subreader.c

   1 /*
   2  * Subtitle reader with format autodetection
   3  *
   4  * Written by laaz
   5  * Some code cleanup & realloc() by A'rpi/ESP-team
   6  *
   7  */
   8
   9
  10 #include <stdio.h>
  11 #include <stdlib.h>
  12 #include <string.h>
  13 #include <ctype.h>
  14
  15 #include <sys/types.h>
  16 #include <dirent.h>
  17
  18 #include "config.h"
  19 #include "mp_msg.h"
  20 #include "subreader.h"
  21
  22 #ifdef HAVE_ENCA
  23 #include <enca.h>
  24 #endif
  25
  26 #define ERR ((void *) -1)
  27
  28 #ifdef USE_ICONV
  29 #include <iconv.h>
  30 char *sub_cp=NULL;
  31 #endif
  32 #ifdef USE_FRIBIDI
  33 #include <fribidi/fribidi.h>
  34 char *fribidi_charset = NULL;   ///character set that will be passed to FriBiDi
  35 int flip_hebrew = 1;            ///flip subtitles using fribidi
  36 int fribidi_flip_commas = 0;    ///flip comma when fribidi is used
  37 #endif
  38
  39 extern char* dvdsub_lang;
  40
  41 /* Maximal length of line of a subtitle */
  42 #define LINE_LEN 1000
  43 static float mpsub_position=0;
  44 static float mpsub_multiplier=1.;
  45 static int sub_slacktime = 20000; //20 sec
  46
  47 int sub_no_text_pp=0;   // 1 => do not apply text post-processing
  48                         // like {\...} elimination in SSA format.
  49
  50 int sub_match_fuzziness=0; // level of sub name matching fuzziness
  51
  52 /* Use the SUB_* constant defined in the header file */
  53 int sub_format=SUB_INVALID;
  54 #ifdef USE_SORTSUB
  55 /*
  56    Some subtitling formats, namely AQT and Subrip09, define the end of a
  57    subtitle as the beginning of the following. Since currently we read one
  58    subtitle at time, for these format we keep two global *subtitle,
  59    previous_aqt_sub and previous_subrip09_sub, pointing to previous subtitle,
  60    so we can change its end when we read current subtitle starting time.
  61    When USE_SORTSUB is defined, we use a single global unsigned long,
  62    previous_sub_end, for both (and even future) formats, to store the end of
  63    the previous sub: it is initialized to 0 in sub_read_file and eventually
  64    modified by sub_read_aqt_line or sub_read_subrip09_line.
  65  */
  66 unsigned long previous_sub_end;
  67 #endif
  68
  69 static int eol(char p) {
  70     return (p=='\r' || p=='\n' || p=='\0');
  71 }
  72
  73 /* Remove leading and trailing space */
  74 static void trail_space(char *s) {
  75         int i = 0;
  76         while (isspace(s[i])) ++i;
  77         if (i) strcpy(s, s + i);
  78         i = strlen(s) - 1;
  79         while (i > 0 && isspace(s[i])) s[i--] = '\0';
  80 }
  81
  82 static char *stristr(const char *haystack, const char *needle) {
  83     int len = 0;
  84     const char *p = haystack;
  85
  86     if (!(haystack && needle)) return NULL;
  87
  88     len=strlen(needle);
  89     while (*p != '\0') {
  90         if (strncasecmp(p, needle, len) == 0) return (char*)p;
  91         p++;
  92     }
  93
  94     return NULL;
  95 }
  96
  97 subtitle *sub_read_line_sami(FILE *fd, subtitle *current) {
  98     static char line[LINE_LEN+1];
  99     static char *s = NULL, *slacktime_s;
 100     char text[LINE_LEN+1], *p=NULL, *q;
 101     int state;
 102
 103     current->lines = current->start = current->end = 0;
 104     current->alignment = SUB_ALIGNMENT_BOTTOMCENTER;
 105     state = 0;
 106
 107     /* read the first line */
 108     if (!s)
 109             if (!(s = fgets(line, LINE_LEN, fd))) return 0;
 110
 111     do {
 112         switch (state) {
 113
 114         case 0: /* find "START=" or "Slacktime:" */
 115             slacktime_s = stristr (s, "Slacktime:");
 116             if (slacktime_s)
 117                 sub_slacktime = strtol (slacktime_s+10, NULL, 0) / 10;
 118
 119             s = stristr (s, "Start=");
 120             if (s) {
 121                 current->start = strtol (s + 6, &s, 0) / 10;
 122                 /* eat '>' */
 123                 for (; *s != '>' && *s != '\0'; s++);
 124                 s++;
 125                 state = 1; continue;
 126             }
 127             break;
 128
 129         case 1: /* find (optionnal) "<P", skip other TAGs */
 130             for  (; *s == ' ' || *s == '\t'; s++); /* strip blanks, if any */
 131             if (*s == '\0') break;
 132             if (*s != '<') { state = 3; p = text; continue; } /* not a TAG */
 133             s++;
 134             if (*s == 'P' || *s == 'p') { s++; state = 2; continue; } /* found '<P' */
 135             for (; *s != '>' && *s != '\0'; s++); /* skip remains of non-<P> TAG */
 136             if (s == '\0')
 137               break;
 138             s++;
 139             continue;
 140
 141         case 2: /* find ">" */
 142             if ((s = strchr (s, '>'))) { s++; state = 3; p = text; continue; }
 143             break;
 144
 145         case 3: /* get all text until '<' appears */
 146             if (*s == '\0') break;
 147             else if (!strncasecmp (s, "<br>", 4)) {
 148                 *p = '\0'; p = text; trail_space (text);
 149                 if (text[0] != '\0')
 150                     current->text[current->lines++] = strdup (text);
 151                 s += 4;
 152             }
 153             else if ((*s == '{') && !sub_no_text_pp) { state = 5; ++s; continue; }
 154             else if (*s == '<') { state = 4; }
 155             else if (!strncasecmp (s, "&nbsp;", 6)) { *p++ = ' '; s += 6; }
 156             else if (*s == '\t') { *p++ = ' '; s++; }
 157             else if (*s == '\r' || *s == '\n') { s++; }
 158             else *p++ = *s++;
 159
 160             /* skip duplicated space */
 161             if (p > text + 2) if (*(p-1) == ' ' && *(p-2) == ' ') p--;
 162
 163             continue;
 164
 165         case 4: /* get current->end or skip <TAG> */
 166             q = stristr (s, "Start=");
 167             if (q) {
 168                 current->end = strtol (q + 6, &q, 0) / 10 - 1;
 169                 *p = '\0'; trail_space (text);
 170                 if (text[0] != '\0')
 171                     current->text[current->lines++] = strdup (text);
 172                 if (current->lines > 0) { state = 99; break; }
 173                 state = 0; continue;
 174             }
 175             s = strchr (s, '>');
 176             if (s) { s++; state = 3; continue; }
 177             break;
 178        case 5: /* get rid of {...} text, but read the alignment code */
 179             if ((*s == '\\') && (*(s + 1) == 'a') && !sub_no_text_pp) {
 180                if (stristr(s, "\\a1") != NULL) {
 181                    current->alignment = SUB_ALIGNMENT_BOTTOMLEFT;
 182                    s = s + 3;
 183                }
 184                if (stristr(s, "\\a2") != NULL) {
 185                    current->alignment = SUB_ALIGNMENT_BOTTOMCENTER;
 186                    s = s + 3;
 187                } else if (stristr(s, "\\a3") != NULL) {
 188                    current->alignment = SUB_ALIGNMENT_BOTTOMRIGHT;
 189                    s = s + 3;
 190                } else if ((stristr(s, "\\a4") != NULL) || (stristr(s, "\\a5") != NULL) || (stristr(s, "\\a8") != NULL)) {
 191                    current->alignment = SUB_ALIGNMENT_TOPLEFT;
 192                    s = s + 3;
 193                } else if (stristr(s, "\\a6") != NULL) {
 194                    current->alignment = SUB_ALIGNMENT_TOPCENTER;
 195                    s = s + 3;
 196                } else if (stristr(s, "\\a7") != NULL) {
 197                    current->alignment = SUB_ALIGNMENT_TOPRIGHT;
 198                    s = s + 3;
 199                } else if (stristr(s, "\\a9") != NULL) {
 200                    current->alignment = SUB_ALIGNMENT_MIDDLELEFT;
 201                    s = s + 3;
 202                } else if (stristr(s, "\\a10") != NULL) {
 203                    current->alignment = SUB_ALIGNMENT_MIDDLECENTER;
 204                    s = s + 4;
 205                } else if (stristr(s, "\\a11") != NULL) {
 206                    current->alignment = SUB_ALIGNMENT_MIDDLERIGHT;
 207                    s = s + 4;
 208                }
 209             }
 210             if (*s == '}') state = 3;
 211             ++s;
 212             continue;
 213         }
 214
 215         /* read next line */
 216         if (state != 99 && !(s = fgets (line, LINE_LEN, fd))) {
 217             if (current->start > 0) {
 218                 break; // if it is the last subtitle
 219             } else {
 220                 return 0;
 221             }
 222         }
 223
 224     } while (state != 99);
 225
 226     // For the last subtitle
 227     if (current->end <= 0) {
 228         current->end = current->start + sub_slacktime;
 229         *p = '\0'; trail_space (text);
 230         if (text[0] != '\0')
 231             current->text[current->lines++] = strdup (text);
 232     }
 233
 234     return current;
 235 }
 236
 237
 238 char *sub_readtext(char *source, char **dest) {
 239     int len=0;
 240     char *p=source;
 241
 242 //    printf("src=%p  dest=%p  \n",source,dest);
 243
 244     while ( !eol(*p) && *p!= '|' ) {
 245         p++,len++;
 246     }
 247
 248     *dest= (char *)malloc (len+1);
 249     if (!dest) {return ERR;}
 250
 251     strncpy(*dest, source, len);
 252     (*dest)[len]=0;
 253
 254     while (*p=='\r' || *p=='\n' || *p=='|') p++;
 255
 256     if (*p) return p;  // not-last text field
 257     else return NULL;  // last text field
 258 }
 259
 260 subtitle *sub_read_line_microdvd(FILE *fd,subtitle *current) {
 261     char line[LINE_LEN+1];
 262     char line2[LINE_LEN+1];
 263     char *p, *next;
 264     int i;
 265
 266     do {
 267         if (!fgets (line, LINE_LEN, fd)) return NULL;
 268     } while ((sscanf (line,
 269                       "{%ld}{}%[^\r\n]",
 270                       &(current->start), line2) < 2) &&
 271              (sscanf (line,
 272                       "{%ld}{%ld}%[^\r\n]",
 273                       &(current->start), &(current->end), line2) < 3));
 274
 275     p=line2;
 276
 277     next=p, i=0;
 278     while ((next =sub_readtext (next, &(current->text[i])))) {
 279         if (current->text[i]==ERR) {return ERR;}
 280         i++;
 281         if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
 282     }
 283     current->lines= ++i;
 284
 285     return current;
 286 }
 287
 288 subtitle *sub_read_line_mpl2(FILE *fd,subtitle *current) {
 289     char line[LINE_LEN+1];
 290     char line2[LINE_LEN+1];
 291     char *p, *next;
 292     int i;
 293
 294     do {
 295         if (!fgets (line, LINE_LEN, fd)) return NULL;
 296     } while ((sscanf (line,
 297                       "[%ld][%ld]%[^\r\n]",
 298                       &(current->start), &(current->end), line2) < 3));
 299     current->start *= 10;
 300     current->end *= 10;
 301     p=line2;
 302
 303     next=p, i=0;
 304     while ((next =sub_readtext (next, &(current->text[i])))) {
 305         if (current->text[i]==ERR) {return ERR;}
 306         i++;
 307         if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
 308     }
 309     current->lines= ++i;
 310
 311     return current;
 312 }
 313
 314 subtitle *sub_read_line_subrip(FILE *fd, subtitle *current) {
 315     char line[LINE_LEN+1];
 316     int a1,a2,a3,a4,b1,b2,b3,b4;
 317     char *p=NULL, *q=NULL;
 318     int len;
 319
 320     while (1) {
 321         if (!fgets (line, LINE_LEN, fd)) return NULL;
 322         if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4) < 8) continue;
 323         current->start = a1*360000+a2*6000+a3*100+a4;
 324         current->end   = b1*360000+b2*6000+b3*100+b4;
 325
 326         if (!fgets (line, LINE_LEN, fd)) return NULL;
 327
 328         p=q=line;
 329         for (current->lines=1; current->lines < SUB_MAX_TEXT; current->lines++) {
 330             for (q=p,len=0; *p && *p!='\r' && *p!='\n' && *p!='|' && strncmp(p,"[br]",4); p++,len++);
 331             current->text[current->lines-1]=(char *)malloc (len+1);
 332             if (!current->text[current->lines-1]) return ERR;
