_line (st, line, LINE_LEN, utf16))) {
232 if (current->start > 0) {
233 break; // if it is the last subtitle
234 } else {
235 return 0;
239 } while (state != 99);
241 // For the last subtitle
242 if (current->end <= 0) {
243 current->end = current->start + sub_slacktime;
244 *p = '\0'; trail_space (text);
245 if (text[0] != '\0')
246 current->text[current->lines++] = strdup (text);
249 return current;
253 static char *sub_readtext(char *source, char **dest) {
254 int len=0;
255 char *p=source;
257 // printf("src=%p dest=%p \n",source,dest);
259 while ( !eol(*p) && *p!= '|' ) {
260 p++,len++;
263 *dest= malloc (len+1);
264 if (!dest) {return ERR;}
266 strncpy(*dest, source, len);
267 (*dest)[len]=0;
269 while (*p=='\r' || *p=='\n' || *p=='|') p++;
271 if (*p) return p; // not-last text field
272 else return NULL; // last text field
275 static subtitle *sub_read_line_microdvd(stream_t *st,subtitle *current, int utf16) {
276 char line[LINE_LEN+1];
277 char line2[LINE_LEN+1];
278 char *p, *next;
279 int i;
281 do {
282 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
283 } while ((sscanf (line,
284 "{%ld}{}%[^\r\n]",
285 &(current->start), line2) < 2) &&
286 (sscanf (line,
287 "{%ld}{%ld}%[^\r\n]",
288 &(current->start), &(current->end), line2) < 3));
290 p=line2;
292 next=p, i=0;
293 while ((next =sub_readtext (next, &(current->text[i])))) {
294 if (current->text[i]==ERR) {return ERR;}
295 i++;
296 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
298 current->lines= ++i;
300 return current;
303 static subtitle *sub_read_line_mpl2(stream_t *st,subtitle *current, int utf16) {
304 char line[LINE_LEN+1];
305 char line2[LINE_LEN+1];
306 char *p, *next;
307 int i;
309 do {
310 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
311 } while ((sscanf (line,
312 "[%ld][%ld]%[^\r\n]",
313 &(current->start), &(current->end), line2) < 3));
314 current->start *= 10;
315 current->end *= 10;
316 p=line2;
318 next=p, i=0;
319 while ((next =sub_readtext (next, &(current->text[i])))) {
320 if (current->text[i]==ERR) {return ERR;}
321 i++;
322 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
324 current->lines= ++i;
326 return current;
329 static subtitle *sub_read_line_subrip(stream_t* st, subtitle *current, int utf16) {
330 char line[LINE_LEN+1];
331 int a1,a2,a3,a4,b1,b2,b3,b4;
332 char *p=NULL, *q=NULL;
333 int len;
335 while (1) {
336 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
337 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4) < 8) continue;
338 current->start = a1*360000+a2*6000+a3*100+a4;
339 current->end = b1*360000+b2*6000+b3*100+b4;
341 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
343 p=q=line;
344 for (current->lines=1; current->lines < SUB_MAX_TEXT; current->lines++) {
345 for (q=p,len=0; *p && *p!='\r' && *p!='\n' && *p!='|' && strncmp(p,"[br]",4); p++,len++);
346 current->text[current->lines-1]=malloc (len+1);
347 if (!current->text[current->lines-1]) return ERR;
348 strncpy (current->text[current->lines-1], q, len);
349 current->text[current->lines-1][len]='\0';
350 if (!*p || *p=='\r' || *p=='\n') break;
351 if (*p=='|') p++;
352 else while (*p++!=']');
354 break;
356 return current;
359 static subtitle *sub_read_line_subviewer(stream_t *st,subtitle *current, int utf16) {
360 char line[LINE_LEN+1];
361 int a1,a2,a3,a4,b1,b2,b3,b4;
362 char *p=NULL;
363 int i,len;
365 while (!current->text[0]) {
366 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
367 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)
368 continue;
369 current->start = a1*360000+a2*6000+a3*100+a4/10;
370 current->end = b1*360000+b2*6000+b3*100+b4/10;
371 for (i=0; i<SUB_MAX_TEXT;) {
372 int blank = 1;
373 if (!stream_read_line (st, line, LINE_LEN, utf16)) break;
374 len=0;
375 for (p=line; *p!='\n' && *p!='\r' && *p; p++,len++)
376 if (*p != ' ' && *p != '\t')
377 blank = 0;
378 if (len && !blank) {
379 int j=0,skip=0;
380 char *curptr=current->text[i]=malloc (len+1);
381 if (!current->text[i]) return ERR;
382 //strncpy (current->text[i], line, len); current->text[i][len]='\0';
383 for(; j<len; j++) {
384 /* let's filter html tags ::atmos */
385 if(line[j]=='>') {
386 skip=0;
387 continue;
389 if(line[j]=='<') {
390 skip=1;
391 continue;
393 if(skip) {
394 continue;
396 *curptr=line[j];
397 curptr++;
399 *curptr='\0';
401 i++;
402 } else {
403 break;
406 current->lines=i;
408 return current;
411 static subtitle *sub_read_line_subviewer2(stream_t *st,subtitle *current, int utf16) {
412 char line[LINE_LEN+1];
413 int a1,a2,a3,a4;
414 char *p=NULL;
415 int i,len;
417 while (!current->text[0]) {
418 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
419 if (line[0]!='{')
420 continue;
421 if ((len=sscanf (line, "{T %d:%d:%d:%d",&a1,&a2,&a3,&a4)) < 4)
422 continue;
423 current->start = a1*360000+a2*6000+a3*100+a4/10;
424 for (i=0; i<SUB_MAX_TEXT;) {
425 if (!stream_read_line (st, line, LINE_LEN, utf16)) break;
426 if (line[0]=='}') break;
427 len=0;
428 for (p=line; *p!='\n' && *p!='\r' && *p; ++p,++len);
429 if (len) {
430 current->text[i]=malloc (len+1);
431 if (!current->text[i]) return ERR;
432 strncpy (current->text[i], line, len); current->text[i][len]='\0';
433 ++i;
434 } else {
435 break;
438 current->lines=i;
440 return current;
444 static subtitle *sub_read_line_vplayer(stream_t *st,subtitle *current, int utf16) {
445 char line[LINE_LEN+1];
446 int a1,a2,a3;
447 char *p=NULL, *next,separator;
448 int i,len,plen;
450 while (!current->text[0]) {
451 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
452 if ((len=sscanf (line, "%d:%d:%d%c%n",&a1,&a2,&a3,&separator,&plen)) < 4)
453 continue;
455 if (!(current->start = a1*360000+a2*6000+a3*100))
456 continue;
457 /* removed by wodzu
458 p=line;
459 // finds the body of the subtitle
460 for (i=0; i<3; i++){
461 p=strchr(p,':');
462 if (p==NULL) break;
463 ++p;
465 if (p==NULL) {
466 printf("SUB: Skipping incorrect subtitle line!\n");
467 continue;
470 // by wodzu: hey! this time we know what length it has! what is
471 // that magic for? it can't deal with space instead of third
472 // colon! look, what simple it can be:
473 p = &line[ plen ];
475 i=0;
476 if (*p!='|') {
478 next = p,i=0;
479 while ((next =sub_readtext (next, &(current->text[i])))) {
480 if (current->text[i]==ERR) {return ERR;}
481 i++;
482 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
484 current->lines=i+1;
487 return current;
490 static subtitle *sub_read_line_rt(stream_t *st,subtitle *current, int utf16) {
491 //TODO: This format uses quite rich (sub/super)set of xhtml
492 // I couldn't check it since DTD is not included.
493 // WARNING: full XML parses can be required for proper parsing
494 char line[LINE_LEN+1];
495 int a1,a2,a3,a4,b1,b2,b3,b4;
496 char *p=NULL,*next=NULL;
497 int i,len,plen;
499 while (!current->text[0]) {
500 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
501 //TODO: it seems that format of time is not easily determined, it may be 1:12, 1:12.0 or 0:1:12.0
502 //to describe the same moment in time. Maybe there are even more formats in use.
503 //if ((len=sscanf (line, "<Time Begin=\"%d:%d:%d.%d\" End=\"%d:%d:%d.%d\"",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8)
504 plen=a1=a2=a3=a4=b1=b2=b3=b4=0;
505 if (
506 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d.%d\" %*[Ee]nd=\"%d.%d\"%*[^<]<clear/>%n",&a3,&a4,&b3,&b4,&plen)) < 4) &&
507 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d.%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a3,&a4,&b2,&b3,&b4,&plen)) < 5) &&
508 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&plen)) < 4) &&
509 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&b4,&plen)) < 5) &&
510 // ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&b2,&b3,&plen)) < 5) &&
511 ((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) &&
512 ((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) &&
513 //now try it without end time
514 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d.%d\"%*[^<]<clear/>%n",&a3,&a4,&plen)) < 2) &&
515 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&plen)) < 2) &&
516 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&plen)) < 3) &&
517 ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d:%d.%d\"%*[^<]<clear/>%n",&a1,&a2,&a3,&a4,&plen)) < 4)
519 continue;
520 current->start = a1*360000+a2*6000+a3*100+a4/10;
521 current->end = b1*360000+b2*6000+b3*100+b4/10;
522 if (b1 == 0 && b2 == 0 && b3 == 0 && b4 == 0)
523 current->end = current->start+200;
524 p=line; p+=plen;i=0;
525 // TODO: I don't know what kind of convention is here for marking multiline subs, maybe <br/> like in xml?
