comparison subreader.c @ 29251:cada7608488b

Get rid of some more trailing whitespace

29250:bc5115226204

int sub_match_fuzziness=0; // level of sub name matching fuzziness

/* Use the SUB_* constant defined in the header file */
int sub_format=SUB_INVALID;
#ifdef CONFIG_SORTSUB
/* 
   Some subtitling formats, namely AQT and Subrip09, define the end of a
   subtitle as the beginning of the following. Since currently we read one
   subtitle at time, for these format we keep two global *subtitle,
   previous_aqt_sub and previous_subrip09_sub, pointing to previous subtitle,
   so we can change its end when we read current subtitle starting time.

    do {
	switch (state) {

	case 0: /* find "START=" or "Slacktime:" */
	    slacktime_s = stristr (s, "Slacktime:");
	    if (slacktime_s) 
                sub_slacktime = strtol (slacktime_s+10, NULL, 0) / 10;

	    s = stristr (s, "Start=");
	    if (s) {
		current->start = strtol (s + 6, &s, 0) / 10;

                for (; *s != '>' && *s != '\0'; s++);
                s++;
		state = 1; continue;
	    }
	    break;
 
	case 1: /* find (optionnal) "<P", skip other TAGs */
	    for  (; *s == ' ' || *s == '\t'; s++); /* strip blanks, if any */
	    if (*s == '\0') break;
	    if (*s != '<') { state = 3; p = text; continue; } /* not a TAG */
	    s++;

	    for (; *s != '>' && *s != '\0'; s++); /* skip remains of non-<P> TAG */
	    if (s == '\0')
	      break;
	    s++;
	    continue;
 
	case 2: /* find ">" */
	    if ((s = strchr (s, '>'))) { s++; state = 3; p = text; continue; }
	    break;
 
	case 3: /* get all text until '<' appears */
	    if (*s == '\0') break;
	    else if (!strncasecmp (s, "<br>", 4)) {
		*p = '\0'; p = text; trail_space (text);
		if (text[0] != '\0')

	    else if (*s == '\r' || *s == '\n') { s++; }
	    else *p++ = *s++;

	    /* skip duplicated space */
	    if (p > text + 2) if (*(p-1) == ' ' && *(p-2) == ' ') p--;
	    
	    continue;

	case 4: /* get current->end or skip <TAG> */
	    q = stristr (s, "Start=");
	    if (q) {

		break; // if it is the last subtitle
	    } else {
		return 0;
	    }
	}
	    
    } while (state != 99);

    // For the last subtitle
    if (current->end <= 0) {
        current->end = current->start + sub_slacktime;
	*p = '\0'; trail_space (text);
	if (text[0] != '\0')
	    current->text[current->lines++] = strdup (text);
    }
    
    return current;
}


static char *sub_readtext(char *source, char **dest) {
    int len=0;
    char *p=source;
    
//    printf("src=%p  dest=%p  \n",source,dest);

    while ( !eol(*p) && *p!= '|' ) {
	p++,len++;
    }
    
    *dest= malloc (len+1);
    if (!dest) {return ERR;}
    
    strncpy(*dest, source, len);
    (*dest)[len]=0;
    
    while (*p=='\r' || *p=='\n' || *p=='|') p++;
    
    if (*p) return p;  // not-last text field
    else return NULL;  // last text field
}

static subtitle *sub_read_line_microdvd(stream_t *st,subtitle *current) {

static subtitle *sub_read_line_subrip(stream_t* st, subtitle *current) {
    char line[LINE_LEN+1];
    int a1,a2,a3,a4,b1,b2,b3,b4;
    char *p=NULL, *q=NULL;
    int len;
    
    while (1) {
	if (!stream_read_line (st, line, LINE_LEN)) return NULL;
	if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4) < 8) continue;
	current->start = a1*360000+a2*6000+a3*100+a4;
	current->end   = b1*360000+b2*6000+b3*100+b4;

static subtitle *sub_read_line_subviewer(stream_t *st,subtitle *current) {
    char line[LINE_LEN+1];
    int a1,a2,a3,a4,b1,b2,b3,b4;
    char *p=NULL;
    int i,len;
    
    while (!current->text[0]) {
	if (!stream_read_line (st, line, LINE_LEN)) return NULL;
	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)
	    continue;
	current->start = a1*360000+a2*6000+a3*100+a4/10;