 333             strncpy (current->text[current->lines-1], q, len);
 334             current->text[current->lines-1][len]='\0';
 335             if (!*p || *p=='\r' || *p=='\n') break;
 336             if (*p=='|') p++;
 337             else while (*p++!=']');
 338         }
 339         break;
 340     }
 341     return current;
 342 }
 343
 344 subtitle *sub_read_line_subviewer(FILE *fd,subtitle *current) {
 345     char line[LINE_LEN+1];
 346     int a1,a2,a3,a4,b1,b2,b3,b4;
 347     char *p=NULL;
 348     int i,len;
 349
 350     while (!current->text[0]) {
 351         if (!fgets (line, LINE_LEN, fd)) return NULL;
 352         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)
 353             continue;
 354         current->start = a1*360000+a2*6000+a3*100+a4/10;
 355         current->end   = b1*360000+b2*6000+b3*100+b4/10;
 356         for (i=0; i<SUB_MAX_TEXT;) {
 357             if (!fgets (line, LINE_LEN, fd)) break;
 358             len=0;
 359             for (p=line; *p!='\n' && *p!='\r' && *p; p++,len++);
 360             if (len) {
 361                 int j=0,skip=0;
 362                 char *curptr=current->text[i]=(char *)malloc (len+1);
 363                 if (!current->text[i]) return ERR;
 364                 //strncpy (current->text[i], line, len); current->text[i][len]='\0';
 365                 for(; j<len; j++) {
 366                     /* let's filter html tags ::atmos */
 367                     if(line[j]=='>') {
 368                         skip=0;
 369                         continue;
 370                     }
 371                     if(line[j]=='<') {
 372                         skip=1;
 373                         continue;
 374                     }
 375                     if(skip) {
 376                         continue;
 377                     }
 378                     *curptr=line[j];
 379                     curptr++;
 380                 }
 381                 *curptr='\0';
 382
 383                 i++;
 384             } else {
 385                 break;
 386             }
 387         }
 388         current->lines=i;
 389     }
 390     return current;
 391 }
 392
 393 subtitle *sub_read_line_subviewer2(FILE *fd,subtitle *current) {
 394     char line[LINE_LEN+1];
 395     int a1,a2,a3,a4;
 396     char *p=NULL;
 397     int i,len;
 398
 399     while (!current->text[0]) {
 400         if (!fgets (line, LINE_LEN, fd)) return NULL;
 401         if (line[0]!='{')
 402             continue;
 403         if ((len=sscanf (line, "{T %d:%d:%d:%d",&a1,&a2,&a3,&a4)) < 4)
 404             continue;
 405         current->start = a1*360000+a2*6000+a3*100+a4/10;
 406         for (i=0; i<SUB_MAX_TEXT;) {
 407             if (!fgets (line, LINE_LEN, fd)) break;
 408             if (line[0]=='}') break;
 409             len=0;
 410             for (p=line; *p!='\n' && *p!='\r' && *p; ++p,++len);
 411             if (len) {
 412                 current->text[i]=(char *)malloc (len+1);
 413                 if (!current->text[i]) return ERR;
 414                 strncpy (current->text[i], line, len); current->text[i][len]='\0';
 415                 ++i;
 416             } else {
 417                 break;
 418             }
 419         }
 420         current->lines=i;
 421     }
 422     return current;
 423 }
 424
 425
 426 subtitle *sub_read_line_vplayer(FILE *fd,subtitle *current) {
 427         char line[LINE_LEN+1];
 428         int a1,a2,a3;
 429         char *p=NULL, *next,separator;
 430         int i,len,plen;
 431
 432         while (!current->text[0]) {
 433                 if (!fgets (line, LINE_LEN, fd)) return NULL;
 434                 if ((len=sscanf (line, "%d:%d:%d%c%n",&a1,&a2,&a3,&separator,&plen)) < 4)
 435                         continue;
 436
 437                 if (!(current->start = a1*360000+a2*6000+a3*100))
 438                         continue;
 439                 /* removed by wodzu
 440                 p=line;
 441                 // finds the body of the subtitle
 442                 for (i=0; i<3; i++){
 443                    p=strchr(p,':');
 444                    if (p==NULL) break;
 445                    ++p;
 446                 }
 447                 if (p==NULL) {
 448                     printf("SUB: Skipping incorrect subtitle line!\n");
 449                     continue;
 450                 }
 451                 */
 452                 // by wodzu: hey! this time we know what length it has! what is
 453                 // that magic for? it can't deal with space instead of third
 454                 // colon! look, what simple it can be:
 455                 p = &line[ plen ];
 456
 457                 i=0;
 458                 if (*p!='|') {
 459                         //
 460                         next = p,i=0;
 461                         while ((next =sub_readtext (next, &(current->text[i])))) {
 462                                 if (current->text[i]==ERR) {return ERR;}
 463                                 i++;
 464                                 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
 465                         }
 466                         current->lines=i+1;
 467                 }
 468         }
 469         return current;
 470 }
 471
 472 subtitle *sub_read_line_rt(FILE *fd,subtitle *current) {
 473         //TODO: This format uses quite rich (sub/super)set of xhtml
 474         // I couldn't check it since DTD is not included.
 475         // WARNING: full XML parses can be required for proper parsing
 476     char line[LINE_LEN+1];
 477     int a1,a2,a3,a4,b1,b2,b3,b4;
 478     char *p=NULL,*next=NULL;
 479     int i,len,plen;
 480
 481     while (!current->text[0]) {
 482         if (!fgets (line, LINE_LEN, fd)) return NULL;
 483         //TODO: it seems that format of time is not easily determined, it may be 1:12, 1:12.0 or 0:1:12.0
 484         //to describe the same moment in time. Maybe there are even more formats in use.
 485         //if ((len=sscanf (line, "<Time Begin=\"%d:%d:%d.%d\" End=\"%d:%d:%d.%d\"",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8)
 486         plen=a1=a2=a3=a4=b1=b2=b3=b4=0;
 487         if (
 488         ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d.%d\" %*[Ee]nd=\"%d.%d\"%*[^<]<clear/>%n",&a3,&a4,&b3,&b4,&plen)) < 4) &&
 489         ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d.%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a3,&a4,&b2,&b3,&b4,&plen)) < 5) &&
 490         ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&plen)) < 4) &&
 491         ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\" %*[Ee]nd=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&b2,&b3,&b4,&plen)) < 5) &&
 492 //      ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\" %*[Ee]nd=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&b2,&b3,&plen)) < 5) &&
 493         ((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) &&
 494         ((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) &&
 495         //now try it without end time
 496         ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d.%d\"%*[^<]<clear/>%n",&a3,&a4,&plen)) < 2) &&
 497         ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&plen)) < 2) &&
 498         ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&plen)) < 3) &&
 499         ((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d:%d.%d\"%*[^<]<clear/>%n",&a1,&a2,&a3,&a4,&plen)) < 4)
 500         )
 501             continue;
 502         current->start = a1*360000+a2*6000+a3*100+a4/10;
 503         current->end   = b1*360000+b2*6000+b3*100+b4/10;
 504         if (b1 == 0 && b2 == 0 && b3 == 0 && b4 == 0)
 505           current->end = current->start+200;
 506         p=line; p+=plen;i=0;
 507         // TODO: I don't know what kind of convention is here for marking multiline subs, maybe <br/> like in xml?
 508         next = strstr(line,"<clear/>");
 509         if(next && strlen(next)>8){
 510           next+=8;i=0;
 511           while ((next =sub_readtext (next, &(current->text[i])))) {
 512                 if (current->text[i]==ERR) {return ERR;}
 513                 i++;
 514                 if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
 515           }
 516         }
 517                         current->lines=i+1;
 518     }
 519     return current;
 520 }
 521
 522 subtitle *sub_read_line_ssa(FILE *fd,subtitle *current) {
 523 /*
 524  * Sub Station Alpha v4 (and v2?) scripts have 9 commas before subtitle
 525  * other Sub Station Alpha scripts have only 8 commas before subtitle
 526  * Reading the "ScriptType:" field is not reliable since many scripts appear
 527  * w/o it
 528  *
 529  * http://www.scriptclub.org is a good place to find more examples
 530  * http://www.eswat.demon.co.uk is where the SSA specs can be found
 531  */
 532         int comma;
 533         static int max_comma = 32; /* let's use 32 for the case that the */
 534                     /*  amount of commas increase with newer SSA versions */
 535
 536         int hour1, min1, sec1, hunsec1,
 537             hour2, min2, sec2, hunsec2, nothing;
 538         int num;
 539
 540         char line[LINE_LEN+1],
 541              line3[LINE_LEN+1],
 542              *line2;
 543         char *tmp;
 544
 545         do {
 546                 if (!fgets (line, LINE_LEN, fd)) return NULL;
 547         } while (sscanf (line, "Dialogue: Marked=%d,%d:%d:%d.%d,%d:%d:%d.%d,"
 548                         "%[^\n\r]", &nothing,
 549                         &hour1, &min1, &sec1, &hunsec1,
 550                         &hour2, &min2, &sec2, &hunsec2,
 551                         line3) < 9
 552                  &&
 553                  sscanf (line, "Dialogue: %d,%d:%d:%d.%d,%d:%d:%d.%d,"
 554                          "%[^\n\r]", &nothing,
 555                          &hour1, &min1, &sec1, &hunsec1,
 556                          &hour2, &min2, &sec2, &hunsec2,
 557                          line3) < 9         );
 558
 559         line2=strchr(line3, ',');
 560
 561         for (comma = 4; comma < max_comma; comma ++)
 562           {
 563             tmp = line2;
 564             if(!(tmp=strchr(++tmp, ','))) break;
 565             if(*(++tmp) == ' ') break;
 566                   /* a space after a comma means we're already in a sentence */
 567             line2 = tmp;
 568           }
 569
 570         if(comma < max_comma)max_comma = comma;
 571         /* eliminate the trailing comma */
 572         if(*line2 == ',') line2++;
 573
 574         current->lines=0;num=0;
 575         current->start = 360000*hour1 + 6000*min1 + 100*sec1 + hunsec1;
 576         current->end   = 360000*hour2 + 6000*min2 + 100*sec2 + hunsec2;
 577
 578         while (((tmp=strstr(line2, "\\n")) != NULL) || ((tmp=strstr(line2, "\\N")) != NULL) ){
 579                 current->text[num]=(char *)malloc(tmp-line2+1);
 580                 strncpy (current->text[num], line2, tmp-line2);
 581                 current->text[num][tmp-line2]='\0';
 582                 line2=tmp+2;
 583                 num++;
 584                 current->lines++;
 585                 if (current->lines >=  SUB_MAX_TEXT) return current;
 586         }
 587
 588         current->text[num]=strdup(line2);
 589         current->lines++;
 590
 591         return current;
 592 }
 593
 594 void sub_pp_ssa(subtitle *sub) {
 595         int l=sub->lines;
 596         char *so,*de,*start;
 597
 598         while (l){
 599                 /* eliminate any text enclosed with {}, they are font and color settings */
 600                 so=de=sub->text[--l];
 601                 while (*so) {
 602                         if(*so == '{' && so[1]=='\\') {
 603                                 for (start=so; *so && *so!='}'; so++);
 604                                 if(*so) so++; else so=start;
 605                         }
 606                         if(*so) {
 607                                 *de=*so;
 608                                 so++; de++;
 609                         }
 610                 }
 611                 *de=*so;
 612         }
 613 }
 614
 615 /*
 616  * PJS subtitles reader.