526 next = strstr(line,"<clear/>");
527 if(next && strlen(next)>8){
528 next+=8;i=0;
529 while ((next =sub_readtext (next, &(current->text[i])))) {
530 if (current->text[i]==ERR) {return ERR;}
531 i++;
532 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
535 current->lines=i+1;
537 return current;
540 static subtitle *sub_read_line_ssa(stream_t *st,subtitle *current, int utf16) {
542 * Sub Station Alpha v4 (and v2?) scripts have 9 commas before subtitle
543 * other Sub Station Alpha scripts have only 8 commas before subtitle
544 * Reading the "ScriptType:" field is not reliable since many scripts appear
545 * w/o it
547 * http://www.scriptclub.org is a good place to find more examples
548 * http://www.eswat.demon.co.uk is where the SSA specs can be found
550 int comma;
551 static int max_comma = 32; /* let's use 32 for the case that the */
552 /* amount of commas increase with newer SSA versions */
554 int hour1, min1, sec1, hunsec1,
555 hour2, min2, sec2, hunsec2, nothing;
556 int num;
558 char line[LINE_LEN+1],
559 line3[LINE_LEN+1],
560 *line2;
561 char *tmp;
563 do {
564 if (!stream_read_line (st, line, LINE_LEN, utf16)) return NULL;
565 } while (sscanf (line, "Dialogue: Marked=%d,%d:%d:%d.%d,%d:%d:%d.%d,"
566 "%[^\n\r]", &nothing,
567 &hour1, &min1, &sec1, &hunsec1,
568 &hour2, &min2, &sec2, &hunsec2,
569 line3) < 9
571 sscanf (line, "Dialogue: %d,%d:%d:%d.%d,%d:%d:%d.%d,"
572 "%[^\n\r]", &nothing,
573 &hour1, &min1, &sec1, &hunsec1,
574 &hour2, &min2, &sec2, &hunsec2,
575 line3) < 9 );
577 line2=strchr(line3, ',');
579 for (comma = 4; comma < max_comma; comma ++)
581 tmp = line2;
582 if(!(tmp=strchr(++tmp, ','))) break;
583 if(*(++tmp) == ' ') break;
584 /* a space after a comma means we're already in a sentence */
585 line2 = tmp;
588 if(comma < max_comma)max_comma = comma;
589 /* eliminate the trailing comma */
590 if(*line2 == ',') line2++;
592 current->lines=0;num=0;
593 current->start = 360000*hour1 + 6000*min1 + 100*sec1 + hunsec1;
594 current->end = 360000*hour2 + 6000*min2 + 100*sec2 + hunsec2;
596 while (((tmp=strstr(line2, "\\n")) != NULL) || ((tmp=strstr(line2, "\\N")) != NULL) ){
597 current->text[num]=malloc(tmp-line2+1);
598 strncpy (current->text[num], line2, tmp-line2);
599 current->text[num][tmp-line2]='\0';
600 line2=tmp+2;
601 num++;
602 current->lines++;
603 if (current->lines >= SUB_MAX_TEXT) return current;
606 current->text[num]=strdup(line2);
607 current->lines++;
609 return current;
612 static void sub_pp_ssa(subtitle *sub) {
613 int l=sub->lines;
614 char *so,*de,*start;
616 while (l){
617 /* eliminate any text enclosed with {}, they are font and color settings */
618 so=de=sub->text[--l];
619 while (*so) {
620 if(*so == '{' && so[1]=='\\') {
621 for (start=so; *so && *so!='}'; so++);
622 if(*so) so++; else so=start;
624 if(*so) {
625 *de=*so;
626 so++; de++;
629 *de=*so;
634 * PJS subtitles reader.
635 * That's the "Phoenix Japanimation Society" format.
636 * I found some of them in http://www.scriptsclub.org/ (used for anime).
637 * The time is in tenths of second.
639 * by set, based on code by szabi (dunnowhat sub format ;-)
641 static subtitle *sub_read_line_pjs(stream_t *st,subtitle *current, int utf16) {
642 char line[LINE_LEN+1];
643 char text[LINE_LEN+1], *s, *d;
645 if (!stream_read_line (st, line, LINE_LEN, utf16))
646 return NULL;
647 /* skip spaces */
648 for (s=line; *s && isspace(*s); s++);
649 /* allow empty lines at the end of the file */
650 if (*s==0)
651 return NULL;
652 /* get the time */
653 if (sscanf (s, "%ld,%ld,", &(current->start),
654 &(current->end)) <2) {
655 return ERR;
657 /* the files I have are in tenths of second */
658 current->start *= 10;
659 current->end *= 10;
660 /* walk to the beggining of the string */
661 for (; *s; s++) if (*s==',') break;
662 if (*s) {
663 for (s++; *s; s++) if (*s==',') break;
664 if (*s) s++;
666 if (*s!='"') {
667 return ERR;
669 /* copy the string to the text buffer */
670 for (s++, d=text; *s && *s!='"'; s++, d++)
671 *d=*s;
672 *d=0;
673 current->text[0] = strdup(text);
674 current->lines = 1;
676 return current;
679 static subtitle *sub_read_line_mpsub(stream_t *st, subtitle *current, int utf16) {
680 char line[LINE_LEN+1];
681 float a,b;
682 int num=0;
683 char *p, *q;
687 if (!stream_read_line(st, line, LINE_LEN, utf16)) return NULL;
688 } while (sscanf (line, "%f %f", &a, &b) !=2);
690 mpsub_position += a*mpsub_multiplier;
691 current->start=(int) mpsub_position;
692 mpsub_position += b*mpsub_multiplier;
693 current->end=(int) mpsub_position;
695 while (num < SUB_MAX_TEXT) {
696 if (!stream_read_line (st, line, LINE_LEN, utf16)) {
697 if (num == 0) return NULL;
698 else return current;
700 p=line;
701 while (isspace(*p)) p++;
702 if (eol(*p) && num > 0) return current;
703 if (eol(*p)) return NULL;
705 for (q=p; !eol(*q); q++);
706 *q='\0';
707 if (strlen(p)) {
708 current->text[num]=strdup(p);
709 // printf (">%s<\n",p);
710 current->lines = ++num;
711 } else {
712 if (num) return current;
713 else return NULL;
716 return NULL; // we should have returned before if it's OK
719 #ifndef CONFIG_SORTSUB
720 //we don't need this if we use previous_sub_end
721 subtitle *previous_aqt_sub = NULL;
722 #endif
724 static subtitle *sub_read_line_aqt(stream_t *st,subtitle *current, int utf16) {
725 char line[LINE_LEN+1];
726 char *next;
727 int i;
729 while (1) {
730 // try to locate next subtitle
731 if (!stream_read_line (st, line, LINE_LEN, utf16))
732 return NULL;
733 if (!(sscanf (line, "-->> %ld", &(current->start)) <1))
734 break;
737 #ifdef CONFIG_SORTSUB
738 previous_sub_end = (current->start) ? current->start - 1 : 0;
739 #else
740 if (previous_aqt_sub != NULL)
741 previous_aqt_sub->end = current->start-1;
743 previous_aqt_sub = current;
744 #endif
746 if (!stream_read_line (st, line, LINE_LEN, utf16))
747 return NULL;
749 sub_readtext((char *) &line,&current->text[0]);
750 current->lines = 1;
751 current->end = current->start; // will be corrected by next subtitle
753 if (!stream_read_line (st, line, LINE_LEN, utf16))
754 return current;
756 next = line,i=1;
757 while ((next =sub_readtext (next, &(current->text[i])))) {
758 if (current->text[i]==ERR) {return ERR;}
759 i++;
760 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
762 current->lines=i+1;
764 if (!strlen(current->text[0]) && !strlen(current->text[1])) {
765 #ifdef CONFIG_SORTSUB
766 previous_sub_end = 0;
767 #else
768 // void subtitle -> end of previous marked and exit
769 previous_aqt_sub = NULL;
770 #endif
771 return NULL;
774 return current;
777 #ifndef CONFIG_SORTSUB
778 subtitle *previous_subrip09_sub = NULL;
779 #endif
781 static subtitle *sub_read_line_subrip09(stream_t *st,subtitle *current, int utf16) {
782 char line[LINE_LEN+1];
783 int a1,a2,a3;
784 char * next=NULL;
785 int i,len;
787 while (1) {
788 // try to locate next subtitle
789 if (!stream_read_line (st, line, LINE_LEN, utf16))
790 return NULL;
791 if (!((len=sscanf (line, "[%d:%d:%d]",&a1,&a2,&a3)) < 3))
792 break;
795 current->start = a1*360000+a2*6000+a3*100;
797 #ifdef CONFIG_SORTSUB