static subtitle *sub_read_line_subviewer2(stream_t *st,subtitle *current) {
    char line[LINE_LEN+1];
    int a1,a2,a3,a4;
    char *p=NULL;
    int i,len;
   
    while (!current->text[0]) {
        if (!stream_read_line (st, line, LINE_LEN)) return NULL;
	if (line[0]!='{')
	    continue;
        if ((len=sscanf (line, "{T %d:%d:%d:%d",&a1,&a2,&a3,&a4)) < 4)

		if (!(current->start = a1*360000+a2*6000+a3*100))
			continue;
                /* removed by wodzu
		p=line;	
 		// finds the body of the subtitle
 		for (i=0; i<3; i++){              
		   p=strchr(p,':');
		   if (p==NULL) break;
		   ++p;
		} 
		if (p==NULL) {
		    printf("SUB: Skipping incorrect subtitle line!\n");
		    continue;
		}
                */

	}
	return current;
}

static subtitle *sub_read_line_rt(stream_t *st,subtitle *current) {
	//TODO: This format uses quite rich (sub/super)set of xhtml 
	// I couldn't check it since DTD is not included.
	// WARNING: full XML parses can be required for proper parsing 
    char line[LINE_LEN+1];
    int a1,a2,a3,a4,b1,b2,b3,b4;
    char *p=NULL,*next=NULL;
    int i,len,plen;
    
    while (!current->text[0]) {
	if (!stream_read_line (st, line, LINE_LEN)) return NULL;
	//TODO: it seems that format of time is not easily determined, it may be 1:12, 1:12.0 or 0:1:12.0
	//to describe the same moment in time. Maybe there are even more formats in use.
	//if ((len=sscanf (line, "<Time Begin=\"%d:%d:%d.%d\" End=\"%d:%d:%d.%d\"",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8)

	((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) &&
	//now try it without end time
	((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d.%d\"%*[^<]<clear/>%n",&a3,&a4,&plen)) < 2) &&
	((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&plen)) < 2) &&
	((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&plen)) < 3) &&
	((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d:%d.%d\"%*[^<]<clear/>%n",&a1,&a2,&a3,&a4,&plen)) < 4) 
	)
	    continue;
	current->start = a1*360000+a2*6000+a3*100+a4/10;
	current->end   = b1*360000+b2*6000+b3*100+b4/10;
	if (b1 == 0 && b2 == 0 && b3 == 0 && b4 == 0)

	do {
		if (!stream_read_line (st, line, LINE_LEN)) return NULL;
	} while (sscanf (line, "Dialogue: Marked=%d,%d:%d:%d.%d,%d:%d:%d.%d,"
			"%[^\n\r]", &nothing,
			&hour1, &min1, &sec1, &hunsec1, 
			&hour2, &min2, &sec2, &hunsec2,
			line3) < 9
		 &&
		 sscanf (line, "Dialogue: %d,%d:%d:%d.%d,%d:%d:%d.%d,"
			 "%[^\n\r]", &nothing,
			 &hour1, &min1, &sec1, &hunsec1, 
			 &hour2, &min2, &sec2, &hunsec2,
			 line3) < 9	    );

        line2=strchr(line3, ',');

        for (comma = 4; comma < max_comma; comma ++)
          {
            tmp = line2;
            if(!(tmp=strchr(++tmp, ','))) break;
            if(*(++tmp) == ' ') break; 
                  /* a space after a comma means we're already in a sentence */
            line2 = tmp;
          }

        if(comma < max_comma)max_comma = comma;

	do
	{
		if (!stream_read_line(st, line, LINE_LEN)) return NULL;
	} while (sscanf (line, "%f %f", &a, &b) !=2);

	mpsub_position += a*mpsub_multiplier; 
	current->start=(int) mpsub_position;
	mpsub_position += b*mpsub_multiplier; 
	current->end=(int) mpsub_position;

	while (num < SUB_MAX_TEXT) {
		if (!stream_read_line (st, line, LINE_LEN)) {
			if (num == 0) return NULL;

        if (!stream_read_line (st, line, LINE_LEN))
		return NULL;
        if (!(sscanf (line, "-->> %ld", &(current->start)) <1))
		break;
    }
    