 617  * That's the "Phoenix Japanimation Society" format.
 618  * I found some of them in http://www.scriptsclub.org/ (used for anime).
 619  * The time is in tenths of second.
 620  *
 621  * by set, based on code by szabi (dunnowhat sub format ;-)
 622  */
 623 subtitle *sub_read_line_pjs(FILE *fd,subtitle *current) {
 624     char line[LINE_LEN+1];
 625     char text[LINE_LEN+1], *s, *d;
 626
 627     if (!fgets (line, LINE_LEN, fd))
 628         return NULL;
 629     /* skip spaces */
 630     for (s=line; *s && isspace(*s); s++);
 631     /* allow empty lines at the end of the file */
 632     if (*s==0)
 633         return NULL;
 634     /* get the time */
 635     if (sscanf (s, "%ld,%ld,", &(current->start),
 636                 &(current->end)) <2) {
 637         return ERR;
 638     }
 639     /* the files I have are in tenths of second */
 640     current->start *= 10;
 641     current->end *= 10;
 642     /* walk to the beggining of the string */
 643     for (; *s; s++) if (*s==',') break;
 644     if (*s) {
 645         for (s++; *s; s++) if (*s==',') break;
 646         if (*s) s++;
 647     }
 648     if (*s!='"') {
 649         return ERR;
 650     }
 651     /* copy the string to the text buffer */
 652     for (s++, d=text; *s && *s!='"'; s++, d++)
 653         *d=*s;
 654     *d=0;
 655     current->text[0] = strdup(text);
 656     current->lines = 1;
 657
 658     return current;
 659 }
 660
 661 subtitle *sub_read_line_mpsub(FILE *fd, subtitle *current) {
 662         char line[LINE_LEN+1];
 663         float a,b;
 664         int num=0;
 665         char *p, *q;
 666
 667         do
 668         {
 669                 if (!fgets(line, LINE_LEN, fd)) return NULL;
 670         } while (sscanf (line, "%f %f", &a, &b) !=2);
 671
 672         mpsub_position += a*mpsub_multiplier;
 673         current->start=(int) mpsub_position;
 674         mpsub_position += b*mpsub_multiplier;
 675         current->end=(int) mpsub_position;
 676
 677         while (num < SUB_MAX_TEXT) {
 678                 if (!fgets (line, LINE_LEN, fd)) {
 679                         if (num == 0) return NULL;
 680                         else return current;
 681                 }
 682                 p=line;
 683                 while (isspace(*p)) p++;
 684                 if (eol(*p) && num > 0) return current;
 685                 if (eol(*p)) return NULL;
 686
 687                 for (q=p; !eol(*q); q++);
 688                 *q='\0';
 689                 if (strlen(p)) {
 690                         current->text[num]=strdup(p);
 691 //                      printf (">%s<\n",p);
 692                         current->lines = ++num;
 693                 } else {
 694                         if (num) return current;
 695                         else return NULL;
 696                 }
 697         }
 698         return NULL; // we should have returned before if it's OK
 699 }
 700
 701 #ifndef USE_SORTSUB
 702 //we don't need this if we use previous_sub_end
 703 subtitle *previous_aqt_sub = NULL;
 704 #endif
 705
 706 subtitle *sub_read_line_aqt(FILE *fd,subtitle *current) {
 707     char line[LINE_LEN+1];
 708     char *next;
 709     int i;
 710
 711     while (1) {
 712     // try to locate next subtitle
 713         if (!fgets (line, LINE_LEN, fd))
 714                 return NULL;
 715         if (!(sscanf (line, "-->> %ld", &(current->start)) <1))
 716                 break;
 717     }
 718
 719 #ifdef USE_SORTSUB
 720     previous_sub_end = (current->start) ? current->start - 1 : 0;
 721 #else
 722     if (previous_aqt_sub != NULL)
 723         previous_aqt_sub->end = current->start-1;
 724
 725     previous_aqt_sub = current;
 726 #endif
 727
 728     if (!fgets (line, LINE_LEN, fd))
 729         return NULL;
 730
 731     sub_readtext((char *) &line,&current->text[0]);
 732     current->lines = 1;
 733     current->end = current->start; // will be corrected by next subtitle
 734
 735     if (!fgets (line, LINE_LEN, fd))
 736         return current;
 737
 738     next = line,i=1;
 739     while ((next =sub_readtext (next, &(current->text[i])))) {
 740         if (current->text[i]==ERR) {return ERR;}
 741         i++;
 742         if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
 743         }
 744     current->lines=i+1;
 745
 746     if ((current->text[0]=="") && (current->text[1]=="")) {
 747 #ifdef USE_SORTSUB
 748         previous_sub_end = 0;
 749 #else
 750         // void subtitle -> end of previous marked and exit
 751         previous_aqt_sub = NULL;
 752 #endif
 753         return NULL;
 754         }
 755
 756     return current;
 757 }
 758
 759 #ifndef USE_SORTSUB
 760 subtitle *previous_subrip09_sub = NULL;
 761 #endif
 762
 763 subtitle *sub_read_line_subrip09(FILE *fd,subtitle *current) {
 764     char line[LINE_LEN+1];
 765     int a1,a2,a3;
 766     char * next=NULL;
 767     int i,len;
 768
 769     while (1) {
 770     // try to locate next subtitle
 771         if (!fgets (line, LINE_LEN, fd))
 772                 return NULL;
 773         if (!((len=sscanf (line, "[%d:%d:%d]",&a1,&a2,&a3)) < 3))
 774                 break;
 775     }
 776
 777     current->start = a1*360000+a2*6000+a3*100;
 778
 779 #ifdef USE_SORTSUB
 780     previous_sub_end = (current->start) ? current->start - 1 : 0;
 781 #else
 782     if (previous_subrip09_sub != NULL)
 783         previous_subrip09_sub->end = current->start-1;
 784
 785     previous_subrip09_sub = current;
 786 #endif
 787
 788     if (!fgets (line, LINE_LEN, fd))
 789         return NULL;
 790
 791     next = line,i=0;
 792
 793     current->text[0]=""; // just to be sure that string is clear
 794
 795     while ((next =sub_readtext (next, &(current->text[i])))) {
 796         if (current->text[i]==ERR) {return ERR;}
 797         i++;
 798         if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
 799         }
 800     current->lines=i+1;
 801
 802     if ((current->text[0]=="") && (i==0)) {
 803 #ifdef USE_SORTSUB
 804         previous_sub_end = 0;
 805 #else
 806         // void subtitle -> end of previous marked and exit
 807         previous_subrip09_sub = NULL;
 808 #endif
 809         return NULL;
 810         }
 811
 812     return current;
 813 }
 814
 815 subtitle *sub_read_line_jacosub(FILE * fd, subtitle * current)
 816 {
 817     char line1[LINE_LEN], line2[LINE_LEN], directive[LINE_LEN], *p, *q;
 818     unsigned a1, a2, a3, a4, b1, b2, b3, b4, comment = 0;
 819     static unsigned jacoTimeres = 30;
 820     static int jacoShift = 0;
 821
 822     memset(current, 0, sizeof(subtitle));
 823     memset(line1, 0, LINE_LEN);
 824     memset(line2, 0, LINE_LEN);
 825     memset(directive, 0, LINE_LEN);
 826     while (!current->text[0]) {
 827         if (!fgets(line1, LINE_LEN, fd)) {
 828             return NULL;
 829         }
 830         if (sscanf
 831             (line1, "%u:%u:%u.%u %u:%u:%u.%u %[^\n\r]", &a1, &a2, &a3, &a4,
 832              &b1, &b2, &b3, &b4, line2) < 9) {
 833             if (sscanf(line1, "@%u @%u %[^\n\r]", &a4, &b4, line2) < 3) {
 834                 if (line1[0] == '#') {
 835                     int hours = 0, minutes = 0, seconds, delta, inverter =
 836                         1;
 837                     unsigned units = jacoShift;
 838                     switch (toupper(line1[1])) {
 839                     case 'S':
 840                         if (isalpha(line1[2])) {
 841                             delta = 6;
 842                         } else {
 843                             delta = 2;
 844                         }
 845                         if (sscanf(&line1[delta], "%d", &hours)) {
 846                             if (hours < 0) {
 847                                 hours *= -1;
 848                                 inverter = -1;
 849                             }
 850                             if (sscanf(&line1[delta], "%*d:%d", &minutes)) {
 851                                 if (sscanf
 852                                     (&line1[delta], "%*d:%*d:%d",
 853                                      &seconds)) {
 854                                     sscanf(&line1[delta], "%*d:%*d:%*d.%d",
 855                                            &units);
 856                                 } else {
 857                                     hours = 0;
 858                                     sscanf(&line1[delta], "%d:%d.%d",
 859                                            &minutes, &seconds, &units);
 860                                     minutes *= inverter;
 861                                 }
 862                             } else {
 863                                 hours = minutes = 0;
 864                                 sscanf(&line1[delta], "%d.%d", &seconds,
 865                                        &units);
 866                                 seconds *= inverter;
 867                             }
 868                             jacoShift =
 869                                 ((hours * 3600 + minutes * 60 +
 870                                   seconds) * jacoTimeres +
 871                                  units) * inverter;
 872                         }
 873                         break;
 874                     case 'T':
 875                         if (isalpha(line1[2])) {
 876                             delta = 8;
 877                         } else {
 878                             delta = 2;
 879                         }
 880                         sscanf(&line1[delta], "%u", &jacoTimeres);
 881                         break;
 882                     }
 883                 }
 884                 continue;
 885             } else {
 886                 current->start =
 887                     (unsigned long) ((a4 + jacoShift) * 100.0 /
 888                                      jacoTimeres);
 889                 current->end =
 890                     (unsigned long) ((b4 + jacoShift) * 100.0 /
 891                                      jacoTimeres);
 892             }
 893         } else {
 894             current->start =
 895                 (unsigned
 896                  long) (((a1 * 3600 + a2 * 60 + a3) * jacoTimeres + a4 +
 897                          jacoShift) * 100.0 / jacoTimeres);
 898             current->end =
 899                 (unsigned
 900                  long) (((b1 * 3600 + b2 * 60 + b3) * jacoTimeres + b4 +
 901                          jacoShift) * 100.0 / jacoTimeres);
 902         }
 903         current->lines = 0;
 904         p = line2;
 905         while ((*p == ' ') || (*p == '\t')) {
 906             ++p;
 907         }
 908         if (isalpha(*p)||*p == '[') {
 909             int cont, jLength;
 910
 911             if (sscanf(p, "%s %[^\n\r]", directive, line1) < 2)
 912                 return (subtitle *) ERR;
 913             jLength = strlen(directive);
 914             for (cont = 0; cont < jLength; ++cont) {
 915                 if (isalpha(*(directive + cont)))
 916                     *(directive + cont) = toupper(*(directive + cont));
 917             }
 918             if ((strstr(directive, "RDB") != NULL)
 919                 || (strstr(directive, "RDC") != NULL)
 920                 || (strstr(directive, "RLB") != NULL)
 921                 || (strstr(directive, "RLG") != NULL)) {
 922                 continue;
 923             }
 924             if (strstr(directive, "JL") != NULL) {
 925                 current->alignment = SUB_ALIGNMENT_BOTTOMLEFT;