798 previous_sub_end = (current->start) ? current->start - 1 : 0;
799 #else
800 if (previous_subrip09_sub != NULL)
801 previous_subrip09_sub->end = current->start-1;
803 previous_subrip09_sub = current;
804 #endif
806 if (!stream_read_line (st, line, LINE_LEN, utf16))
807 return NULL;
809 next = line,i=0;
811 current->text[0]=""; // just to be sure that string is clear
813 while ((next =sub_readtext (next, &(current->text[i])))) {
814 if (current->text[i]==ERR) {return ERR;}
815 i++;
816 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
818 current->lines=i+1;
820 if (!strlen(current->text[0]) && (i==0)) {
821 #ifdef CONFIG_SORTSUB
822 previous_sub_end = 0;
823 #else
824 // void subtitle -> end of previous marked and exit
825 previous_subrip09_sub = NULL;
826 #endif
827 return NULL;
830 return current;
833 static subtitle *sub_read_line_jacosub(stream_t* st, subtitle * current, int utf16)
835 char line1[LINE_LEN], line2[LINE_LEN], directive[LINE_LEN], *p, *q;
836 unsigned a1, a2, a3, a4, b1, b2, b3, b4, comment = 0;
837 static unsigned jacoTimeres = 30;
838 static int jacoShift = 0;
840 memset(current, 0, sizeof(subtitle));
841 memset(line1, 0, LINE_LEN);
842 memset(line2, 0, LINE_LEN);
843 memset(directive, 0, LINE_LEN);
844 while (!current->text[0]) {
845 if (!stream_read_line(st, line1, LINE_LEN, utf16)) {
846 return NULL;
848 if (sscanf
849 (line1, "%u:%u:%u.%u %u:%u:%u.%u %[^\n\r]", &a1, &a2, &a3, &a4,
850 &b1, &b2, &b3, &b4, line2) < 9) {
851 if (sscanf(line1, "@%u @%u %[^\n\r]", &a4, &b4, line2) < 3) {
852 if (line1[0] == '#') {
853 int hours = 0, minutes = 0, seconds, delta, inverter =
855 unsigned units = jacoShift;
856 switch (toupper(line1[1])) {
857 case 'S':
858 if (isalpha(line1[2])) {
859 delta = 6;
860 } else {
861 delta = 2;
863 if (sscanf(&line1[delta], "%d", &hours)) {
864 if (hours < 0) {
865 hours *= -1;
866 inverter = -1;
868 if (sscanf(&line1[delta], "%*d:%d", &minutes)) {
869 if (sscanf
870 (&line1[delta], "%*d:%*d:%d",
871 &seconds)) {
872 sscanf(&line1[delta], "%*d:%*d:%*d.%d",
873 &units);
874 } else {
875 hours = 0;
876 sscanf(&line1[delta], "%d:%d.%d",
877 &minutes, &seconds, &units);
878 minutes *= inverter;
880 } else {
881 hours = minutes = 0;
882 sscanf(&line1[delta], "%d.%d", &seconds,
883 &units);
884 seconds *= inverter;
886 jacoShift =
887 ((hours * 3600 + minutes * 60 +
888 seconds) * jacoTimeres +
889 units) * inverter;
891 break;
892 case 'T':
893 if (isalpha(line1[2])) {
894 delta = 8;
895 } else {
896 delta = 2;
898 sscanf(&line1[delta], "%u", &jacoTimeres);
899 break;
902 continue;
903 } else {
904 current->start =
905 (unsigned long) ((a4 + jacoShift) * 100.0 /
906 jacoTimeres);
907 current->end =
908 (unsigned long) ((b4 + jacoShift) * 100.0 /
909 jacoTimeres);
911 } else {
912 current->start =
913 (unsigned
914 long) (((a1 * 3600 + a2 * 60 + a3) * jacoTimeres + a4 +
915 jacoShift) * 100.0 / jacoTimeres);
916 current->end =
917 (unsigned
918 long) (((b1 * 3600 + b2 * 60 + b3) * jacoTimeres + b4 +
919 jacoShift) * 100.0 / jacoTimeres);
921 current->lines = 0;
922 p = line2;
923 while ((*p == ' ') || (*p == '\t')) {
924 ++p;
926 if (isalpha(*p)||*p == '[') {
927 int cont, jLength;
929 if (sscanf(p, "%s %[^\n\r]", directive, line1) < 2)
930 return (subtitle *) ERR;
931 jLength = strlen(directive);
932 for (cont = 0; cont < jLength; ++cont) {
933 if (isalpha(*(directive + cont)))
934 *(directive + cont) = toupper(*(directive + cont));
936 if ((strstr(directive, "RDB") != NULL)
937 || (strstr(directive, "RDC") != NULL)
938 || (strstr(directive, "RLB") != NULL)
939 || (strstr(directive, "RLG") != NULL)) {
940 continue;
942 if (strstr(directive, "JL") != NULL) {
943 current->alignment = SUB_ALIGNMENT_BOTTOMLEFT;
944 } else if (strstr(directive, "JR") != NULL) {
945 current->alignment = SUB_ALIGNMENT_BOTTOMRIGHT;
946 } else {
947 current->alignment = SUB_ALIGNMENT_BOTTOMCENTER;
949 strcpy(line2, line1);
950 p = line2;
952 for (q = line1; (!eol(*p)) && (current->lines < SUB_MAX_TEXT); ++p) {
953 switch (*p) {
954 case '{':
955 comment++;
956 break;
957 case '}':
958 if (comment) {
959 --comment;
960 //the next line to get rid of a blank after the comment
961 if ((*(p + 1)) == ' ')
962 p++;
964 break;
965 case '~':
966 if (!comment) {
967 *q = ' ';
968 ++q;
970 break;
971 case ' ':
972 case '\t':
973 if ((*(p + 1) == ' ') || (*(p + 1) == '\t'))
974 break;
975 if (!comment) {
976 *q = ' ';
977 ++q;
979 break;
980 case '\\':
981 if (*(p + 1) == 'n') {
982 *q = '\0';
983 q = line1;
984 current->text[current->lines++] = strdup(line1);
985 ++p;
986 break;
988 if ((toupper(*(p + 1)) == 'C')
989 || (toupper(*(p + 1)) == 'F')) {
990 ++p,++p;
991 break;
993 if ((*(p + 1) == 'B') || (*(p + 1) == 'b') || (*(p + 1) == 'D') || //actually this means "insert current date here"
994 (*(p + 1) == 'I') || (*(p + 1) == 'i') || (*(p + 1) == 'N') || (*(p + 1) == 'T') || //actually this means "insert current time here"
995 (*(p + 1) == 'U') || (*(p + 1) == 'u')) {
996 ++p;
997 break;
999 if ((*(p + 1) == '\\') ||
1000 (*(p + 1) == '~') || (*(p + 1) == '{')) {
1001 ++p;
1002 } else if (eol(*(p + 1))) {
1003 if (!stream_read_line(st, directive, LINE_LEN, utf16))
1004 return NULL;
1005 trail_space(directive);
1006 av_strlcat(line2, directive, LINE_LEN);
1007 break;
1009 default:
1010 if (!comment) {
1011 *q = *p;
1012 ++q;
1014 } //-- switch
1015 } //-- for
1016 *q = '\0';
1017 current->text[current->lines] = strdup(line1);
1018 } //-- while
1019 current->lines++;
1020 return current;
1023 static int sub_autodetect (stream_t* st, int *uses_time, int utf16) {
1024 char line[LINE_LEN+1];
1025 int i,j=0;
1026 char p;
1028 while (j < 100) {
1029 j++;
1030 if (!stream_read_line (st, line, LINE_LEN, utf16))
1031 return SUB_INVALID;
1033 if (sscanf (line, "{%d}{%d}", &i, &i)==2)
1034 {*uses_time=0;return SUB_MICRODVD;}
1035 if (sscanf (line, "{%d}{}", &i)==1)
1036 {*uses_time=0;return SUB_MICRODVD;}
1037 if (sscanf (line, "[%d][%d]", &i, &i)==2)
1038 {*uses_time=1;return SUB_MPL2;}
1039 if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d", &i, &i, &i, &i, &i, &i, &i, &i)==8)
1040 {*uses_time=1;return SUB_SUBRIP;}
1041 if (sscanf (line, "%d:%d:%d%[,.:]%d --> %d:%d:%d%[,.:]%d", &i, &i, &i, (char *)&i, &i, &i, &i, &i, (char *)&i, &i)==10)
1042 {*uses_time=1;return SUB_SUBVIEWER;}
1043 if (sscanf (line, "{T %d:%d:%d:%d",&i, &i, &i, &i)==4)
1044 {*uses_time=1;return SUB_SUBVIEWER2;}
1045 if (strstr (line, "<SAMI>"))
1046 {*uses_time=1; return SUB_SAMI;}
1047 if (sscanf(line, "%d:%d:%d.%d %d:%d:%d.%d", &i, &i, &i, &i, &i, &i, &i, &i) == 8)
1048 {*uses_time = 1; return SUB_JACOSUB;}
1049 if (sscanf(line, "@%d @%d", &i, &i) == 2)
1050 {*uses_time = 1; return SUB_JACOSUB;}
1051 if (sscanf (line, "%d:%d:%d:", &i, &i, &i )==3)
1052 {*uses_time=1;return SUB_VPLAYER;}
1053 if (sscanf (line, "%d:%d:%d ", &i, &i, &i )==3)
1054 {*uses_time=1;return SUB_VPLAYER;}
1055 //TODO: just checking if first line of sub starts with "<" is WAY
1056 // too weak test for RT
1057 // Please someone who knows the format of RT... FIX IT!!!