#ifdef CONFIG_SORTSUB
    previous_sub_end = (current->start) ? current->start - 1 : 0; 
#else
    if (previous_aqt_sub != NULL) 
	previous_aqt_sub->end = current->start-1;
    
    previous_aqt_sub = current;
#endif

    if (!stream_read_line (st, line, LINE_LEN))
	return NULL;

static subtitle *sub_read_line_subrip09(stream_t *st,subtitle *current) {
    char line[LINE_LEN+1];
    int a1,a2,a3;
    char * next=NULL;
    int i,len;
   
    while (1) {
    // try to locate next subtitle
        if (!stream_read_line (st, line, LINE_LEN))
		return NULL;
        if (!((len=sscanf (line, "[%d:%d:%d]",&a1,&a2,&a3)) < 3))
		break;
    }

    current->start = a1*360000+a2*6000+a3*100;
    
#ifdef CONFIG_SORTSUB
    previous_sub_end = (current->start) ? current->start - 1 : 0; 
#else
    if (previous_subrip09_sub != NULL) 
	previous_subrip09_sub->end = current->start-1;
    
    previous_subrip09_sub = current;
#endif

    if (!stream_read_line (st, line, LINE_LEN))
	return NULL;

    next = line,i=0;
    
    current->text[0]=""; // just to be sure that string is clear
    
    while ((next =sub_readtext (next, &(current->text[i])))) {
	if (current->text[i]==ERR) {return ERR;}
	i++;
	if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
	}

static int sub_autodetect (stream_t* st, int *uses_time) {
    char line[LINE_LEN+1];
    int i,j=0;
    char p;
    
    while (j < 100) {
	j++;
	if (!stream_read_line (st, line, LINE_LEN))
	    return SUB_INVALID;

		  if (st && st->flags & STREAM_SEEK ) {
		    cp_tmp = guess_cp(st, enca_lang, enca_fallback);
		  } else {
		    cp_tmp = enca_fallback;
		    if (st)
		      mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: enca failed, stream must be seekable.\n"); 
		  }
		}
#endif
		if ((icdsc = iconv_open (tocp, cp_tmp)) != (iconv_t)(-1)){
			mp_msg(MSGT_SUBREADER,MSGL_V,"SUB: opened iconv descriptor.\n");

  int char_set_num;
  fribidi_boolean log2vis;
  if(flip_hebrew) { // Please fix the indentation someday
  fribidi_set_mirroring(1);
  fribidi_set_reorder_nsm(0);
   
  if( sub_utf8 == 0 ) {
    char_set_num = fribidi_parse_charset (fribidi_charset?fribidi_charset:"ISO8859-8");
  }else {
    char_set_num = fribidi_parse_charset ("UTF-8");
  }

    mp_msg(MSGT_SUBREADER, MSGL_V, "ENCA supported languages: ");
    for (i = 0; i < langcnt; i++) {
	mp_msg(MSGT_SUBREADER, MSGL_V, "%s ", languages[i]);
    }
    mp_msg(MSGT_SUBREADER, MSGL_V, "\n");
    
    for (i = 0; i < langcnt; i++) {
	if (strcasecmp(languages[i], preferred_language) != 0) continue;
	analyser = enca_analyser_alloc(languages[i]);
	encoding = enca_analyse_const(analyser, buffer, buflen);
	enca_analyser_free(analyser);
	if (encoding.charset != ENCA_CS_UNKNOWN) {
	    detected_sub_cp = enca_charset_name(encoding.charset, ENCA_NAME_STYLE_ICONV);
	    break;
	}
    }
    
    free(languages);

    if (!detected_sub_cp) {
	detected_sub_cp = fallback;
	mp_msg(MSGT_SUBREADER, MSGL_INFO, "ENCA detection failed: fallback to %s\n", fallback);

    buffer = malloc(MAX_GUESS_BUFFER_SIZE);
    buflen = stream_read(st,buffer, MAX_GUESS_BUFFER_SIZE);

    detected_sub_cp = guess_buffer_cp(buffer, buflen, preferred_language, fallback);
    
    free(buffer);
    stream_reset(st);
    stream_seek(st,0);

    return detected_sub_cp;

	    { sub_read_line_subrip09, NULL, "subrip 0.9" },
	    { sub_read_line_jacosub, NULL, "jacosub" },
	    { sub_read_line_mpl2, NULL, "mpl2" }
    };
    struct subreader *srp;
    
    if(filename==NULL) return NULL; //qnx segfault
    i = 0;
    fd=open_stream (filename, NULL, &i); if (!fd) return NULL;
    
    sub_format=sub_autodetect (fd, &uses_time);
    mpsub_multiplier = (uses_time ? 100.0 : 1.0);
    if (sub_format==SUB_INVALID) {mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: Could not determine file format\n");return NULL;}
    srp=sr+sub_format;
    mp_msg(MSGT_SUBREADER, MSGL_V, "SUB: Detected subtitle file format: %s\n", srp->name);
    
    stream_reset(fd);
    stream_seek(fd,0);