 926             } else if (strstr(directive, "JR") != NULL) {
 927                 current->alignment = SUB_ALIGNMENT_BOTTOMRIGHT;
 928             } else {
 929                 current->alignment = SUB_ALIGNMENT_BOTTOMCENTER;
 930             }
 931             strcpy(line2, line1);
 932             p = line2;
 933         }
 934         for (q = line1; (!eol(*p)) && (current->lines < SUB_MAX_TEXT); ++p) {
 935             switch (*p) {
 936             case '{':
 937                 comment++;
 938                 break;
 939             case '}':
 940                 if (comment) {
 941                     --comment;
 942                     //the next line to get rid of a blank after the comment
 943                     if ((*(p + 1)) == ' ')
 944                         p++;
 945                 }
 946                 break;
 947             case '~':
 948                 if (!comment) {
 949                     *q = ' ';
 950                     ++q;
 951                 }
 952                 break;
 953             case ' ':
 954             case '\t':
 955                 if ((*(p + 1) == ' ') || (*(p + 1) == '\t'))
 956                     break;
 957                 if (!comment) {
 958                     *q = ' ';
 959                     ++q;
 960                 }
 961                 break;
 962             case '\\':
 963                 if (*(p + 1) == 'n') {
 964                     *q = '\0';
 965                     q = line1;
 966                     current->text[current->lines++] = strdup(line1);
 967                     ++p;
 968                     break;
 969                 }
 970                 if ((toupper(*(p + 1)) == 'C')
 971                     || (toupper(*(p + 1)) == 'F')) {
 972                     ++p,++p;
 973                     break;
 974                 }
 975                 if ((*(p + 1) == 'B') || (*(p + 1) == 'b') || (*(p + 1) == 'D') ||      //actually this means "insert current date here"
 976                     (*(p + 1) == 'I') || (*(p + 1) == 'i') || (*(p + 1) == 'N') || (*(p + 1) == 'T') || //actually this means "insert current time here"
 977                     (*(p + 1) == 'U') || (*(p + 1) == 'u')) {
 978                     ++p;
 979                     break;
 980                 }
 981                 if ((*(p + 1) == '\\') ||
 982                     (*(p + 1) == '~') || (*(p + 1) == '{')) {
 983                     ++p;
 984                 } else if (eol(*(p + 1))) {
 985                     if (!fgets(directive, LINE_LEN, fd))
 986                         return NULL;
 987                     trail_space(directive);
 988                     strncat(line2, directive,
 989                             (LINE_LEN > 511) ? LINE_LEN : 511);
 990                     break;
 991                 }
 992             default:
 993                 if (!comment) {
 994                     *q = *p;
 995                     ++q;
 996                 }
 997             }                   //-- switch
 998         }                       //-- for
 999         *q = '\0';
1000         current->text[current->lines] = strdup(line1);
1001     }                           //-- while
1002     current->lines++;
1003     return current;
1004 }
1005
1006 int sub_autodetect (FILE *fd, int *uses_time) {
1007     char line[LINE_LEN+1];
1008     int i,j=0;
1009     char p;
1010
1011     while (j < 100) {
1012         j++;
1013         if (!fgets (line, LINE_LEN, fd))
1014             return SUB_INVALID;
1015
1016         if (sscanf (line, "{%d}{%d}", &i, &i)==2)
1017                 {*uses_time=0;return SUB_MICRODVD;}
1018         if (sscanf (line, "{%d}{}", &i)==1)
1019                 {*uses_time=0;return SUB_MICRODVD;}
1020         if (sscanf (line, "[%d][%d]", &i, &i)==2)
1021                 {*uses_time=1;return SUB_MPL2;}
1022         if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d",     &i, &i, &i, &i, &i, &i, &i, &i)==8)
1023                 {*uses_time=1;return SUB_SUBRIP;}
1024         if (sscanf (line, "%d:%d:%d%[,.:]%d --> %d:%d:%d%[,.:]%d", &i, &i, &i, (char *)&i, &i, &i, &i, &i, (char *)&i, &i)==10)
1025                 {*uses_time=1;return SUB_SUBVIEWER;}
1026         if (sscanf (line, "{T %d:%d:%d:%d",&i, &i, &i, &i)==4)
1027                 {*uses_time=1;return SUB_SUBVIEWER2;}
1028         if (strstr (line, "<SAMI>"))
1029                 {*uses_time=1; return SUB_SAMI;}
1030         if (sscanf(line, "%d:%d:%d.%d %d:%d:%d.%d", &i, &i, &i, &i, &i, &i, &i, &i) == 8)
1031                 {*uses_time = 1; return SUB_JACOSUB;}
1032         if (sscanf(line, "@%d @%d", &i, &i) == 2)
1033                 {*uses_time = 1; return SUB_JACOSUB;}
1034         if (sscanf (line, "%d:%d:%d:",     &i, &i, &i )==3)
1035                 {*uses_time=1;return SUB_VPLAYER;}
1036         if (sscanf (line, "%d:%d:%d ",     &i, &i, &i )==3)
1037                 {*uses_time=1;return SUB_VPLAYER;}
1038         //TODO: just checking if first line of sub starts with "<" is WAY
1039         // too weak test for RT
1040         // Please someone who knows the format of RT... FIX IT!!!
1041         // It may conflict with other sub formats in the future (actually it doesn't)
1042         if ( *line == '<' )
1043                 {*uses_time=1;return SUB_RT;}
1044
1045         if (!memcmp(line, "Dialogue: Marked", 16))
1046                 {*uses_time=1; return SUB_SSA;}
1047         if (!memcmp(line, "Dialogue: ", 10))
1048                 {*uses_time=1; return SUB_SSA;}
1049         if (sscanf (line, "%d,%d,\"%c", &i, &i, (char *) &i) == 3)
1050                 {*uses_time=1;return SUB_PJS;}
1051         if (sscanf (line, "FORMAT=%d", &i) == 1)
1052                 {*uses_time=0; return SUB_MPSUB;}
1053         if (sscanf (line, "FORMAT=TIM%c", &p)==1 && p=='E')
1054                 {*uses_time=1; return SUB_MPSUB;}
1055         if (strstr (line, "-->>"))
1056                 {*uses_time=0; return SUB_AQTITLE;}
1057         if (sscanf (line, "[%d:%d:%d]", &i, &i, &i)==3)
1058                 {*uses_time=1;return SUB_SUBRIP09;}
1059     }
1060
1061     return SUB_INVALID;  // too many bad lines
1062 }
1063
1064 #ifdef DUMPSUBS
1065 int sub_utf8=0;
1066 #else
1067 extern int sub_utf8;
1068 int sub_utf8_prev=0;
1069 #endif
1070
1071 extern float sub_delay;
1072 extern float sub_fps;
1073
1074 #ifdef USE_ICONV
1075 static iconv_t icdsc = (iconv_t)(-1);
1076
1077 void    subcp_open (FILE *enca_fd)
1078 {
1079         char *tocp = "UTF-8";
1080
1081         if (sub_cp){
1082                 char *cp_tmp = sub_cp;
1083 #ifdef HAVE_ENCA
1084                 char enca_lang[3], enca_fallback[100];
1085                 int free_cp_tmp = 0;
1086                 if (sscanf(sub_cp, "enca:%2s:%99s", enca_lang, enca_fallback) == 2
1087                      || sscanf(sub_cp, "ENCA:%2s:%99s", enca_lang, enca_fallback) == 2) {
1088                   if (enca_fd) {
1089                     cp_tmp = guess_cp(enca_fd, enca_lang, enca_fallback);
1090                     free_cp_tmp = 1;
1091                   } else {
1092                     cp_tmp = enca_fallback;
1093                   }
1094                 }
1095 #endif
1096                 if ((icdsc = iconv_open (tocp, cp_tmp)) != (iconv_t)(-1)){
1097                         mp_msg(MSGT_SUBREADER,MSGL_V,"SUB: opened iconv descriptor.\n");
1098                         sub_utf8 = 2;
1099                 } else
1100                         mp_msg(MSGT_SUBREADER,MSGL_ERR,"SUB: error opening iconv descriptor.\n");
1101 #ifdef HAVE_ENCA
1102                 if (free_cp_tmp && cp_tmp) free(cp_tmp);
1103 #endif
1104         }
1105 }
1106
1107 void    subcp_close (void)
1108 {
1109         if (icdsc != (iconv_t)(-1)){
1110                 (void) iconv_close (icdsc);
1111                 icdsc = (iconv_t)(-1);
1112                 mp_msg(MSGT_SUBREADER,MSGL_V,"SUB: closed iconv descriptor.\n");
1113         }
1114 }
1115
1116 #define ICBUFFSIZE 512
1117 static char icbuffer[ICBUFFSIZE];
1118
1119 subtitle* subcp_recode (subtitle *sub)
1120 {
1121         int l=sub->lines;
1122         size_t ileft, oleft;
1123         char *op, *ip, *ot;
1124
1125         while (l){
1126                 op = icbuffer;
1127                 ip = sub->text[--l];
1128                 ileft = strlen(ip);
1129                 oleft = ICBUFFSIZE - 1;
1130
1131                 if (iconv(icdsc, &ip, &ileft,
1132                           &op, &oleft) == (size_t)(-1)) {
1133                         mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: error recoding line (1).\n");
1134                         l++;
1135                         break;
1136                 }
1137                 if (!(ot = (char *)malloc(op - icbuffer + 1))){
1138                         mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: error allocating mem.\n");
1139                         l++;
1140                         break;
1141                 }
1142                 *op='\0' ;
1143                 strcpy (ot, icbuffer);
1144                 free (sub->text[l]);
1145                 sub->text[l] = ot;
1146         }
1147         if (l){
1148                 for (l = sub->lines; l;)
1149                         free (sub->text[--l]);
1150                 return ERR;
1151         }
1152         return sub;
1153 }
1154
1155 // for demux_ogg.c:
1156 subtitle* subcp_recode1 (subtitle *sub)
1157 {
1158   int l=sub->lines;
1159   size_t ileft, oleft;
1160
1161   if(icdsc == (iconv_t)(-1)) return sub;
1162
1163   while (l){
1164      char *ip = icbuffer;
1165      char *op = sub->text[--l];
1166      strlcpy(ip, op, ICBUFFSIZE);
1167      ileft = strlen(ip);
1168      oleft = ICBUFFSIZE - 1;
1169
1170      if (iconv(icdsc, &ip, &ileft,
1171               &op, &oleft) == (size_t)(-1)) {
1172         mp_msg(MSGT_SUBREADER,MSGL_V,"SUB: error recoding line (2).\n");
1173         return sub;
1174      }
1175      *op='\0' ;
1176   }
1177   return sub;
1178 }
1179 #endif
1180
1181 #ifdef USE_FRIBIDI
1182 #ifndef max
1183 #define max(a,b)  (((a)>(b))?(a):(b))
1184 #endif
1185 subtitle* sub_fribidi (subtitle *sub, int sub_utf8)
1186 {
1187   FriBidiChar logical[LINE_LEN+1], visual[LINE_LEN+1]; // Hopefully these two won't smash the stack
1188   char        *ip      = NULL, *op     = NULL;
1189   FriBidiCharType base;
1190   size_t len,orig_len;
1191   int l=sub->lines;
1192   int char_set_num;
1193   fribidi_boolean log2vis;
1194   if(flip_hebrew) { // Please fix the indentation someday
1195   fribidi_set_mirroring (FRIBIDI_TRUE);
1196   fribidi_set_reorder_nsm (FRIBIDI_FALSE);
1197
1198   if( sub_utf8 == 0 ) {
1199     char_set_num = fribidi_parse_charset (fribidi_charset?fribidi_charset:"ISO8859-8");
1200   }else {
1201     char_set_num = fribidi_parse_charset ("UTF-8");
1202   }
1203   while (l) {
1204     ip = sub->text[--l];
1205     orig_len = len = strlen( ip ); // We assume that we don't use full unicode, only UTF-8 or ISO8859-x
1206     if(len > LINE_LEN) {
1207       mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: sub->text is longer than LINE_LEN.\n");
1208       l++;
1209       break;
1210     }
1211     len = fribidi_charset_to_unicode (char_set_num, ip, len, logical);
1212     base = fribidi_flip_commas?FRIBIDI_TYPE_ON:FRIBIDI_TYPE_L;
1213     log2vis = fribidi_log2vis (logical, len, &base,
1214                                /* output */
1215                                visual, NULL, NULL, NULL);