1058 // It may conflict with other sub formats in the future (actually it doesn't)
1059 if ( *line == '<' )
1060 {*uses_time=1;return SUB_RT;}
1062 if (!memcmp(line, "Dialogue: Marked", 16))
1063 {*uses_time=1; return SUB_SSA;}
1064 if (!memcmp(line, "Dialogue: ", 10))
1065 {*uses_time=1; return SUB_SSA;}
1066 if (sscanf (line, "%d,%d,\"%c", &i, &i, (char *) &i) == 3)
1067 {*uses_time=1;return SUB_PJS;}
1068 if (sscanf (line, "FORMAT=%d", &i) == 1)
1069 {*uses_time=0; return SUB_MPSUB;}
1070 if (sscanf (line, "FORMAT=TIM%c", &p)==1 && p=='E')
1071 {*uses_time=1; return SUB_MPSUB;}
1072 if (strstr (line, "-->>"))
1073 {*uses_time=0; return SUB_AQTITLE;}
1074 if (sscanf (line, "[%d:%d:%d]", &i, &i, &i)==3)
1075 {*uses_time=1;return SUB_SUBRIP09;}
1078 return SUB_INVALID; // too many bad lines
1081 extern int sub_utf8;
1082 int sub_utf8_prev=0;
1084 extern float sub_delay;
1085 extern float sub_fps;
1087 #ifdef CONFIG_ICONV
1088 static iconv_t icdsc = (iconv_t)(-1);
1090 void subcp_open (stream_t *st)
1092 char *tocp = "UTF-8";
1094 if (sub_cp){
1095 const char *cp_tmp = sub_cp;
1096 #ifdef CONFIG_ENCA
1097 char enca_lang[3], enca_fallback[100];
1098 if (sscanf(sub_cp, "enca:%2s:%99s", enca_lang, enca_fallback) == 2
1099 || sscanf(sub_cp, "ENCA:%2s:%99s", enca_lang, enca_fallback) == 2) {
1100 if (st && st->flags & MP_STREAM_SEEK ) {
1101 cp_tmp = guess_cp(st, enca_lang, enca_fallback);
1102 } else {
1103 cp_tmp = enca_fallback;
1104 if (st)
1105 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: enca failed, stream must be seekable.\n");
1108 #endif
1109 if ((icdsc = iconv_open (tocp, cp_tmp)) != (iconv_t)(-1)){
1110 mp_msg(MSGT_SUBREADER,MSGL_V,"SUB: opened iconv descriptor.\n");
1111 sub_utf8 = 2;
1112 } else
1113 mp_msg(MSGT_SUBREADER,MSGL_ERR,"SUB: error opening iconv descriptor.\n");
1117 void subcp_close (void)
1119 if (icdsc != (iconv_t)(-1)){
1120 (void) iconv_close (icdsc);
1121 icdsc = (iconv_t)(-1);
1122 mp_msg(MSGT_SUBREADER,MSGL_V,"SUB: closed iconv descriptor.\n");
1126 subtitle* subcp_recode (subtitle *sub)
1128 int l=sub->lines;
1129 size_t ileft, oleft;
1130 char *op, *ip, *ot;
1131 if(icdsc == (iconv_t)(-1)) return sub;
1133 while (l){
1134 ip = sub->text[--l];
1135 ileft = strlen(ip);
1136 oleft = 4 * ileft;
1138 if (!(ot = malloc(oleft + 1))){
1139 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: error allocating mem.\n");
1140 continue;
1142 op = ot;
1143 if (iconv(icdsc, &ip, &ileft,
1144 &op, &oleft) == (size_t)(-1)) {
1145 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: error recoding line.\n");
1146 free(ot);
1147 continue;
1149 // In some stateful encodings, we must clear the state to handle the last character
1150 if (iconv(icdsc, NULL, NULL,
1151 &op, &oleft) == (size_t)(-1)) {
1152 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: error recoding line, can't clear encoding state.\n");
1154 *op='\0' ;
1155 free (sub->text[l]);
1156 sub->text[l] = ot;
1158 return sub;
1160 #endif
1162 #ifdef CONFIG_FRIBIDI
1164 * Do conversion necessary for right-to-left language support via fribidi.
1165 * @param sub subtitle to convert
1166 * @param sub_utf8 whether the subtitle is encoded in UTF-8
1167 * @param from first new subtitle, all lines before this are assumed to be already converted
1169 static subtitle* sub_fribidi (subtitle *sub, int sub_utf8, int from)
1171 FriBidiChar logical[LINE_LEN+1], visual[LINE_LEN+1]; // Hopefully these two won't smash the stack
1172 char *ip = NULL, *op = NULL;
1173 size_t len,orig_len;
1174 int l=sub->lines;
1175 int char_set_num;
1176 fribidi_boolean log2vis;
1177 if (!flip_hebrew)
1178 return sub;
1179 fribidi_set_mirroring(1);
1180 fribidi_set_reorder_nsm(0);
1182 if( sub_utf8 == 0 ) {
1183 char_set_num = fribidi_parse_charset (fribidi_charset?fribidi_charset:"ISO8859-8");
1184 }else {
1185 char_set_num = fribidi_parse_charset ("UTF-8");
1187 while (l > from) {
1188 ip = sub->text[--l];
1189 orig_len = len = strlen( ip ); // We assume that we don't use full unicode, only UTF-8 or ISO8859-x
1190 if(len > LINE_LEN) {
1191 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: sub->text is longer than LINE_LEN.\n");
1192 l++;
1193 break;
1195 len = fribidi_charset_to_unicode (char_set_num, ip, len, logical);
1196 #if FRIBIDI_INTERFACE_VERSION < 3
1197 FriBidiCharType base = fribidi_flip_commas?FRIBIDI_TYPE_ON:FRIBIDI_TYPE_L;
1198 #else
1199 FriBidiParType base = fribidi_flip_commas?FRIBIDI_TYPE_ON:FRIBIDI_TYPE_L;
1200 #endif
1201 log2vis = fribidi_log2vis (logical, len, &base,
1202 /* output */
1203 visual, NULL, NULL, NULL);
1204 if(log2vis) {
1205 len = fribidi_remove_bidi_marks (visual, len, NULL, NULL,
1206 NULL);
1207 if((op = malloc((FFMAX(2*orig_len,2*len) + 1))) == NULL) {
1208 mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: error allocating mem.\n");
1209 l++;
1210 break;
1212 fribidi_unicode_to_charset ( char_set_num, visual, len,op);
1213 free (ip);
1214 sub->text[l] = op;
1217 if (!from && l){
1218 for (l = sub->lines; l;)
1219 free (sub->text[--l]);
1220 return ERR;
1222 return sub;
1225 #endif
1227 static void adjust_subs_time(subtitle* sub, float subtime, float fps, int block,
1228 int sub_num, int sub_uses_time) {
1229 int n,m;
1230 subtitle* nextsub;
1231 int i = sub_num;
1232 unsigned long subfms = (sub_uses_time ? 100 : fps) * subtime;
1233 unsigned long overlap = (sub_uses_time ? 100 : fps) / 5; // 0.2s
1235 n=m=0;
1236 if (i) for (;;){
1237 if (sub->end <= sub->start){
1238 sub->end = sub->start + subfms;
1239 m++;
1240 n++;
1242 if (!--i) break;
1243 nextsub = sub + 1;
1244 if(block){
1245 if ((sub->end > nextsub->start) && (sub->end <= nextsub->start + overlap)) {
1246 // these subtitles overlap for less than 0.2 seconds
1247 // and would result in very short overlapping subtitle
1248 // so let's fix the problem here, before overlapping code
1249 // get its hands on them
1250 unsigned delta = sub->end - nextsub->start, half = delta / 2;
1251 sub->end -= half + 1;
1252 nextsub->start += delta - half;
1254 if (sub->end >= nextsub->start){
1255 sub->end = nextsub->start - 1;
1256 if (sub->end - sub->start > subfms)
1257 sub->end = sub->start + subfms;
1258 if (!m)
1259 n++;
1263 /* Theory:
1264 * Movies are often converted from FILM (24 fps)
1265 * to PAL (25) by simply speeding it up, so we
1266 * to multiply the original timestmaps by
1267 * (Movie's FPS / Subtitle's (guessed) FPS)
1268 * so eg. for 23.98 fps movie and PAL time based
1269 * subtitles we say -subfps 25 and we're fine!