#ifdef CONFIG_ICONV
    sub_utf8_prev=sub_utf8;

	  subcp_close();
          sub_utf8=sub_utf8_prev;
#endif
	    return NULL;
    }
    
#ifdef CONFIG_SORTSUB
    sub = malloc(sizeof(subtitle));
    //This is to deal with those formats (AQT & Subrip) which define the end of a subtitle
    //as the beginning of the following
    previous_sub_end = 0;
#endif    
    while(1){
        if(sub_num>=n_max){
            n_max+=16;
            first=realloc(first,n_max*sizeof(subtitle));
        }

	 {
#ifdef CONFIG_ICONV
          subcp_close();
#endif
    	  if ( first ) free(first);
	  return NULL; 
	 }
        // Apply any post processing that needs recoding first
        if ((sub!=ERR) && !sub_no_text_pp && srp->post) srp->post(sub);
#ifdef CONFIG_SORTSUB
	if(!sub_num || (first[sub_num - 1].start <= sub->start)){

	    }
	}
#endif	
        if(sub==ERR) ++sub_errs; else ++sub_num; // Error vs. Valid
    }
    
    free_stream(fd);

#ifdef CONFIG_ICONV
    subcp_close();
#endif

#if 0
char * strreplace( char * in,char * what,char * whereof )
{
 int i;
 char * tmp;
 
 if ( ( in == NULL )||( what == NULL )||( whereof == NULL )||( ( tmp=strstr( in,what ) ) == NULL ) ) return NULL;
 for( i=0;i<strlen( whereof );i++ ) tmp[i]=whereof[i];
 if ( strlen( what ) > strlen( whereof ) ) tmp[i]=0;
 return in;
}

	if (*s == 0) break;
	*d++ = ' ';
    }
    *d = 0;
}
 
static void strcpy_strip_ext(char *d, char *s)
{
    char *tmp = strrchr(s,'.');
    if (!tmp) {
	strcpy(d, s);

    while (*d) {
	*d = tolower(*d);
	d++;
    }
}
 
static void strcpy_get_ext(char *d, char *s)
{
    char *tmp = strrchr(s,'.');
    if (!tmp) {
	strcpy(d, "");

char** sub_filenames(const char* path, char *fname)
{
    char *f_dir, *f_fname, *f_fname_noext, *f_fname_trim, *tmp, *tmp_sub_id;
    char *tmp_fname_noext, *tmp_fname_trim, *tmp_fname_ext, *tmpresult;
 
    int len, pos, found, i, j;
    char * sub_exts[] = {  "utf", "utf8", "utf-8", "sub", "srt", "smi", "rt", "txt", "ssa", "aqt", "jss", "js", "ass", NULL};
    subfn *result;
    char **result2;
    
    int subcnt;
 
    FILE *f;

    DIR *d;
    struct dirent *de;

    tmpresult = malloc(len);

    result = malloc(sizeof(subfn)*MAX_SUBTITLE_FILES);
    memset(result, 0, sizeof(subfn)*MAX_SUBTITLE_FILES);
    
    subcnt = 0;
    
    tmp = strrchr(fname,'/');
#if defined(__MINGW32__) || defined(__CYGWIN__) || defined(__OS2__)
    if(!tmp)tmp = strrchr(fname,'\\');
    if(!tmp)tmp = strrchr(fname,':');
#endif
    
    // extract filename & dirname from fname
    if (tmp) {
	strcpy(f_fname, tmp+1);
	pos = tmp - fname;
	strncpy(f_dir, fname, pos+1);
	f_dir[pos+1] = 0;
    } else {
	strcpy(f_fname, fname);
	strcpy(f_dir, "./");
    }
 
    strcpy_strip_ext(f_fname_noext, f_fname);
    strcpy_trim(f_fname_trim, f_fname_noext);

    tmp_sub_id = NULL;
    if (dvdsub_lang && !whiteonly(dvdsub_lang)) {

		    if (strcasecmp(sub_exts[i], tmp_fname_ext) == 0) {
			found = 1;
			break;
		    }
		}
 