1216     if(log2vis) {
1217       len = fribidi_remove_bidi_marks (visual, len, NULL, NULL,
1218                                        NULL);
1219       if((op = (char*)malloc(sizeof(char)*(max(2*orig_len,2*len) + 1))) == NULL) {
1220         mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: error allocating mem.\n");
1221         l++;
1222         break;
1223       }
1224       fribidi_unicode_to_charset ( char_set_num, visual, len,op);
1225       free (ip);
1226       sub->text[l] = op;
1227     }
1228   }
1229   if (l){
1230     for (l = sub->lines; l;)
1231       free (sub->text[--l]);
1232     return ERR;
1233   }
1234   }
1235   return sub;
1236 }
1237
1238 #endif
1239
1240 static void adjust_subs_time(subtitle* sub, float subtime, float fps, int block,
1241                              int sub_num, int sub_uses_time) {
1242         int n,m;
1243         subtitle* nextsub;
1244         int i = sub_num;
1245         unsigned long subfms = (sub_uses_time ? 100 : fps) * subtime;
1246         unsigned long overlap = (sub_uses_time ? 100 : fps) / 5; // 0.2s
1247
1248         n=m=0;
1249         if (i)  for (;;){
1250                 if (sub->end <= sub->start){
1251                         sub->end = sub->start + subfms;
1252                         m++;
1253                         n++;
1254                 }
1255                 if (!--i) break;
1256                 nextsub = sub + 1;
1257             if(block){
1258                 if ((sub->end > nextsub->start) && (sub->end <= nextsub->start + overlap)) {
1259                     // these subtitles overlap for less than 0.2 seconds
1260                     // and would result in very short overlapping subtitle
1261                     // so let's fix the problem here, before overlapping code
1262                     // get its hands on them
1263                     unsigned delta = sub->end - nextsub->start, half = delta / 2;
1264                     sub->end -= half + 1;
1265                     nextsub->start += delta - half;
1266                 }
1267                 if (sub->end >= nextsub->start){
1268                         sub->end = nextsub->start - 1;
1269                         if (sub->end - sub->start > subfms)
1270                                 sub->end = sub->start + subfms;
1271                         if (!m)
1272                                 n++;
1273                 }
1274             }
1275
1276                 /* Theory:
1277                  * Movies are often converted from FILM (24 fps)
1278                  * to PAL (25) by simply speeding it up, so we
1279                  * to multiply the original timestmaps by
1280                  * (Movie's FPS / Subtitle's (guessed) FPS)
1281                  * so eg. for 23.98 fps movie and PAL time based
1282                  * subtitles we say -subfps 25 and we're fine!
1283                  */
1284
1285                 /* timed sub fps correction ::atmos */
1286                 if(sub_uses_time && sub_fps) {
1287                         sub->start *= sub_fps/fps;
1288                         sub->end   *= sub_fps/fps;
1289                 }
1290
1291                 sub = nextsub;
1292                 m = 0;
1293         }
1294         if (n) mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Adjusted %d subtitle(s).\n", n);
1295 }
1296
1297 struct subreader {
1298     subtitle * (*read)(FILE *fd,subtitle *dest);
1299     void       (*post)(subtitle *dest);
1300     const char *name;
1301 };
1302
1303 #ifdef HAVE_ENCA
1304 #define MAX_GUESS_BUFFER_SIZE (256*1024)
1305 void* guess_cp(FILE *fd, char *preferred_language, char *fallback)
1306 {
1307     const char **languages;
1308     size_t langcnt, buflen;
1309     EncaAnalyser analyser;
1310     EncaEncoding encoding;
1311     unsigned char *buffer;
1312     char *detected_sub_cp = NULL;
1313     int i;
1314
1315     buffer = (unsigned char*)malloc(MAX_GUESS_BUFFER_SIZE*sizeof(char));
1316     buflen = fread(buffer, 1, MAX_GUESS_BUFFER_SIZE, fd);
1317
1318     languages = enca_get_languages(&langcnt);
1319     mp_msg(MSGT_SUBREADER, MSGL_V, "ENCA supported languages: ");
1320     for (i = 0; i < langcnt; i++) {
1321         mp_msg(MSGT_SUBREADER, MSGL_V, "%s ", languages[i]);
1322     }
1323     mp_msg(MSGT_SUBREADER, MSGL_V, "\n");
1324
1325     for (i = 0; i < langcnt; i++) {
1326         if (strcasecmp(languages[i], preferred_language) != 0) continue;
1327         analyser = enca_analyser_alloc(languages[i]);
1328         encoding = enca_analyse_const(analyser, buffer, buflen);
1329         mp_msg(MSGT_SUBREADER, MSGL_INFO, "ENCA detected charset: %s\n", enca_charset_name(encoding.charset, ENCA_NAME_STYLE_ICONV));
1330         detected_sub_cp = strdup(enca_charset_name(encoding.charset, ENCA_NAME_STYLE_ICONV));
1331         enca_analyser_free(analyser);
1332     }
1333
1334     free(languages);
1335     free(buffer);
1336     rewind(fd);
1337
1338     if (!detected_sub_cp) detected_sub_cp = strdup(fallback);
1339
1340     return detected_sub_cp;
1341 }
1342 #endif
1343
1344 sub_data* sub_read_file (char *filename, float fps) {
1345         //filename is assumed to be malloc'ed,  free() is used in sub_free()
1346     FILE *fd;
1347     int n_max, n_first, i, j, sub_first, sub_orig;
1348     subtitle *first, *second, *sub, *return_sub;
1349     sub_data *subt_data;
1350     int uses_time = 0, sub_num = 0, sub_errs = 0;
1351     struct subreader sr[]=
1352     {
1353             { sub_read_line_microdvd, NULL, "microdvd" },
1354             { sub_read_line_subrip, NULL, "subrip" },
1355             { sub_read_line_subviewer, NULL, "subviewer" },
1356             { sub_read_line_sami, NULL, "sami" },
1357             { sub_read_line_vplayer, NULL, "vplayer" },
1358             { sub_read_line_rt, NULL, "rt" },
1359             { sub_read_line_ssa, sub_pp_ssa, "ssa" },
1360             { sub_read_line_pjs, NULL, "pjs" },
1361             { sub_read_line_mpsub, NULL, "mpsub" },
1362             { sub_read_line_aqt, NULL, "aqt" },
1363             { sub_read_line_subviewer2, NULL, "subviewer 2.0" },
1364             { sub_read_line_subrip09, NULL, "subrip 0.9" },
1365             { sub_read_line_jacosub, NULL, "jacosub" },
1366             { sub_read_line_mpl2, NULL, "mpl2" }
1367     };
1368     struct subreader *srp;
1369
1370     if(filename==NULL) return NULL; //qnx segfault
1371     fd=fopen (filename, "r"); if (!fd) return NULL;
1372
1373     sub_format=sub_autodetect (fd, &uses_time);
1374     mpsub_multiplier = (uses_time ? 100.0 : 1.0);
1375     if (sub_format==SUB_INVALID) {mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: Could not determine file format\n");return NULL;}
1376     srp=sr+sub_format;
1377     mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Detected subtitle file format: %s\n", srp->name);
1378
1379     rewind (fd);
1380
1381 #ifdef USE_ICONV
1382     sub_utf8_prev=sub_utf8;
1383     {
1384             int l,k;
1385             k = -1;
1386             if ((l=strlen(filename))>4){
1387                     char *exts[] = {".utf", ".utf8", ".utf-8" };
1388                     for (k=3;--k>=0;)
1389                         if (!strcasecmp(filename+(l - strlen(exts[k])), exts[k])){
1390                             sub_utf8 = 1;
1391                             break;
1392                         }
1393             }
1394             if (k<0) subcp_open(fd);
1395     }
1396 #endif
1397
1398     sub_num=0;n_max=32;
1399     first=(subtitle *)malloc(n_max*sizeof(subtitle));
1400     if(!first){
1401 #ifdef USE_ICONV
1402           subcp_close();
1403           sub_utf8=sub_utf8_prev;
1404 #endif
1405             return NULL;
1406     }
1407
1408 #ifdef USE_SORTSUB
1409     sub = (subtitle *)malloc(sizeof(subtitle));
1410     //This is to deal with those formats (AQT & Subrip) which define the end of a subtitle
1411     //as the beginning of the following
1412     previous_sub_end = 0;
1413 #endif
1414     while(1){
1415         if(sub_num>=n_max){
1416             n_max+=16;
1417             first=realloc(first,n_max*sizeof(subtitle));
1418         }
1419 #ifndef USE_SORTSUB
1420         sub = &first[sub_num];
1421 #endif
1422         memset(sub, '\0', sizeof(subtitle));
1423         sub=srp->read(fd,sub);
1424         if(!sub) break;   // EOF
1425 #ifdef USE_ICONV
1426         if ((sub!=ERR) && (sub_utf8 & 2)) sub=subcp_recode(sub);
1427 #endif
1428 #ifdef USE_FRIBIDI
1429         if (sub!=ERR) sub=sub_fribidi(sub,sub_utf8);
1430 #endif
1431         if ( sub == ERR )
1432          {
1433 #ifdef USE_ICONV
1434           subcp_close();
1435 #endif
1436           if ( first ) free(first);
1437           return NULL;
1438          }
1439         // Apply any post processing that needs recoding first
1440         if ((sub!=ERR) && !sub_no_text_pp && srp->post) srp->post(sub);
1441 #ifdef USE_SORTSUB
1442         if(!sub_num || (first[sub_num - 1].start <= sub->start)){
1443             first[sub_num].start = sub->start;
1444             first[sub_num].end   = sub->end;
1445             first[sub_num].lines = sub->lines;
1446             first[sub_num].alignment = sub->alignment;
1447             for(i = 0; i < sub->lines; ++i){
1448                 first[sub_num].text[i] = sub->text[i];
1449             }
1450             if (previous_sub_end){
1451                 first[sub_num - 1].end = previous_sub_end;
1452                 previous_sub_end = 0;
1453             }
1454         } else {
1455             for(j = sub_num - 1; j >= 0; --j){
1456                 first[j + 1].start = first[j].start;
1457                 first[j + 1].end   = first[j].end;
1458                 first[j + 1].lines = first[j].lines;
1459                 first[j + 1].alignment = first[j].alignment;
1460                 for(i = 0; i < first[j].lines; ++i){
1461                     first[j + 1].text[i] = first[j].text[i];
1462                 }
1463                 if(!j || (first[j - 1].start <= sub->start)){
1464                     first[j].start = sub->start;
1465                     first[j].end   = sub->end;
1466                     first[j].lines = sub->lines;
1467                     first[j].alignment = sub->alignment;
1468                     for(i = 0; i < SUB_MAX_TEXT; ++i){
1469                         first[j].text[i] = sub->text[i];
1470                     }
1471                     if (previous_sub_end){
1472                         first[j].end = first[j - 1].end;
1473                         first[j - 1].end = previous_sub_end;
1474                         previous_sub_end = 0;
1475                     }
1476                     break;
1477                 }
1478             }
1479         }
1480 #endif
1481         if(sub==ERR) ++sub_errs; else ++sub_num; // Error vs. Valid
1482     }
1483
1484     fclose(fd);
1485
1486 #ifdef USE_ICONV
1487     subcp_close();
1488 #endif
1489
1490 //    printf ("SUB: Subtitle format %s time.\n", uses_time?"uses":"doesn't use");
1491     mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Read %i subtitles", sub_num);
1492     if (sub_errs) mp_msg(MSGT_SUBREADER,MSGL_INFO,", %i bad line(s).\n", sub_errs);
1493     else          mp_msg(MSGT_SUBREADER,MSGL_INFO,".\n");
1494
1495     if(sub_num<=0){
1496         free(first);
1497         return NULL;
1498     }
1499
1500     // we do overlap if the user forced it (suboverlap_enable == 2) or
1501     // the user didn't forced no-overlapsub and the format is Jacosub or Ssa.