1272 /* timed sub fps correction ::atmos */
1273 /* the frame-based case is handled in mpcommon.c
1274 * where find_sub is called */
1275 if(sub_uses_time && sub_fps) {
1276 sub->start *= sub_fps/fps;
1277 sub->end *= sub_fps/fps;
1280 sub = nextsub;
1281 m = 0;
1283 if (n) mp_msg(MSGT_SUBREADER,MSGL_V,"SUB: Adjusted %d subtitle(s).\n", n);
1286 struct subreader {
1287 subtitle * (*read)(stream_t *st,subtitle *dest,int utf16);
1288 void (*post)(subtitle *dest);
1289 const char *name;
1292 #ifdef CONFIG_ENCA
1293 const char* guess_buffer_cp(unsigned char* buffer, int buflen, const char *preferred_language, const char *fallback)
1295 const char **languages;
1296 size_t langcnt;
1297 EncaAnalyser analyser;
1298 EncaEncoding encoding;
1299 const char *detected_sub_cp = NULL;
1300 int i;
1302 languages = enca_get_languages(&langcnt);
1303 mp_msg(MSGT_SUBREADER, MSGL_V, "ENCA supported languages: ");
1304 for (i = 0; i < langcnt; i++) {
1305 mp_msg(MSGT_SUBREADER, MSGL_V, "%s ", languages[i]);
1307 mp_msg(MSGT_SUBREADER, MSGL_V, "\n");
1309 for (i = 0; i < langcnt; i++) {
1310 if (strcasecmp(languages[i], preferred_language) != 0) continue;
1311 analyser = enca_analyser_alloc(languages[i]);
1312 encoding = enca_analyse_const(analyser, buffer, buflen);
1313 enca_analyser_free(analyser);
1314 if (encoding.charset != ENCA_CS_UNKNOWN) {
1315 detected_sub_cp = enca_charset_name(encoding.charset, ENCA_NAME_STYLE_ICONV);
1316 break;
1320 free(languages);
1322 if (!detected_sub_cp) {
1323 detected_sub_cp = fallback;
1324 mp_msg(MSGT_SUBREADER, MSGL_INFO, "ENCA detection failed: fallback to %s\n", fallback);
1325 }else{
1326 mp_msg(MSGT_SUBREADER, MSGL_INFO, "ENCA detected charset: %s\n", detected_sub_cp);
1329 return detected_sub_cp;
1332 #define MAX_GUESS_BUFFER_SIZE (256*1024)
1333 const char* guess_cp(stream_t *st, const char *preferred_language, const char *fallback)
1335 size_t buflen;
1336 unsigned char *buffer;
1337 const char *detected_sub_cp = NULL;
1339 buffer = malloc(MAX_GUESS_BUFFER_SIZE);
1340 buflen = stream_read(st,buffer, MAX_GUESS_BUFFER_SIZE);
1342 detected_sub_cp = guess_buffer_cp(buffer, buflen, preferred_language, fallback);
1344 free(buffer);
1345 stream_reset(st);
1346 stream_seek(st,0);
1348 return detected_sub_cp;
1350 #undef MAX_GUESS_BUFFER_SIZE
1351 #endif
1353 sub_data* sub_read_file (char *filename, float fps) {
1354 int utf16;
1355 stream_t* fd;
1356 int n_max, n_first, i, j, sub_first, sub_orig;
1357 subtitle *first, *second, *sub, *return_sub, *alloced_sub = NULL;
1358 sub_data *subt_data;
1359 int uses_time = 0, sub_num = 0, sub_errs = 0;
1360 static const struct subreader sr[]=
1362 { sub_read_line_microdvd, NULL, "microdvd" },
1363 { sub_read_line_subrip, NULL, "subrip" },
1364 { sub_read_line_subviewer, NULL, "subviewer" },
1365 { sub_read_line_sami, NULL, "sami" },
1366 { sub_read_line_vplayer, NULL, "vplayer" },
1367 { sub_read_line_rt, NULL, "rt" },
1368 { sub_read_line_ssa, sub_pp_ssa, "ssa" },
1369 { sub_read_line_pjs, NULL, "pjs" },
1370 { sub_read_line_mpsub, NULL, "mpsub" },
1371 { sub_read_line_aqt, NULL, "aqt" },
1372 { sub_read_line_subviewer2, NULL, "subviewer 2.0" },
1373 { sub_read_line_subrip09, NULL, "subrip 0.9" },
1374 { sub_read_line_jacosub, NULL, "jacosub" },
1375 { sub_read_line_mpl2, NULL, "mpl2" }
1377 const struct subreader *srp;
1379 if(filename==NULL) return NULL; //qnx segfault
1380 i = 0;
1381 fd=open_stream (filename, NULL, &i); if (!fd) return NULL;
1383 sub_format = SUB_INVALID;
1384 for (utf16 = 0; sub_format == SUB_INVALID && utf16 < 3; utf16++) {
1385 sub_format=sub_autodetect (fd, &uses_time, utf16);
1386 stream_reset(fd);
1387 stream_seek(fd,0);
1389 utf16--;
1391 mpsub_multiplier = (uses_time ? 100.0 : 1.0);
1392 if (sub_format==SUB_INVALID) {mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: Could not determine file format\n");return NULL;}
1393 srp=sr+sub_format;
1394 mp_msg(MSGT_SUBREADER, MSGL_V, "SUB: Detected subtitle file format: %s\n", srp->name);
1396 #ifdef CONFIG_ICONV
1397 sub_utf8_prev=sub_utf8;
1399 int l,k;
1400 k = -1;
1401 if ((l=strlen(filename))>4){
1402 char *exts[] = {".utf", ".utf8", ".utf-8" };
1403 for (k=3;--k>=0;)
1404 if (l >= strlen(exts[k]) && !strcasecmp(filename+(l - strlen(exts[k])), exts[k])){
1405 sub_utf8 = 1;
1406 break;
1409 if (k<0) subcp_open(fd);
1411 #endif
1413 sub_num=0;n_max=32;
1414 first=malloc(n_max*sizeof(subtitle));
1415 if(!first){
1416 #ifdef CONFIG_ICONV
1417 subcp_close();
1418 sub_utf8=sub_utf8_prev;
1419 #endif
1420 return NULL;
1423 #ifdef CONFIG_SORTSUB
1424 alloced_sub =
1425 sub = malloc(sizeof(subtitle));
1426 //This is to deal with those formats (AQT & Subrip) which define the end of a subtitle
1427 //as the beginning of the following
1428 previous_sub_end = 0;
1429 #endif
1430 while(1){
1431 if(sub_num>=n_max){
1432 n_max+=16;
1433 first=realloc(first,n_max*sizeof(subtitle));
1435 #ifndef CONFIG_SORTSUB
1436 sub = &first[sub_num];
1437 #endif
1438 memset(sub, '\0', sizeof(subtitle));
1439 sub=srp->read(fd,sub,utf16);
1440 if(!sub) break; // EOF
1441 #ifdef CONFIG_ICONV
1442 if ((sub!=ERR) && sub_utf8 == 2) sub=subcp_recode(sub);
1443 #endif
1444 #ifdef CONFIG_FRIBIDI
1445 if (sub!=ERR) sub=sub_fribidi(sub,sub_utf8,0);
1446 #endif
1447 if ( sub == ERR )
1449 #ifdef CONFIG_ICONV
1450 subcp_close();
1451 #endif
1452 if ( first ) free(first);
1453 free(alloced_sub);
1454 return NULL;
1456 // Apply any post processing that needs recoding first
1457 if ((sub!=ERR) && !sub_no_text_pp && srp->post) srp->post(sub);
1458 #ifdef CONFIG_SORTSUB
1459 if(!sub_num || (first[sub_num - 1].start <= sub->start)){
1460 first[sub_num].start = sub->start;
1461 first[sub_num].end = sub->end;
1462 first[sub_num].lines = sub->lines;
1463 first[sub_num].alignment = sub->alignment;
1464 for(i = 0; i < sub->lines; ++i){
1465 first[sub_num].text[i] = sub->text[i];
1467 if (previous_sub_end){
1468 first[sub_num - 1].end = previous_sub_end;
1469 previous_sub_end = 0;
1471 } else {
1472 for(j = sub_num - 1; j >= 0; --j){
1473 first[j + 1].start = first[j].start;
1474 first[j + 1].end = first[j].end;
1475 first[j + 1].lines = first[j].lines;
1476 first[j + 1].alignment = first[j].alignment;
1477 for(i = 0; i < first[j].lines; ++i){
1478 first[j + 1].text[i] = first[j].text[i];
1480 if(!j || (first[j - 1].start <= sub->start)){
1481 first[j].start = sub->start;
1482 first[j].end = sub->end;
1483 first[j].lines = sub->lines;
1484 first[j].alignment = sub->alignment;
1485 for(i = 0; i < SUB_MAX_TEXT; ++i){
1486 first[j].text[i] = sub->text[i];
1488 if (previous_sub_end){
1489 first[j].end = first[j - 1].end;
1490 first[j - 1].end = previous_sub_end;
1491 previous_sub_end = 0;
1493 break;
1497 #endif
1498 if(sub==ERR) ++sub_errs; else ++sub_num; // Error vs. Valid
1501 free_stream(fd);
1503 #ifdef CONFIG_ICONV
1504 subcp_close();
1505 #endif
1506 free(alloced_sub);
1508 // printf ("SUB: Subtitle format %s time.\n", uses_time?"uses":"doesn't use");
1509 mp_msg(MSGT_SUBREADER, MSGL_V,"SUB: Read %i subtitles, %i bad line(s).\n",
1510 sub_num, sub_errs);
1512 if(sub_num<=0){
1513 free(first);
1514 return NULL;
1517 // we do overlap if the user forced it (suboverlap_enable == 2) or
1518 // the user didn't forced no-overlapsub and the format is Jacosub or Ssa.
1519 // this is because usually overlapping subtitles are found in these formats,
1520 // while in others they are probably result of bad timing
1521 if ((suboverlap_enabled == 2) ||
1522 ((suboverlap_enabled) && ((sub_format == SUB_JACOSUB) || (sub_format == SUB_SSA)))) {
1523 adjust_subs_time(first, 6.0, fps, 0, sub_num, uses_time);/*~6 secs AST*/
1524 // here we manage overlapping subtitles
1525 sub_orig = sub_num;
1526 n_first = sub_num;
1527 sub_num = 0;
1528 second = NULL;
1529 // for each subtitle in first[] we deal with its 'block' of
1530 // bonded subtitles
1531 for (sub_first = 0; sub_first < n_first; ++sub_first) {
1532 unsigned long global_start = first[sub_first].start,
1533 global_end = first[sub_first].end, local_start, local_end;
1534 int lines_to_add = first[sub_first].lines, sub_to_add = 0,
1535 **placeholder = NULL, higher_line = 0, counter, start_block_sub = sub_num;
1536 char real_block = 1;
1538 // here we find the number of subtitles inside the 'block'
1539 // and its span interval. this works well only with sorted
1540 // subtitles
1541 while ((sub_first + sub_to_add + 1 < n_first) && (first[sub_first + sub_to_add + 1].start < global_end)) {
1542 ++sub_to_add;
1543 lines_to_add += first[sub_first + sub_to_add].lines;
1544 if (first[sub_first + sub_to_add].start < global_start) {
1545 global_start = first[sub_first + sub_to_add].start;
1547 if (first[sub_first + sub_to_add].end > global_end) {
1548 global_end = first[sub_first + sub_to_add].end;
1552 /* Avoid n^2 memory use for the "placeholder" data structure
1553 * below with subtitles that have a huge number of
1554 * consecutive overlapping lines. */
1555 lines_to_add = FFMIN(lines_to_add, SUB_MAX_TEXT);
1557 // we need a structure to keep trace of the screen lines
1558 // used by the subs, a 'placeholder'
1559 counter = 2 * sub_to_add + 1; // the maximum number of subs derived
1560 // from a block of sub_to_add+1 subs
1561 placeholder = malloc(sizeof(int *) * counter);
1562 for (i = 0; i < counter; ++i) {
1563 placeholder[i] = malloc(sizeof(int) * lines_to_add);
1564 for (j = 0; j < lines_to_add; ++j) {
1565 placeholder[i][j] = -1;
1569 counter = 0;
1570 local_end = global_start - 1;
1571 do {
1572 int ls;
1574 // here we find the beginning and the end of a new
1575 // subtitle in the block
1576 local_start = local_end + 1;
1577 local_end = global_end;
1578 for (j = 0; j <= sub_to_add; ++j) {
1579 if ((first[sub_first + j].start - 1 > local_start) && (first[sub_first + j].start - 1 < local_end)) {
1580 local_end = first[sub_first + j].start - 1;
1581 } else if ((first[sub_first + j].end > local_start) && (first[sub_first + j].end < local_end)) {
1582 local_end = first[sub_first + j].end;
1585 // here we allocate the screen lines to subs we must
1586 // display in current local_start-local_end interval.