		// we have a (likely) subtitle file
		if (found) {
		    int prio = 0;
		    if (!prio && tmp_sub_id)
		    {
			sprintf(tmpresult, "%s %s", f_fname_trim, tmp_sub_id);
			mp_msg(MSGT_SUBREADER,MSGL_INFO,"dvdsublang...%s\n", tmpresult);
			if (strcmp(tmp_fname_trim, tmpresult) == 0 && sub_match_fuzziness >= 1) {
			    // matches the movie name + lang extension
			    prio = 5;
			}		    
		    }
		    if (!prio && strcmp(tmp_fname_trim, f_fname_trim) == 0) {
			// matches the movie name
			prio = 4;
		    }

		}
		if (subcnt >= MAX_SUBTITLE_FILES) break;
	    }
	    closedir(d);
	}
 
    }

    if (tmp_sub_id) free(tmp_sub_id);
    
    free(f_dir);
    free(f_fname);
    free(f_fname_noext);
    free(f_fname_trim);

    for (i = 0; i < subcnt; i++) {
	result2[i] = result[i].fname;
    }
    result2[subcnt] = NULL;
    
    free(result);

    return result2;
}

	    mp_msg(MSGT_SUBREADER,MSGL_INFO,"\t\t%d: %s%s", i,egysub->text[i], i==egysub->lines-1?"":" \n ");
	}
	mp_msg(MSGT_SUBREADER,MSGL_INFO,"\n");
    }

    mp_msg(MSGT_SUBREADER,MSGL_INFO,"Subtitle format %s time.\n", 
                                  subd->sub_uses_time ? "uses":"doesn't use");
    mp_msg(MSGT_SUBREADER,MSGL_INFO,"Read %i subtitles, %i errors.\n", subd->sub_num, subd->sub_errs);
}

void dump_srt(sub_data* subd, float fps){

    if (!subd->sub_uses_time && sub_fps == 0)
	sub_fps = fps;
    fd=fopen("dumpsub.srt","w");
    if(!fd)
    { 
	perror("dump_srt: fopen");
	return;
    }
    for(i=0; i < subd->sub_num; i++)
    {

			b=((egysub->end-egysub->start)/100.0);
			if ( (float)((int)a) == a)
			fprintf (fd, "%.0f",a);
			else
			fprintf (fd, "%.2f",a);
			    
			if ( (float)((int)b) == b)
			fprintf (fd, " %.0f\n",b);
			else
			fprintf (fd, " %.2f\n",b);
		} else {

	    end = end * sub_fps / fps;
	}
	start -= delay;
	end -= delay;
	fprintf(fd, "{%d}{%d}", start, end);
	for (j = 0; j < subs[i].lines; ++j) 
	    fprintf(fd, "%s%s", j ? "|" : "", subs[i].text[j]);
	fprintf(fd, "\n");
    }
    fclose(fd);
    mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.sub\'.\n");

    if (!subd->sub_uses_time && sub_fps == 0)
	sub_fps = fps;
    fd=fopen("dumpsub.jss","w");
    if(!fd)
    { 
	perror("dump_jacosub: fopen");
	return;
    }
    fprintf(fd, "#TIMERES %d\n", (subd->sub_uses_time) ? 100 : (int)sub_fps); 
    for(i=0; i < subd->sub_num; i++)
    {
        onesub=subs+i;    //=&subs[i];

	temp=onesub->start;

    if (!subd->sub_uses_time && sub_fps == 0)
	sub_fps = fps;
    fd=fopen("dumpsub.smi","w");
    if(!fd)
    { 
	perror("dump_jacosub: fopen");
	return;
    }
    fprintf(fd, "<SAMI>\n"
		"<HEAD>\n"

}

void sub_free( sub_data * subd )
{
 int i;
 
    if ( !subd ) return;
 
    if (subd->subtitles) {
	for (i=0; i < subd->subtitles->lines; i++) free( subd->subtitles->text[i] );
	free( subd->subtitles );
    }
    if (subd->filename) free( subd->filename );

      }
      double_newline = 1;
      buf[pos] = 0;
      sub->lines++;
      pos = 0;
      buf = malloc(MAX_SUBLINE + 1);      
      sub->text[sub->lines] = buf;
      sub->endpts[sub->lines] = endpts;
    }
  }
  buf[pos] = 0;