1502     // this is because usually overlapping subtitles are found in these formats,
1503     // while in others they are probably result of bad timing
1504 if ((suboverlap_enabled == 2) ||
1505     ((suboverlap_enabled) && ((sub_format == SUB_JACOSUB) || (sub_format == SUB_SSA)))) {
1506     adjust_subs_time(first, 6.0, fps, 0, sub_num, uses_time);/*~6 secs AST*/
1507 // here we manage overlapping subtitles
1508     sub_orig = sub_num;
1509     n_first = sub_num;
1510     sub_num = 0;
1511     second = NULL;
1512     // for each subtitle in first[] we deal with its 'block' of
1513     // bonded subtitles
1514     for (sub_first = 0; sub_first < n_first; ++sub_first) {
1515         unsigned long global_start = first[sub_first].start,
1516                 global_end = first[sub_first].end, local_start, local_end;
1517         int lines_to_add = first[sub_first].lines, sub_to_add = 0,
1518                 **placeholder = NULL, higher_line = 0, counter, start_block_sub = sub_num;
1519         char real_block = 1;
1520
1521         // here we find the number of subtitles inside the 'block'
1522         // and its span interval. this works well only with sorted
1523         // subtitles
1524         while ((sub_first + sub_to_add + 1 < n_first) && (first[sub_first + sub_to_add + 1].start < global_end)) {
1525             ++sub_to_add;
1526             lines_to_add += first[sub_first + sub_to_add].lines;
1527             if (first[sub_first + sub_to_add].start < global_start) {
1528                 global_start = first[sub_first + sub_to_add].start;
1529             }
1530             if (first[sub_first + sub_to_add].end > global_end) {
1531                 global_end = first[sub_first + sub_to_add].end;
1532             }
1533         }
1534
1535         // we need a structure to keep trace of the screen lines
1536         // used by the subs, a 'placeholder'
1537         counter = 2 * sub_to_add + 1;  // the maximum number of subs derived
1538                                        // from a block of sub_to_add+1 subs
1539         placeholder = (int **) malloc(sizeof(int *) * counter);
1540         for (i = 0; i < counter; ++i) {
1541             placeholder[i] = (int *) malloc(sizeof(int) * lines_to_add);
1542             for (j = 0; j < lines_to_add; ++j) {
1543                 placeholder[i][j] = -1;
1544             }
1545         }
1546
1547         counter = 0;
1548         local_end = global_start - 1;
1549         do {
1550             int ls;
1551
1552             // here we find the beginning and the end of a new
1553             // subtitle in the block
1554             local_start = local_end + 1;
1555             local_end   = global_end;
1556             for (j = 0; j <= sub_to_add; ++j) {
1557                 if ((first[sub_first + j].start - 1 > local_start) && (first[sub_first + j].start - 1 < local_end)) {
1558                     local_end = first[sub_first + j].start - 1;
1559                 } else if ((first[sub_first + j].end > local_start) && (first[sub_first + j].end < local_end)) {
1560                     local_end = first[sub_first + j].end;
1561                 }
1562             }
1563             // here we allocate the screen lines to subs we must
1564             // display in current local_start-local_end interval.
1565             // if the subs were yet presents in the previous interval
1566             // they keep the same lines, otherside they get unused lines
1567             for (j = 0; j <= sub_to_add; ++j) {
1568                 if ((first[sub_first + j].start <= local_end) && (first[sub_first + j].end > local_start)) {
1569                     unsigned long sub_lines = first[sub_first + j].lines, fragment_length = lines_to_add + 1,
1570                         tmp = 0;
1571                     char boolean = 0;
1572                     int fragment_position = -1;
1573
1574                     // if this is not the first new sub of the block
1575                     // we find if this sub was present in the previous
1576                     // new sub
1577                     if (counter)
1578                         for (i = 0; i < lines_to_add; ++i) {
1579                             if (placeholder[counter - 1][i] == sub_first + j) {
1580                                 placeholder[counter][i] = sub_first + j;
1581                                 boolean = 1;
1582                             }
1583                         }
1584                     if (boolean)
1585                         continue;
1586
1587                     // we are looking for the shortest among all groups of
1588                     // sequential blank lines whose length is greater than or
1589                     // equal to sub_lines. we store in fragment_position the
1590                     // position of the shortest group, in fragment_length its
1591                     // length, and in tmp the length of the group currently
1592                     // examinated
1593                     for (i = 0; i < lines_to_add; ++i) {
1594                         if (placeholder[counter][i] == -1) {
1595                             // placeholder[counter][i] is part of the current group
1596                             // of blank lines
1597                             ++tmp;
1598                         } else {
1599                             if (tmp == sub_lines) {
1600                                 // current group's size fits exactly the one we
1601                                 // need, so we stop looking
1602                                 fragment_position = i - tmp;
1603                                 tmp = 0;
1604                                 break;
1605                             }
1606                             if ((tmp) && (tmp > sub_lines) && (tmp < fragment_length)) {
1607                                 // current group is the best we found till here,
1608                                 // but is still bigger than the one we are looking
1609                                 // for, so we keep on looking
1610                                 fragment_length = tmp;
1611                                 fragment_position = i - tmp;
1612                                 tmp = 0;
1613                             } else {
1614                                 // current group doesn't fit at all, so we forget it
1615                                 tmp = 0;
1616                             }
1617                         }
1618                     }
1619                     if (tmp) {
1620                         // last screen line is blank, a group ends with it
1621                         if ((tmp >= sub_lines) && (tmp < fragment_length)) {
1622                             fragment_position = i - tmp;
1623                         }
1624                     }
1625                     if (fragment_position == -1) {
1626                         // it was not possible to find free screen line(s) for a subtitle,
1627                         // usually this means a bug in the code; however we do not overlap
1628                         mp_msg(MSGT_SUBREADER, MSGL_WARN, "SUB: we could not find a suitable position for an overlapping subtitle\n");
1629                         higher_line = SUB_MAX_TEXT + 1;
1630                         break;
1631                     } else {
1632                         for (tmp = 0; tmp < sub_lines; ++tmp) {
1633                             placeholder[counter][fragment_position + tmp] = sub_first + j;
1634                         }
1635                     }
1636                 }
1637             }
1638             for (j = higher_line + 1; j < lines_to_add; ++j) {
1639                 if (placeholder[counter][j] != -1)
1640                     higher_line = j;
1641                 else
1642                     break;
1643             }
1644             if (higher_line >= SUB_MAX_TEXT) {
1645                 // the 'block' has too much lines, so we don't overlap the
1646                 // subtitles
1647                 second = (subtitle *) realloc(second, (sub_num + sub_to_add + 1) * sizeof(subtitle));
1648                 for (j = 0; j <= sub_to_add; ++j) {
1649                     int ls;
1650                     memset(&second[sub_num + j], '\0', sizeof(subtitle));
1651                     second[sub_num + j].start = first[sub_first + j].start;
1652                     second[sub_num + j].end   = first[sub_first + j].end;
1653                     second[sub_num + j].lines = first[sub_first + j].lines;
1654                     second[sub_num + j].alignment = first[sub_first + j].alignment;
1655                     for (ls = 0; ls < second[sub_num + j].lines; ls++) {
1656                         second[sub_num + j].text[ls] = strdup(first[sub_first + j].text[ls]);
1657                     }
1658                 }
1659                 sub_num += sub_to_add + 1;
1660                 sub_first += sub_to_add;
1661                 real_block = 0;
1662                 break;
1663             }
1664
1665             // we read the placeholder structure and create the new
1666             // subs.