1587 // if the subs were yet presents in the previous interval
1588 // they keep the same lines, otherside they get unused lines
1589 for (j = 0; j <= sub_to_add; ++j) {
1590 if ((first[sub_first + j].start <= local_end) && (first[sub_first + j].end > local_start)) {
1591 unsigned long sub_lines = first[sub_first + j].lines, fragment_length = lines_to_add + 1,
1592 tmp = 0;
1593 char boolean = 0;
1594 int fragment_position = -1;
1596 // if this is not the first new sub of the block
1597 // we find if this sub was present in the previous
1598 // new sub
1599 if (counter)
1600 for (i = 0; i < lines_to_add; ++i) {
1601 if (placeholder[counter - 1][i] == sub_first + j) {
1602 placeholder[counter][i] = sub_first + j;
1603 boolean = 1;
1606 if (boolean)
1607 continue;
1609 // we are looking for the shortest among all groups of
1610 // sequential blank lines whose length is greater than or
1611 // equal to sub_lines. we store in fragment_position the
1612 // position of the shortest group, in fragment_length its
1613 // length, and in tmp the length of the group currently
1614 // examinated
1615 for (i = 0; i < lines_to_add; ++i) {
1616 if (placeholder[counter][i] == -1) {
1617 // placeholder[counter][i] is part of the current group
1618 // of blank lines
1619 ++tmp;
1620 } else {
1621 if (tmp == sub_lines) {
1622 // current group's size fits exactly the one we
1623 // need, so we stop looking
1624 fragment_position = i - tmp;
1625 tmp = 0;
1626 break;
1628 if ((tmp) && (tmp > sub_lines) && (tmp < fragment_length)) {
1629 // current group is the best we found till here,
1630 // but is still bigger than the one we are looking
1631 // for, so we keep on looking
1632 fragment_length = tmp;
1633 fragment_position = i - tmp;
1634 tmp = 0;
1635 } else {
1636 // current group doesn't fit at all, so we forget it
1637 tmp = 0;
1641 if (tmp) {
1642 // last screen line is blank, a group ends with it
1643 if ((tmp >= sub_lines) && (tmp < fragment_length)) {
1644 fragment_position = i - tmp;
1647 if (fragment_position == -1) {
1648 // it was not possible to find free screen line(s) for a subtitle,
1649 // usually this means a bug in the code; however we do not overlap
1650 mp_msg(MSGT_SUBREADER, MSGL_WARN, "SUB: we could not find a suitable position for an overlapping subtitle\n");
1651 higher_line = SUB_MAX_TEXT + 1;
1652 break;
1653 } else {
1654 for (tmp = 0; tmp < sub_lines; ++tmp) {
1655 placeholder[counter][fragment_position + tmp] = sub_first + j;
1660 for (j = higher_line + 1; j < lines_to_add; ++j) {
1661 if (placeholder[counter][j] != -1)
1662 higher_line = j;
1663 else
1664 break;
1666 if (higher_line >= SUB_MAX_TEXT) {
1667 // the 'block' has too much lines, so we don't overlap the
1668 // subtitles
1669 second = realloc(second, (sub_num + sub_to_add + 1) * sizeof(subtitle));
1670 for (j = 0; j <= sub_to_add; ++j) {
1671 int ls;
1672 memset(&second[sub_num + j], '\0', sizeof(subtitle));
1673 second[sub_num + j].start = first[sub_first + j].start;
1674 second[sub_num + j].end = first[sub_first + j].end;
1675 second[sub_num + j].lines = first[sub_first + j].lines;
1676 second[sub_num + j].alignment = first[sub_first + j].alignment;
1677 for (ls = 0; ls < second[sub_num + j].lines; ls++) {
1678 second[sub_num + j].text[ls] = strdup(first[sub_first + j].text[ls]);
1681 sub_num += sub_to_add + 1;
1682 sub_first += sub_to_add;
1683 real_block = 0;
1684 break;
1687 // we read the placeholder structure and create the new
1688 // subs.
1689 second = realloc(second, (sub_num + 1) * sizeof(subtitle));
1690 memset(&second[sub_num], '\0', sizeof(subtitle));
1691 second[sub_num].start = local_start;
1692 second[sub_num].end = local_end;
1693 second[sub_num].alignment = first[sub_first].alignment;
1694 n_max = (lines_to_add < SUB_MAX_TEXT) ? lines_to_add : SUB_MAX_TEXT;
1695 for (i = 0, j = 0; j < n_max; ++j) {
1696 if (placeholder[counter][j] != -1) {
1697 int lines = first[placeholder[counter][j]].lines;
1698 for (ls = 0; ls < lines; ++ls) {
1699 second[sub_num].text[i++] = strdup(first[placeholder[counter][j]].text[ls]);
1701 j += lines - 1;
1702 } else {
1703 second[sub_num].text[i++] = strdup(" ");
1706 ++sub_num;
1707 ++counter;
1708 } while (local_end < global_end);
1709 if (real_block)
1710 for (i = 0; i < counter; ++i)
1711 second[start_block_sub + i].lines = higher_line + 1;
1713 counter = 2 * sub_to_add + 1;
1714 for (i = 0; i < counter; ++i) {
1715 free(placeholder[i]);
1717 free(placeholder);
1718 sub_first += sub_to_add;
1721 for (j = sub_orig - 1; j >= 0; --j) {
1722 for (i = first[j].lines - 1; i >= 0; --i) {
1723 free(first[j].text[i]);
1726 free(first);
1728 return_sub = second;
1729 } else { //if(suboverlap_enabled)
1730 adjust_subs_time(first, 6.0, fps, 1, sub_num, uses_time);/*~6 secs AST*/
1731 return_sub = first;
1733 if (return_sub == NULL) return NULL;
1734 subt_data = malloc(sizeof(sub_data));
1735 subt_data->filename = strdup(filename);
1736 subt_data->sub_uses_time = uses_time;
1737 subt_data->sub_num = sub_num;
1738 subt_data->sub_errs = sub_errs;
1739 subt_data->subtitles = return_sub;
1740 return subt_data;
1743 #if 0
1744 char * strreplace( char * in,char * what,char * whereof )
1746 int i;
1747 char * tmp;
1749 if ( ( in == NULL )||( what == NULL )||( whereof == NULL )||( ( tmp=strstr( in,what ) ) == NULL ) ) return NULL;
1750 for( i=0;i<strlen( whereof );i++ ) tmp[i]=whereof[i];
1751 if ( strlen( what ) > strlen( whereof ) ) tmp[i]=0;
1752 return in;
1754 #endif
1757 static void strcpy_trim(char *d, char *s)
1759 // skip leading whitespace
1760 while (*s && isspace(*s)) {
1761 s++;
1763 for (;;) {
1764 // copy word
1765 while (*s && !isspace(*s)) {
1766 *d = tolower(*s);
1767 s++; d++;
1769 if (*s == 0) break;
1770 // trim excess whitespace
1771 while (*s && isspace(*s)) {
1772 s++;
1774 if (*s == 0) break;
1775 *d++ = ' ';
1777 *d = 0;
1780 static void strcpy_strip_ext(char *d, char *s)
1782 char *tmp = strrchr(s,'.');
1783 if (!tmp) {
1784 strcpy(d, s);
1785 return;
1786 } else {
1787 strncpy(d, s, tmp-s);
1788 d[tmp-s] = 0;
1790 while (*d) {
1791 *d = tolower(*d);
1792 d++;
1796 static void strcpy_get_ext(char *d, char *s)
1798 char *tmp = strrchr(s,'.');
1799 if (!tmp) {
1800 strcpy(d, "");
1801 return;
1802 } else {
1803 strcpy(d, tmp+1);
1807 static int whiteonly(char *s)
1809 while (*s) {
1810 if (!isspace(*s)) return 0;
1811 s++;
1813 return 1;
1816 typedef struct subfn
1818 int priority;
1819 char *fname;
1820 } subfn;
1822 static int compare_sub_priority(const void *a, const void *b)
1824 if (((const subfn*)a)->priority > ((const subfn*)b)->priority) {
1825 return -1;
1826 } else if (((const subfn*)a)->priority < ((const subfn*)b)->priority) {
1827 return 1;
1828 } else {
1829 return strcoll(((const subfn*)a)->fname, ((const subfn*)b)->fname);
1833 char** sub_filenames(const char* path, char *fname)
1835 char *f_dir, *f_fname, *f_fname_noext, *f_fname_trim, *tmp, *tmp_sub_id;
1836 char *tmp_fname_noext, *tmp_fname_trim, *tmp_fname_ext, *tmpresult;
1838 int len, pos, found, i, j;
1839 char * sub_exts[] = { "utf", "utf8", "utf-8", "sub", "srt", "smi", "rt", "txt", "ssa", "aqt", "jss", "js", "ass", NULL};
1840 subfn *result;
1841 char **result2;
1843 int subcnt;
1845 FILE *f;
1847 DIR *d;
1848 struct dirent *de;
1850 len = (strlen(fname) > 256 ? strlen(fname) : 256)
1851 +(strlen(path) > 256 ? strlen(path) : 256)+2;
1853 f_dir = malloc(len);
1854 f_fname = malloc(len);
1855 f_fname_noext = malloc(len);
1856 f_fname_trim = malloc(len);
1858 tmp_fname_noext = malloc(len);
1859 tmp_fname_trim = malloc(len);
1860 tmp_fname_ext = malloc(len);
1862 tmpresult = malloc(len);
1864 result = malloc(sizeof(subfn)*MAX_SUBTITLE_FILES);
1865 memset(result, 0, sizeof(subfn)*MAX_SUBTITLE_FILES);
1867 subcnt = 0;
1869 tmp = strrchr(fname,'/');
1870 #if HAVE_DOS_PATHS
1871 if(!tmp)tmp = strrchr(fname,'\\');
1872 if(!tmp)tmp = strrchr(fname,':');
1873 #endif
1875 // extract filename & dirname from fname
1876 if (tmp) {
1877 strcpy(f_fname, tmp+1);
1878 pos = tmp - fname;
1879 strncpy(f_dir, fname, pos+1);
1880 f_dir[pos+1] = 0;
1881 } else {
1882 strcpy(f_fname, fname);
1883 strcpy(f_dir, "./");
1886 strcpy_strip_ext(f_fname_noext, f_fname);
1887 strcpy_trim(f_fname_trim, f_fname_noext);
1889 /* The code using sub language here is broken - it assumes strict
1890 * "videoname languagename" syntax for the subtitle file, which is
1891 * very unlikely to match especially if language name uses "en,de"
1892 * syntax... */
1893 tmp_sub_id = NULL;
1894 #if 0
1895 if (dvdsub_lang && !whiteonly(dvdsub_lang)) {
1896 tmp_sub_id = malloc(strlen(dvdsub_lang)+1);
1897 strcpy_trim(tmp_sub_id, dvdsub_lang);
1899 #endif
1901 // 0 = nothing
1902 // 1 = any subtitle file
1903 // 2 = any sub file containing movie name
1904 // 3 = sub file containing movie name and the lang extension
1905 for (j = 0; j <= 1; j++) {
1906 d = opendir(j == 0 ? f_dir : path);
1907 if (d) {
1908 while ((de = readdir(d))) {
1909 // retrieve various parts of the filename
1910 strcpy_strip_ext(tmp_fname_noext, de->d_name);
1911 strcpy_get_ext(tmp_fname_ext, de->d_name);
1912 strcpy_trim(tmp_fname_trim, tmp_fname_noext);
1914 // does it end with a subtitle extension?