29251:cada7608488b

int sub_match_fuzziness=0; // level of sub name matching fuzziness

/* Use the SUB_* constant defined in the header file */
int sub_format=SUB_INVALID;
#ifdef CONFIG_SORTSUB
/*
   Some subtitling formats, namely AQT and Subrip09, define the end of a
   subtitle as the beginning of the following. Since currently we read one
   subtitle at time, for these format we keep two global *subtitle,
   previous_aqt_sub and previous_subrip09_sub, pointing to previous subtitle,
   so we can change its end when we read current subtitle starting time.

    do {
	switch (state) {

	case 0: /* find "START=" or "Slacktime:" */
	    slacktime_s = stristr (s, "Slacktime:");
	    if (slacktime_s)
                sub_slacktime = strtol (slacktime_s+10, NULL, 0) / 10;

	    s = stristr (s, "Start=");
	    if (s) {
		current->start = strtol (s + 6, &s, 0) / 10;

                for (; *s != '>' && *s != '\0'; s++);
                s++;
		state = 1; continue;
	    }
	    break;

	case 1: /* find (optionnal) "<P", skip other TAGs */
	    for  (; *s == ' ' || *s == '\t'; s++); /* strip blanks, if any */
	    if (*s == '\0') break;
	    if (*s != '<') { state = 3; p = text; continue; } /* not a TAG */
	    s++;

	    for (; *s != '>' && *s != '\0'; s++); /* skip remains of non-<P> TAG */
	    if (s == '\0')
	      break;
	    s++;
	    continue;

	case 2: /* find ">" */
	    if ((s = strchr (s, '>'))) { s++; state = 3; p = text; continue; }
	    break;

	case 3: /* get all text until '<' appears */
	    if (*s == '\0') break;
	    else if (!strncasecmp (s, "<br>", 4)) {
		*p = '\0'; p = text; trail_space (text);
		if (text[0] != '\0')

	    else if (*s == '\r' || *s == '\n') { s++; }
	    else *p++ = *s++;

	    /* skip duplicated space */
	    if (p > text + 2) if (*(p-1) == ' ' && *(p-2) == ' ') p--;

	    continue;

	case 4: /* get current->end or skip <TAG> */
	    q = stristr (s, "Start=");
	    if (q) {

		break; // if it is the last subtitle
	    } else {
		return 0;
	    }
	}

    } while (state != 99);

    // For the last subtitle
    if (current->end <= 0) {
        current->end = current->start + sub_slacktime;
	*p = '\0'; trail_space (text);
	if (text[0] != '\0')
	    current->text[current->lines++] = strdup (text);
    }

    return current;
}


static char *sub_readtext(char *source, char **dest) {
    int len=0;
    char *p=source;

//    printf("src=%p  dest=%p  \n",source,dest);

    while ( !eol(*p) && *p!= '|' ) {
	p++,len++;
    }

    *dest= malloc (len+1);
    if (!dest) {return ERR;}

    strncpy(*dest, source, len);
    (*dest)[len]=0;

    while (*p=='\r' || *p=='\n' || *p=='|') p++;

    if (*p) return p;  // not-last text field
    else return NULL;  // last text field
}

static subtitle *sub_read_line_microdvd(stream_t *st,subtitle *current) {

static subtitle *sub_read_line_subrip(stream_t* st, subtitle *current) {
    char line[LINE_LEN+1];
    int a1,a2,a3,a4,b1,b2,b3,b4;
    char *p=NULL, *q=NULL;
    int len;

    while (1) {
	if (!stream_read_line (st, line, LINE_LEN)) return NULL;
	if (sscanf (line, "%d:%d:%d.%d,%d:%d:%d.%d",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4) < 8) continue;
	current->start = a1*360000+a2*6000+a3*100+a4;
	current->end   = b1*360000+b2*6000+b3*100+b4;

static subtitle *sub_read_line_subviewer(stream_t *st,subtitle *current) {
    char line[LINE_LEN+1];
    int a1,a2,a3,a4,b1,b2,b3,b4;
    char *p=NULL;
    int i,len;

    while (!current->text[0]) {
	if (!stream_read_line (st, line, LINE_LEN)) return NULL;
	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)
	    continue;
	current->start = a1*360000+a2*6000+a3*100+a4/10;

static subtitle *sub_read_line_subviewer2(stream_t *st,subtitle *current) {
    char line[LINE_LEN+1];
    int a1,a2,a3,a4;
    char *p=NULL;
    int i,len;

    while (!current->text[0]) {
        if (!stream_read_line (st, line, LINE_LEN)) return NULL;
	if (line[0]!='{')
	    continue;
        if ((len=sscanf (line, "{T %d:%d:%d:%d",&a1,&a2,&a3,&a4)) < 4)

		if (!(current->start = a1*360000+a2*6000+a3*100))
			continue;
                /* removed by wodzu
		p=line;	
 		// finds the body of the subtitle
 		for (i=0; i<3; i++){
		   p=strchr(p,':');
		   if (p==NULL) break;
		   ++p;
		}
		if (p==NULL) {
		    printf("SUB: Skipping incorrect subtitle line!\n");
		    continue;
		}
                */