1667             second = (subtitle *) realloc(second, (sub_num + 1) * sizeof(subtitle));
1668             memset(&second[sub_num], '\0', sizeof(subtitle));
1669             second[sub_num].start = local_start;
1670             second[sub_num].end   = local_end;
1671             second[sub_num].alignment = first[sub_first].alignment;
1672             n_max = (lines_to_add < SUB_MAX_TEXT) ? lines_to_add : SUB_MAX_TEXT;
1673             for (i = 0, j = 0; j < n_max; ++j) {
1674                 if (placeholder[counter][j] != -1) {
1675                     int lines = first[placeholder[counter][j]].lines;
1676                     for (ls = 0; ls < lines; ++ls) {
1677                         second[sub_num].text[i++] = strdup(first[placeholder[counter][j]].text[ls]);
1678                     }
1679                     j += lines - 1;
1680                 } else {
1681                     second[sub_num].text[i++] = strdup(" ");
1682                 }
1683             }
1684             ++sub_num;
1685             ++counter;
1686         } while (local_end < global_end);
1687         if (real_block)
1688             for (i = 0; i < counter; ++i)
1689                 second[start_block_sub + i].lines = higher_line + 1;
1690
1691         counter = 2 * sub_to_add + 1;
1692         for (i = 0; i < counter; ++i) {
1693             free(placeholder[i]);
1694         }
1695         free(placeholder);
1696         sub_first += sub_to_add;
1697     }
1698
1699     for (j = sub_orig - 1; j >= 0; --j) {
1700         for (i = first[j].lines - 1; i >= 0; --i) {
1701             free(first[j].text[i]);
1702         }
1703     }
1704     free(first);
1705
1706     return_sub = second;
1707 } else { //if(suboverlap_enabled)
1708     adjust_subs_time(first, 6.0, fps, 1, sub_num, uses_time);/*~6 secs AST*/
1709     return_sub = first;
1710 }
1711     if (return_sub == NULL) return NULL;
1712     subt_data = (sub_data *)malloc(sizeof(sub_data));
1713     subt_data->filename = filename;
1714     subt_data->sub_uses_time = uses_time;
1715     subt_data->sub_num = sub_num;
1716     subt_data->sub_errs = sub_errs;
1717     subt_data->subtitles = return_sub;
1718     return subt_data;
1719 }
1720
1721 #if 0
1722 char * strreplace( char * in,char * what,char * whereof )
1723 {
1724  int i;
1725  char * tmp;
1726
1727  if ( ( in == NULL )||( what == NULL )||( whereof == NULL )||( ( tmp=strstr( in,what ) ) == NULL ) ) return NULL;
1728  for( i=0;i<strlen( whereof );i++ ) tmp[i]=whereof[i];
1729  if ( strlen( what ) > strlen( whereof ) ) tmp[i]=0;
1730  return in;
1731 }
1732 #endif
1733
1734
1735 static void strcpy_trim(char *d, char *s)
1736 {
1737     // skip leading whitespace
1738     while (*s && !isalnum(*s)) {
1739         s++;
1740     }
1741     for (;;) {
1742         // copy word
1743         while (*s && isalnum(*s)) {
1744             *d = tolower(*s);
1745             s++; d++;
1746         }
1747         if (*s == 0) break;
1748         // trim excess whitespace
1749         while (*s && !isalnum(*s)) {
1750             s++;
1751         }
1752         if (*s == 0) break;
1753         *d++ = ' ';
1754     }
1755     *d = 0;
1756 }
1757
1758 static void strcpy_strip_ext(char *d, char *s)
1759 {
1760     char *tmp = strrchr(s,'.');
1761     if (!tmp) {
1762         strcpy(d, s);
1763         return;
1764     } else {
1765         strncpy(d, s, tmp-s);
1766         d[tmp-s] = 0;
1767     }
1768     while (*d) {
1769         *d = tolower(*d);
1770         d++;
1771     }
1772 }
1773
1774 static void strcpy_get_ext(char *d, char *s)
1775 {
1776     char *tmp = strrchr(s,'.');
1777     if (!tmp) {
1778         strcpy(d, "");
1779         return;
1780     } else {
1781         strcpy(d, tmp+1);
1782    }
1783 }
1784
1785 static int whiteonly(char *s)
1786 {
1787     while (*s) {
1788         if (isalnum(*s)) return 0;
1789         s++;
1790   }
1791     return 1;
1792 }
1793
1794 typedef struct _subfn
1795 {
1796     int priority;
1797     char *fname;
1798 } subfn;
1799
1800 static int compare_sub_priority(const void *a, const void *b)
1801 {
1802     if (((subfn*)a)->priority > ((subfn*)b)->priority) {
1803         return -1;
1804     } else if (((subfn*)a)->priority < ((subfn*)b)->priority) {
1805         return 1;
1806     } else {
1807         return strcoll(((subfn*)a)->fname, ((subfn*)b)->fname);
1808     }
1809 }
1810
1811 char** sub_filenames(char* path, char *fname)
1812 {
1813     char *f_dir, *f_fname, *f_fname_noext, *f_fname_trim, *tmp, *tmp_sub_id;
1814     char *tmp_fname_noext, *tmp_fname_trim, *tmp_fname_ext, *tmpresult;
1815
1816     int len, pos, found, i, j;
1817     char * sub_exts[] = {  "utf", "utf8", "utf-8", "sub", "srt", "smi", "rt", "txt", "ssa", "aqt", "jss", "js", "ass", NULL};
1818     subfn *result;
1819     char **result2;
1820
1821     int subcnt;
1822
1823     FILE *f;
1824
1825     DIR *d;
1826     struct dirent *de;
1827
1828     len = (strlen(fname) > 256 ? strlen(fname) : 256)
1829         +(strlen(path) > 256 ? strlen(path) : 256)+2;
1830
1831     f_dir = (char*)malloc(len);
1832     f_fname = (char*)malloc(len);
1833     f_fname_noext = (char*)malloc(len);
1834     f_fname_trim = (char*)malloc(len);
1835
1836     tmp_fname_noext = (char*)malloc(len);
1837     tmp_fname_trim = (char*)malloc(len);
1838     tmp_fname_ext = (char*)malloc(len);
1839
1840     tmpresult = (char*)malloc(len);
1841
1842     result = (subfn*)malloc(sizeof(subfn)*MAX_SUBTITLE_FILES);
1843     memset(result, 0, sizeof(subfn)*MAX_SUBTITLE_FILES);
1844
1845     subcnt = 0;
1846
1847     tmp = strrchr(fname,'/');
1848 #ifdef WIN32
1849     if(!tmp)tmp = strrchr(fname,'\\');
1850 #endif
1851
1852     // extract filename & dirname from fname
1853     if (tmp) {
1854         strcpy(f_fname, tmp+1);
1855         pos = tmp - fname;
1856         strncpy(f_dir, fname, pos+1);
1857         f_dir[pos+1] = 0;
1858     } else {
1859         strcpy(f_fname, fname);
1860         strcpy(f_dir, "./");
1861     }
1862
1863     strcpy_strip_ext(f_fname_noext, f_fname);
1864     strcpy_trim(f_fname_trim, f_fname_noext);
1865
1866     tmp_sub_id = NULL;
1867     if (dvdsub_lang && !whiteonly(dvdsub_lang)) {
1868         tmp_sub_id = (char*)malloc(strlen(dvdsub_lang)+1);
1869         strcpy_trim(tmp_sub_id, dvdsub_lang);
1870     }
1871
1872     // 0 = nothing
1873     // 1 = any subtitle file
1874     // 2 = any sub file containing movie name
1875     // 3 = sub file containing movie name and the lang extension
1876     for (j = 0; j <= 1; j++) {
1877         d = opendir(j == 0 ? f_dir : path);
1878         if (d) {
1879             while ((de = readdir(d))) {
1880                 // retrieve various parts of the filename
1881                 strcpy_strip_ext(tmp_fname_noext, de->d_name);
1882                 strcpy_get_ext(tmp_fname_ext, de->d_name);
1883                 strcpy_trim(tmp_fname_trim, tmp_fname_noext);
1884
1885                 // does it end with a subtitle extension?
1886                 found = 0;
1887 #ifdef USE_ICONV
1888 #ifdef HAVE_ENCA
1889                 for (i = ((sub_cp && strncasecmp(sub_cp, "enca", 4) != 0) ? 3 : 0); sub_exts[i]; i++) {
1890 #else
1891                 for (i = (sub_cp ? 3 : 0); sub_exts[i]; i++) {
1892 #endif
1893 #else
1894                 for (i = 0; sub_exts[i]; i++) {
1895 #endif
1896                     if (strcasecmp(sub_exts[i], tmp_fname_ext) == 0) {
1897                         found = 1;
1898                         break;
1899                     }
1900                 }
1901
1902                 // we have a (likely) subtitle file
1903                 if (found) {
1904                     int prio = 0;
1905                     if (!prio && tmp_sub_id)
1906                     {
1907                         sprintf(tmpresult, "%s %s", f_fname_trim, tmp_sub_id);
1908                         printf("dvdsublang...%s\n", tmpresult);
1909                         if (strcmp(tmp_fname_trim, tmpresult) == 0 && sub_match_fuzziness >= 1) {
1910                             // matches the movie name + lang extension
1911                             prio = 5;
1912                         }
1913                     }
1914                     if (!prio && strcmp(tmp_fname_trim, f_fname_trim) == 0) {
1915                         // matches the movie name
1916                         prio = 4;
1917                     }
1918                     if (!prio && (tmp = strstr(tmp_fname_trim, f_fname_trim)) && (sub_match_fuzziness >= 1)) {
1919                         // contains the movie name
1920                         tmp += strlen(f_fname_trim);
1921                         if (tmp_sub_id && strstr(tmp, tmp_sub_id)) {
1922                             // with sub_id specified prefer localized subtitles
1923                             prio = 3;
1924                         } else if ((tmp_sub_id == NULL) && whiteonly(tmp)) {
1925                             // without sub_id prefer "plain" name
1926                             prio = 3;
1927                         } else {
1928                             // with no localized subs found, try any else instead
1929                             prio = 2;
1930                         }
1931                     }
1932                     if (!prio) {
1933                         // doesn't contain the movie name
1934                         // don't try in the mplayer subtitle directory
1935                         if ((j == 0) && (sub_match_fuzziness >= 2)) {
1936                             prio = 1;
1937                         }
1938                     }
1939
1940                     if (prio) {
1941                         prio += prio;
1942 #ifdef USE_ICONV
1943                         if (i<3){ // prefer UTF-8 coded
1944                             prio++;
1945                         }
1946 #endif
1947                         sprintf(tmpresult, "%s%s", j == 0 ? f_dir : path, de->d_name);
1948 //                      fprintf(stderr, "%s priority %d\n", tmpresult, prio);
1949                         if ((f = fopen(tmpresult, "rt"))) {
1950                             fclose(f);
1951                             result[subcnt].priority = prio;
1952                             result[subcnt].fname = strdup(tmpresult);
1953                             subcnt++;
1954                         }
1955                     }
1956
1957                 }
1958                 if (subcnt >= MAX_SUBTITLE_FILES) break;
1959             }
1960             closedir(d);
1961         }
1962
1963     }
1964
1965     if (tmp_sub_id) free(tmp_sub_id);
1966