1915 found = 0;
1916 #ifdef CONFIG_ICONV
1917 #ifdef CONFIG_ENCA
1918 for (i = ((sub_cp && strncasecmp(sub_cp, "enca", 4) != 0) ? 3 : 0); sub_exts[i]; i++) {
1919 #else
1920 for (i = (sub_cp ? 3 : 0); sub_exts[i]; i++) {
1921 #endif
1922 #else
1923 for (i = 0; sub_exts[i]; i++) {
1924 #endif
1925 if (strcasecmp(sub_exts[i], tmp_fname_ext) == 0) {
1926 found = 1;
1927 break;
1931 // we have a (likely) subtitle file
1932 if (found) {
1933 int prio = 0;
1934 if (!prio && tmp_sub_id)
1936 sprintf(tmpresult, "%s %s", f_fname_trim, tmp_sub_id);
1937 if (strcmp(tmp_fname_trim, tmpresult) == 0 && sub_match_fuzziness >= 1) {
1938 // matches the movie name + lang extension
1939 prio = 5;
1942 if (!prio && strcmp(tmp_fname_trim, f_fname_trim) == 0) {
1943 // matches the movie name
1944 prio = 4;
1946 if (!prio && (tmp = strstr(tmp_fname_trim, f_fname_trim)) && (sub_match_fuzziness >= 1)) {
1947 // contains the movie name
1948 tmp += strlen(f_fname_trim);
1949 if (tmp_sub_id && strstr(tmp, tmp_sub_id)) {
1950 // with sub_id specified prefer localized subtitles
1951 prio = 3;
1952 } else if ((tmp_sub_id == NULL) && whiteonly(tmp)) {
1953 // without sub_id prefer "plain" name
1954 prio = 3;
1955 } else {
1956 // with no localized subs found, try any else instead
1957 prio = 2;
1960 if (!prio) {
1961 // doesn't contain the movie name
1962 // don't try in the mplayer subtitle directory
1963 if ((j == 0) && (sub_match_fuzziness >= 2)) {
1964 prio = 1;
1968 mp_msg(MSGT_SUBREADER, MSGL_DBG2, "Potential sub file: "
1969 "\"%s\" Priority: %d\n", de->d_name, prio);
1970 if (prio) {
1971 prio += prio;
1972 #ifdef CONFIG_ICONV
1973 if (i<3){ // prefer UTF-8 coded
1974 prio++;
1976 #endif
1977 sprintf(tmpresult, "%s%s", j == 0 ? f_dir : path, de->d_name);
1978 // fprintf(stderr, "%s priority %d\n", tmpresult, prio);
1979 if ((f = fopen(tmpresult, "rt"))) {
1980 fclose(f);
1981 result[subcnt].priority = prio;
1982 result[subcnt].fname = strdup(tmpresult);
1983 subcnt++;
1988 if (subcnt >= MAX_SUBTITLE_FILES) break;
1990 closedir(d);
1995 if (tmp_sub_id) free(tmp_sub_id);
1997 free(f_dir);
1998 free(f_fname);
1999 free(f_fname_noext);
2000 free(f_fname_trim);
2002 free(tmp_fname_noext);
2003 free(tmp_fname_trim);
2004 free(tmp_fname_ext);
2006 free(tmpresult);
2008 qsort(result, subcnt, sizeof(subfn), compare_sub_priority);
2010 result2 = malloc(sizeof(char*)*(subcnt+1));
2011 memset(result2, 0, sizeof(char*)*(subcnt+1));
2013 for (i = 0; i < subcnt; i++) {
2014 result2[i] = result[i].fname;
2016 result2[subcnt] = NULL;
2018 free(result);
2020 return result2;
2023 void list_sub_file(sub_data* subd){
2024 int i,j;
2025 subtitle *subs = subd->subtitles;
2027 for(j=0; j < subd->sub_num; j++){
2028 subtitle* egysub=&subs[j];
2029 mp_msg(MSGT_SUBREADER,MSGL_INFO,"%i line%c (%li-%li)\n",
2030 egysub->lines,
2031 (1==egysub->lines)?' ':'s',
2032 egysub->start,
2033 egysub->end);
2034 for (i=0; i<egysub->lines; i++) {
2035 mp_msg(MSGT_SUBREADER,MSGL_INFO,"\t\t%d: %s%s", i,egysub->text[i], i==egysub->lines-1?"":" \n ");
2037 mp_msg(MSGT_SUBREADER,MSGL_INFO,"\n");
2040 mp_msg(MSGT_SUBREADER,MSGL_INFO,"Subtitle format %s time.\n",
2041 subd->sub_uses_time ? "uses":"doesn't use");
2042 mp_msg(MSGT_SUBREADER,MSGL_INFO,"Read %i subtitles, %i errors.\n", subd->sub_num, subd->sub_errs);
2045 void dump_srt(sub_data* subd, float fps){
2046 int i,j;
2047 int h,m,s,ms;
2048 FILE * fd;
2049 subtitle * onesub;
2050 unsigned long temp;
2051 subtitle *subs = subd->subtitles;
2053 if (!subd->sub_uses_time && sub_fps == 0)
2054 sub_fps = fps;
2055 fd=fopen("dumpsub.srt","w");
2056 if(!fd)
2058 perror("dump_srt: fopen");
2059 return;
2061 for(i=0; i < subd->sub_num; i++)
2063 onesub=subs+i; //=&subs[i];
2064 fprintf(fd,"%d\n",i+1);//line number
2066 temp=onesub->start;
2067 if (!subd->sub_uses_time)
2068 temp = temp * 100 / sub_fps;
2069 temp -= sub_delay * 100;
2070 h=temp/360000;temp%=360000; //h =1*100*60*60
2071 m=temp/6000; temp%=6000; //m =1*100*60
2072 s=temp/100; temp%=100; //s =1*100
2073 ms=temp*10; //ms=1*10
2074 fprintf(fd,"%02d:%02d:%02d,%03d --> ",h,m,s,ms);
2076 temp=onesub->end;
2077 if (!subd->sub_uses_time)
2078 temp = temp * 100 / sub_fps;
2079 temp -= sub_delay * 100;
2080 h=temp/360000;temp%=360000;
2081 m=temp/6000; temp%=6000;
2082 s=temp/100; temp%=100;
2083 ms=temp*10;
2084 fprintf(fd,"%02d:%02d:%02d,%03d\n",h,m,s,ms);
2086 for(j=0;j<onesub->lines;j++)
2087 fprintf(fd,"%s\n",onesub->text[j]);
2089 fprintf(fd,"\n");
2091 fclose(fd);
2092 mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.srt\'.\n");
2095 void dump_mpsub(sub_data* subd, float fps){
2096 int i,j;
2097 FILE *fd;
2098 float a,b;
2099 subtitle *subs = subd->subtitles;
2101 mpsub_position = subd->sub_uses_time? (sub_delay*100) : (sub_delay*fps);
2102 if (sub_fps==0) sub_fps=fps;
2104 fd=fopen ("dump.mpsub", "w");
2105 if (!fd) {
2106 perror ("dump_mpsub: fopen");
2107 return;
2111 if (subd->sub_uses_time) fprintf (fd,"FORMAT=TIME\n\n");
2112 else fprintf (fd, "FORMAT=%5.2f\n\n", fps);
2114 for(j=0; j < subd->sub_num; j++){
2115 subtitle* egysub=&subs[j];
2116 if (subd->sub_uses_time) {
2117 a=((egysub->start-mpsub_position)/100.0);
2118 b=((egysub->end-egysub->start)/100.0);
2119 if ( (float)((int)a) == a)
2120 fprintf (fd, "%.0f",a);
2121 else
2122 fprintf (fd, "%.2f",a);
2124 if ( (float)((int)b) == b)
2125 fprintf (fd, " %.0f\n",b);
2126 else
2127 fprintf (fd, " %.2f\n",b);
2128 } else {
2129 fprintf (fd, "%ld %ld\n", (long)((egysub->start*(fps/sub_fps))-((mpsub_position*(fps/sub_fps)))),
2130 (long)(((egysub->end)-(egysub->start))*(fps/sub_fps)));
2133 mpsub_position = egysub->end;
2134 for (i=0; i<egysub->lines; i++) {
2135 fprintf (fd, "%s\n",egysub->text[i]);
2137 fprintf (fd, "\n");
2139 fclose (fd);
2140 mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dump.mpsub\'.\n");
2143 void dump_microdvd(sub_data* subd, float fps) {
2144 int i, delay;
2145 FILE *fd;
2146 subtitle *subs = subd->subtitles;
2147 if (sub_fps == 0)
2148 sub_fps = fps;
2149 fd = fopen("dumpsub.sub", "w");
2150 if (!fd) {
2151 perror("dumpsub.sub: fopen");
2152 return;
2154 delay = sub_delay * sub_fps;
2155 for (i = 0; i < subd->sub_num; ++i) {