	}
	return current;
}

static subtitle *sub_read_line_rt(stream_t *st,subtitle *current) {
	//TODO: This format uses quite rich (sub/super)set of xhtml
	// I couldn't check it since DTD is not included.
	// WARNING: full XML parses can be required for proper parsing
    char line[LINE_LEN+1];
    int a1,a2,a3,a4,b1,b2,b3,b4;
    char *p=NULL,*next=NULL;
    int i,len,plen;

    while (!current->text[0]) {
	if (!stream_read_line (st, line, LINE_LEN)) return NULL;
	//TODO: it seems that format of time is not easily determined, it may be 1:12, 1:12.0 or 0:1:12.0
	//to describe the same moment in time. Maybe there are even more formats in use.
	//if ((len=sscanf (line, "<Time Begin=\"%d:%d:%d.%d\" End=\"%d:%d:%d.%d\"",&a1,&a2,&a3,&a4,&b1,&b2,&b3,&b4)) < 8)

	((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) &&
	//now try it without end time
	((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d.%d\"%*[^<]<clear/>%n",&a3,&a4,&plen)) < 2) &&
	((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d\"%*[^<]<clear/>%n",&a2,&a3,&plen)) < 2) &&
	((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d.%d\"%*[^<]<clear/>%n",&a2,&a3,&a4,&plen)) < 3) &&
	((len=sscanf (line, "<%*[tT]ime %*[bB]egin=\"%d:%d:%d.%d\"%*[^<]<clear/>%n",&a1,&a2,&a3,&a4,&plen)) < 4)
	)
	    continue;
	current->start = a1*360000+a2*6000+a3*100+a4/10;
	current->end   = b1*360000+b2*6000+b3*100+b4/10;
	if (b1 == 0 && b2 == 0 && b3 == 0 && b4 == 0)

	do {
		if (!stream_read_line (st, line, LINE_LEN)) return NULL;
	} while (sscanf (line, "Dialogue: Marked=%d,%d:%d:%d.%d,%d:%d:%d.%d,"
			"%[^\n\r]", &nothing,
			&hour1, &min1, &sec1, &hunsec1,
			&hour2, &min2, &sec2, &hunsec2,
			line3) < 9
		 &&
		 sscanf (line, "Dialogue: %d,%d:%d:%d.%d,%d:%d:%d.%d,"
			 "%[^\n\r]", &nothing,
			 &hour1, &min1, &sec1, &hunsec1,
			 &hour2, &min2, &sec2, &hunsec2,
			 line3) < 9	    );

        line2=strchr(line3, ',');

        for (comma = 4; comma < max_comma; comma ++)
          {
            tmp = line2;
            if(!(tmp=strchr(++tmp, ','))) break;
            if(*(++tmp) == ' ') break;
                  /* a space after a comma means we're already in a sentence */
            line2 = tmp;
          }

        if(comma < max_comma)max_comma = comma;

	do
	{
		if (!stream_read_line(st, line, LINE_LEN)) return NULL;
	} while (sscanf (line, "%f %f", &a, &b) !=2);

	mpsub_position += a*mpsub_multiplier;
	current->start=(int) mpsub_position;
	mpsub_position += b*mpsub_multiplier;
	current->end=(int) mpsub_position;

	while (num < SUB_MAX_TEXT) {
		if (!stream_read_line (st, line, LINE_LEN)) {
			if (num == 0) return NULL;

        if (!stream_read_line (st, line, LINE_LEN))
		return NULL;
        if (!(sscanf (line, "-->> %ld", &(current->start)) <1))
		break;
    }