1967     free(f_dir);
1968     free(f_fname);
1969     free(f_fname_noext);
1970     free(f_fname_trim);
1971
1972     free(tmp_fname_noext);
1973     free(tmp_fname_trim);
1974     free(tmp_fname_ext);
1975
1976     free(tmpresult);
1977
1978     qsort(result, subcnt, sizeof(subfn), compare_sub_priority);
1979
1980     result2 = (char**)malloc(sizeof(char*)*(subcnt+1));
1981     memset(result2, 0, sizeof(char*)*(subcnt+1));
1982
1983     for (i = 0; i < subcnt; i++) {
1984         result2[i] = result[i].fname;
1985     }
1986     result2[subcnt] = NULL;
1987
1988     free(result);
1989
1990     return result2;
1991 }
1992
1993 void list_sub_file(sub_data* subd){
1994     int i,j;
1995     subtitle *subs = subd->subtitles;
1996
1997     for(j=0; j < subd->sub_num; j++){
1998         subtitle* egysub=&subs[j];
1999         printf ("%i line%c (%li-%li)\n",
2000                     egysub->lines,
2001                     (1==egysub->lines)?' ':'s',
2002                     egysub->start,
2003                     egysub->end);
2004         for (i=0; i<egysub->lines; i++) {
2005             printf ("\t\t%d: %s%s", i,egysub->text[i], i==egysub->lines-1?"":" \n ");
2006         }
2007         printf ("\n");
2008     }
2009
2010     printf ("Subtitle format %s time.\n",
2011                                   subd->sub_uses_time ? "uses":"doesn't use");
2012     printf ("Read %i subtitles, %i errors.\n", subd->sub_num, subd->sub_errs);
2013 }
2014
2015 void dump_srt(sub_data* subd, float fps){
2016     int i,j;
2017     int h,m,s,ms;
2018     FILE * fd;
2019     subtitle * onesub;
2020     unsigned long temp;
2021     subtitle *subs = subd->subtitles;
2022
2023     if (!subd->sub_uses_time && sub_fps == 0)
2024         sub_fps = fps;
2025     fd=fopen("dumpsub.srt","w");
2026     if(!fd)
2027     {
2028         perror("dump_srt: fopen");
2029         return;
2030     }
2031     for(i=0; i < subd->sub_num; i++)
2032     {
2033         onesub=subs+i;    //=&subs[i];
2034         fprintf(fd,"%d\n",i+1);//line number
2035
2036         temp=onesub->start;
2037         if (!subd->sub_uses_time)
2038             temp = temp * 100 / sub_fps;
2039         temp -= sub_delay * 100;
2040         h=temp/360000;temp%=360000;     //h =1*100*60*60
2041         m=temp/6000;  temp%=6000;       //m =1*100*60
2042         s=temp/100;   temp%=100;        //s =1*100
2043         ms=temp*10;                     //ms=1*10
2044         fprintf(fd,"%02d:%02d:%02d,%03d --> ",h,m,s,ms);
2045
2046         temp=onesub->end;
2047         if (!subd->sub_uses_time)
2048             temp = temp * 100 / sub_fps;
2049         temp -= sub_delay * 100;
2050         h=temp/360000;temp%=360000;
2051         m=temp/6000;  temp%=6000;
2052         s=temp/100;   temp%=100;
2053         ms=temp*10;
2054         fprintf(fd,"%02d:%02d:%02d,%03d\n",h,m,s,ms);
2055
2056         for(j=0;j<onesub->lines;j++)
2057             fprintf(fd,"%s\n",onesub->text[j]);
2058
2059         fprintf(fd,"\n");
2060     }
2061     fclose(fd);
2062     mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.srt\'.\n");
2063 }
2064
2065 void dump_mpsub(sub_data* subd, float fps){
2066         int i,j;
2067         FILE *fd;
2068         float a,b;
2069         subtitle *subs = subd->subtitles;
2070
2071         mpsub_position = subd->sub_uses_time? (sub_delay*100) : (sub_delay*fps);
2072         if (sub_fps==0) sub_fps=fps;
2073
2074         fd=fopen ("dump.mpsub", "w");
2075         if (!fd) {
2076                 perror ("dump_mpsub: fopen");
2077                 return;
2078         }
2079
2080
2081         if (subd->sub_uses_time) fprintf (fd,"FORMAT=TIME\n\n");
2082         else fprintf (fd, "FORMAT=%5.2f\n\n", fps);
2083
2084         for(j=0; j < subd->sub_num; j++){
2085                 subtitle* egysub=&subs[j];
2086                 if (subd->sub_uses_time) {
2087                         a=((egysub->start-mpsub_position)/100.0);
2088                         b=((egysub->end-egysub->start)/100.0);
2089                         if ( (float)((int)a) == a)
2090                         fprintf (fd, "%.0f",a);
2091                         else
2092                         fprintf (fd, "%.2f",a);
2093
2094                         if ( (float)((int)b) == b)
2095                         fprintf (fd, " %.0f\n",b);
2096                         else
2097                         fprintf (fd, " %.2f\n",b);
2098                 } else {
2099                         fprintf (fd, "%ld %ld\n", (long)((egysub->start*(fps/sub_fps))-((mpsub_position*(fps/sub_fps)))),
2100                                         (long)(((egysub->end)-(egysub->start))*(fps/sub_fps)));
2101                 }
2102
2103                 mpsub_position = egysub->end;
2104                 for (i=0; i<egysub->lines; i++) {
2105                         fprintf (fd, "%s\n",egysub->text[i]);
2106                 }
2107                 fprintf (fd, "\n");
2108         }
2109         fclose (fd);
2110         mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dump.mpsub\'.\n");
2111 }
2112
2113 void dump_microdvd(sub_data* subd, float fps) {
2114     int i, delay;
2115     FILE *fd;
2116     subtitle *subs = subd->subtitles;
2117     if (sub_fps == 0)
2118         sub_fps = fps;
2119     fd = fopen("dumpsub.txt", "w");
2120     if (!fd) {
2121         perror("dumpsub.txt: fopen");
2122         return;
2123     }
2124     delay = sub_delay * sub_fps;
2125     for (i = 0; i < subd->sub_num; ++i) {
2126         int j, start, end;
2127         start = subs[i].start;
2128         end = subs[i].end;
2129         if (subd->sub_uses_time) {
2130             start = start * sub_fps / 100 ;
2131             end = end * sub_fps / 100;
2132         }
2133         else {
2134             start = start * sub_fps / fps;
2135             end = end * sub_fps / fps;
2136         }
2137         start -= delay;
2138         end -= delay;
2139         fprintf(fd, "{%d}{%d}", start, end);
2140         for (j = 0; j < subs[i].lines; ++j)
2141             fprintf(fd, "%s%s", j ? "|" : "", subs[i].text[j]);
2142         fprintf(fd, "\n");
2143     }
2144     fclose(fd);
2145     mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.txt\'.\n");
2146 }
2147
2148 void dump_jacosub(sub_data* subd, float fps) {
2149     int i,j;
2150     int h,m,s,cs;
2151     FILE * fd;
2152     subtitle * onesub;
2153     unsigned long temp;
2154     subtitle *subs = subd->subtitles;
2155
2156     if (!subd->sub_uses_time && sub_fps == 0)
2157         sub_fps = fps;
2158     fd=fopen("dumpsub.jss","w");
2159     if(!fd)
2160     {
2161         perror("dump_jacosub: fopen");
2162         return;
2163     }
2164     fprintf(fd, "#TIMERES %d\n", (subd->sub_uses_time) ? 100 : (int)sub_fps);
2165     for(i=0; i < subd->sub_num; i++)
2166     {
2167         onesub=subs+i;    //=&subs[i];
2168
2169         temp=onesub->start;
2170         if (!subd->sub_uses_time)
2171             temp = temp * 100 / sub_fps;
2172         temp -= sub_delay * 100;
2173         h=temp/360000;temp%=360000;     //h =1*100*60*60
2174         m=temp/6000;  temp%=6000;       //m =1*100*60
2175         s=temp/100;   temp%=100;        //s =1*100
2176         cs=temp;                        //cs=1*10
2177         fprintf(fd,"%02d:%02d:%02d.%02d ",h,m,s,cs);
2178
2179         temp=onesub->end;
2180         if (!subd->sub_uses_time)
2181             temp = temp * 100 / sub_fps;
2182         temp -= sub_delay * 100;
2183         h=temp/360000;temp%=360000;
2184         m=temp/6000;  temp%=6000;
2185         s=temp/100;   temp%=100;
2186         cs=temp;
2187         fprintf(fd,"%02d:%02d:%02d.%02d {~} ",h,m,s,cs);
2188
2189         for(j=0;j<onesub->lines;j++)
2190             fprintf(fd,"%s%s",j ? "\\n" : "", onesub->text[j]);
2191
2192         fprintf(fd,"\n");
2193     }
2194     fclose(fd);
2195     mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.js\'.\n");
2196 }
2197
2198 void dump_sami(sub_data* subd, float fps) {
2199     int i,j;
2200     FILE * fd;
2201     subtitle * onesub;
2202     unsigned long temp;
2203     subtitle *subs = subd->subtitles;
2204
2205     if (!subd->sub_uses_time && sub_fps == 0)
2206         sub_fps = fps;
2207     fd=fopen("dumpsub.smi","w");
2208     if(!fd)
2209     {
2210         perror("dump_jacosub: fopen");
2211         return;
2212     }
2213     fprintf(fd, "<SAMI>\n"
2214                 "<HEAD>\n"
2215                 "       <STYLE TYPE=\"Text/css\">\n"
2216                 "       <!--\n"
2217                 "         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"
2218                 "         .SUBTTL {Name: 'Subtitles'; Lang: en-US; SAMIType: CC;}\n"
2219                 "       -->\n"
2220                 "       </STYLE>\n"
2221                 "</HEAD>\n"
2222                 "<BODY>\n");
2223     for(i=0; i < subd->sub_num; i++)
2224     {
2225         onesub=subs+i;    //=&subs[i];
2226
2227         temp=onesub->start;
2228         if (!subd->sub_uses_time)
2229             temp = temp * 100 / sub_fps;
2230         temp -= sub_delay * 100;
2231         fprintf(fd,"\t<SYNC Start=%lu>\n"
2232                     "\t  <P>", temp * 10);
2233
2234         for(j=0;j<onesub->lines;j++)
2235             fprintf(fd,"%s%s",j ? "<br>" : "", onesub->text[j]);
2236
2237         fprintf(fd,"\n");
2238
2239         temp=onesub->end;
2240         if (!subd->sub_uses_time)
2241             temp = temp * 100 / sub_fps;
2242         temp -= sub_delay * 100;
2243         fprintf(fd,"\t<SYNC Start=%lu>\n"
2244                     "\t  <P>&nbsp;\n", temp * 10);
2245     }
2246     fprintf(fd, "</BODY>\n"
2247                 "</SAMI>\n");
2248     fclose(fd);
2249     mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.smi\'.\n");
2250 }
2251
2252 void sub_free( sub_data * subd )
2253 {
2254  int i;
2255
2256     if ( !subd ) return;
2257
2258     if (subd->subtitles) {
2259         for (i=0; i < subd->subtitles->lines; i++) free( subd->subtitles->text[i] );
2260         free( subd->subtitles );
2261     }
2262     if (subd->filename) free( subd->filename );
2263     free( subd );
2264 }
2265
2266 #ifdef DUMPSUBS
2267 int main(int argc, char **argv) {  // for testing
2268     sub_data *subd;
2269
2270     if(argc<2){
2271         printf("\nUsage: subreader filename.sub\n\n");
2272         exit(1);
2273     }
2274     sub_cp = argv[2];
2275     subd = sub_read_file(argv[1]);
2276     if(!subd){
2277         printf("Couldn't load file.\n");
2278         exit(1);
2279     }
2280
2281     list_sub_file(subd);
2282
2283     return 0;
2284 }
2285 #endif