2156 int j, start, end;
2157 start = subs[i].start;
2158 end = subs[i].end;
2159 if (subd->sub_uses_time) {
2160 start = start * sub_fps / 100 ;
2161 end = end * sub_fps / 100;
2163 else {
2164 start = start * sub_fps / fps;
2165 end = end * sub_fps / fps;
2167 start -= delay;
2168 end -= delay;
2169 fprintf(fd, "{%d}{%d}", start, end);
2170 for (j = 0; j < subs[i].lines; ++j)
2171 fprintf(fd, "%s%s", j ? "|" : "", subs[i].text[j]);
2172 fprintf(fd, "\n");
2174 fclose(fd);
2175 mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.sub\'.\n");
2178 void dump_jacosub(sub_data* subd, float fps) {
2179 int i,j;
2180 int h,m,s,cs;
2181 FILE * fd;
2182 subtitle * onesub;
2183 unsigned long temp;
2184 subtitle *subs = subd->subtitles;
2186 if (!subd->sub_uses_time && sub_fps == 0)
2187 sub_fps = fps;
2188 fd=fopen("dumpsub.jss","w");
2189 if(!fd)
2191 perror("dump_jacosub: fopen");
2192 return;
2194 fprintf(fd, "#TIMERES %d\n", (subd->sub_uses_time) ? 100 : (int)sub_fps);
2195 for(i=0; i < subd->sub_num; i++)
2197 onesub=subs+i; //=&subs[i];
2199 temp=onesub->start;
2200 if (!subd->sub_uses_time)
2201 temp = temp * 100 / sub_fps;
2202 temp -= sub_delay * 100;
2203 h=temp/360000;temp%=360000; //h =1*100*60*60
2204 m=temp/6000; temp%=6000; //m =1*100*60
2205 s=temp/100; temp%=100; //s =1*100
2206 cs=temp; //cs=1*10
2207 fprintf(fd,"%02d:%02d:%02d.%02d ",h,m,s,cs);
2209 temp=onesub->end;
2210 if (!subd->sub_uses_time)
2211 temp = temp * 100 / sub_fps;
2212 temp -= sub_delay * 100;
2213 h=temp/360000;temp%=360000;
2214 m=temp/6000; temp%=6000;
2215 s=temp/100; temp%=100;
2216 cs=temp;
2217 fprintf(fd,"%02d:%02d:%02d.%02d {~} ",h,m,s,cs);
2219 for(j=0;j<onesub->lines;j++)
2220 fprintf(fd,"%s%s",j ? "\\n" : "", onesub->text[j]);
2222 fprintf(fd,"\n");
2224 fclose(fd);
2225 mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.js\'.\n");
2228 void dump_sami(sub_data* subd, float fps) {
2229 int i,j;
2230 FILE * fd;
2231 subtitle * onesub;
2232 unsigned long temp;
2233 subtitle *subs = subd->subtitles;
2235 if (!subd->sub_uses_time && sub_fps == 0)
2236 sub_fps = fps;
2237 fd=fopen("dumpsub.smi","w");
2238 if(!fd)
2240 perror("dump_jacosub: fopen");
2241 return;
2243 fprintf(fd, "<SAMI>\n"
2244 "<HEAD>\n"
2245 " <STYLE TYPE=\"Text/css\">\n"
2246 " <!--\n"
2247 " 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"
2248 " .SUBTTL {Name: 'Subtitles'; Lang: en-US; SAMIType: CC;}\n"
2249 " -->\n"
2250 " </STYLE>\n"
2251 "</HEAD>\n"
2252 "<BODY>\n");
2253 for(i=0; i < subd->sub_num; i++)
2255 onesub=subs+i; //=&subs[i];
2257 temp=onesub->start;
2258 if (!subd->sub_uses_time)
2259 temp = temp * 100 / sub_fps;
2260 temp -= sub_delay * 100;
2261 fprintf(fd,"\t<SYNC Start=%lu>\n"
2262 "\t <P>", temp * 10);
2264 for(j=0;j<onesub->lines;j++)
2265 fprintf(fd,"%s%s",j ? "<br>" : "", onesub->text[j]);
2267 fprintf(fd,"\n");
2269 temp=onesub->end;
2270 if (!subd->sub_uses_time)
2271 temp = temp * 100 / sub_fps;
2272 temp -= sub_delay * 100;
2273 fprintf(fd,"\t<SYNC Start=%lu>\n"
2274 "\t <P>&nbsp;\n", temp * 10);
2276 fprintf(fd, "</BODY>\n"
2277 "</SAMI>\n");
2278 fclose(fd);
2279 mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.smi\'.\n");
2282 void sub_free( sub_data * subd )
2284 int i, j;
2286 if ( !subd ) return;
2288 for (i = 0; i < subd->sub_num; i++)
2289 for (j = 0; j < subd->subtitles[i].lines; j++)
2290 free( subd->subtitles[i].text[j] );
2291 free( subd->subtitles );
2292 free( subd->filename );
2293 free( subd );
2296 #define MAX_SUBLINE 512
2298 * \brief parse text and append it to subtitle in sub
2299 * \param sub subtitle struct to add text to
2300 * \param txt text to parse
2301 * \param len length of text in txt
2302 * \param endpts pts at which this subtitle text should be removed again
2304 * <> and {} are interpreted as comment delimiters, "\n", "\N", '\n', '\r'
2305 * and '\0' are interpreted as newlines, duplicate, leading and trailing
2306 * newlines are ignored.
2308 void sub_add_text(subtitle *sub, const char *txt, int len, double endpts) {
2309 int comment = 0;
2310 int double_newline = 1; // ignore newlines at the beginning
2311 int i, pos;
2312 char *buf;
2313 int orig_lines = sub->lines;
2314 if (sub->lines >= SUB_MAX_TEXT) return;
2315 pos = 0;
2316 buf = malloc(MAX_SUBLINE + 1);
2317 sub->text[sub->lines] = buf;
2318 sub->endpts[sub->lines] = endpts;
2319 for (i = 0; i < len && pos < MAX_SUBLINE; i++) {
2320 char c = txt[i];
2321 if (c == '<') comment |= 1;
2322 if (c == '{') comment |= 2;
2323 if (comment) {
2324 if (c == '}') comment &= ~2;
2325 if (c == '>') comment &= ~1;
2326 continue;
2328 if (pos == MAX_SUBLINE - 1) {
2329 i--;
2330 c = 0;
2332 if (c == '\\' && i + 1 < len) {
2333 c = txt[++i];
2334 if (c == 'n' || c == 'N') c = 0;
2336 if (c == '\n' || c == '\r') c = 0;
2337 if (c) {
2338 double_newline = 0;
2339 buf[pos++] = c;
2340 } else if (!double_newline) {
2341 if (sub->lines >= SUB_MAX_TEXT - 1) {
2342 mp_msg(MSGT_VO, MSGL_WARN, "Too many subtitle lines\n");
2343 break;
2345 double_newline = 1;
2346 buf[pos] = 0;
2347 sub->lines++;
2348 pos = 0;
2349 buf = malloc(MAX_SUBLINE + 1);
2350 sub->text[sub->lines] = buf;
2351 sub->endpts[sub->lines] = endpts;
2354 buf[pos] = 0;
2355 if (sub->lines < SUB_MAX_TEXT &&
2356 strlen(sub->text[sub->lines]))
2357 sub->lines++;
2358 #ifdef CONFIG_FRIBIDI
2359 sub = sub_fribidi(sub, sub_utf8, orig_lines);
2360 #endif
2363 #define MP_NOPTS_VALUE (-1LL<<63)
2365 * \brief remove outdated subtitle lines.
2366 * \param sub subtitle struct to modify
2367 * \param pts current pts. All lines with endpts <= this will be removed.
2368 * Use MP_NOPTS_VALUE to remove all lines
2369 * \return 1 if sub was modified, 0 otherwise.
2371 int sub_clear_text(subtitle *sub, double pts) {
2372 int i = 0;
2373 int changed = 0;
2374 while (i < sub->lines) {
2375 double endpts = sub->endpts[i];
2376 if (pts == MP_NOPTS_VALUE || (endpts != MP_NOPTS_VALUE && pts >= endpts)) {
2377 int j;
2378 free(sub->text[i]);
2379 for (j = i + 1; j < sub->lines; j++) {
2380 sub->text[j - 1] = sub->text[j];
2381 sub->endpts[j - 1] = sub->endpts[j];
2383 sub->lines--;
2384 changed = 1;
2385 } else
2386 i++;
2388 return changed;