#ifdef CONFIG_SORTSUB
    previous_sub_end = (current->start) ? current->start - 1 : 0;
#else
    if (previous_aqt_sub != NULL)
	previous_aqt_sub->end = current->start-1;

    previous_aqt_sub = current;
#endif

    if (!stream_read_line (st, line, LINE_LEN))
	return NULL;

static subtitle *sub_read_line_subrip09(stream_t *st,subtitle *current) {
    char line[LINE_LEN+1];
    int a1,a2,a3;
    char * next=NULL;
    int i,len;

    while (1) {
    // try to locate next subtitle
        if (!stream_read_line (st, line, LINE_LEN))
		return NULL;
        if (!((len=sscanf (line, "[%d:%d:%d]",&a1,&a2,&a3)) < 3))
		break;
    }

    current->start = a1*360000+a2*6000+a3*100;

#ifdef CONFIG_SORTSUB
    previous_sub_end = (current->start) ? current->start - 1 : 0;
#else
    if (previous_subrip09_sub != NULL)
	previous_subrip09_sub->end = current->start-1;

    previous_subrip09_sub = current;
#endif

    if (!stream_read_line (st, line, LINE_LEN))
	return NULL;

    next = line,i=0;

    current->text[0]=""; // just to be sure that string is clear

    while ((next =sub_readtext (next, &(current->text[i])))) {
	if (current->text[i]==ERR) {return ERR;}
	i++;
	if (i>=SUB_MAX_TEXT) { mp_msg(MSGT_SUBREADER,MSGL_WARN,"Too many lines in a subtitle\n");current->lines=i;return current;}
	}

static int sub_autodetect (stream_t* st, int *uses_time) {
    char line[LINE_LEN+1];
    int i,j=0;
    char p;

    while (j < 100) {
	j++;
	if (!stream_read_line (st, line, LINE_LEN))
	    return SUB_INVALID;

		  if (st && st->flags & STREAM_SEEK ) {
		    cp_tmp = guess_cp(st, enca_lang, enca_fallback);
		  } else {
		    cp_tmp = enca_fallback;
		    if (st)
		      mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: enca failed, stream must be seekable.\n");
		  }
		}
#endif
		if ((icdsc = iconv_open (tocp, cp_tmp)) != (iconv_t)(-1)){
			mp_msg(MSGT_SUBREADER,MSGL_V,"SUB: opened iconv descriptor.\n");

  int char_set_num;
  fribidi_boolean log2vis;
  if(flip_hebrew) { // Please fix the indentation someday
  fribidi_set_mirroring(1);
  fribidi_set_reorder_nsm(0);

  if( sub_utf8 == 0 ) {
    char_set_num = fribidi_parse_charset (fribidi_charset?fribidi_charset:"ISO8859-8");
  }else {
    char_set_num = fribidi_parse_charset ("UTF-8");
  }

    mp_msg(MSGT_SUBREADER, MSGL_V, "ENCA supported languages: ");
    for (i = 0; i < langcnt; i++) {
	mp_msg(MSGT_SUBREADER, MSGL_V, "%s ", languages[i]);
    }
    mp_msg(MSGT_SUBREADER, MSGL_V, "\n");

    for (i = 0; i < langcnt; i++) {
	if (strcasecmp(languages[i], preferred_language) != 0) continue;
	analyser = enca_analyser_alloc(languages[i]);
	encoding = enca_analyse_const(analyser, buffer, buflen);
	enca_analyser_free(analyser);
	if (encoding.charset != ENCA_CS_UNKNOWN) {
	    detected_sub_cp = enca_charset_name(encoding.charset, ENCA_NAME_STYLE_ICONV);
	    break;
	}
    }

    free(languages);

    if (!detected_sub_cp) {
	detected_sub_cp = fallback;
	mp_msg(MSGT_SUBREADER, MSGL_INFO, "ENCA detection failed: fallback to %s\n", fallback);

    buffer = malloc(MAX_GUESS_BUFFER_SIZE);
    buflen = stream_read(st,buffer, MAX_GUESS_BUFFER_SIZE);

    detected_sub_cp = guess_buffer_cp(buffer, buflen, preferred_language, fallback);

    free(buffer);
    stream_reset(st);
    stream_seek(st,0);

    return detected_sub_cp;

	    { sub_read_line_subrip09, NULL, "subrip 0.9" },
	    { sub_read_line_jacosub, NULL, "jacosub" },
	    { sub_read_line_mpl2, NULL, "mpl2" }
    };
    struct subreader *srp;

    if(filename==NULL) return NULL; //qnx segfault
    i = 0;
    fd=open_stream (filename, NULL, &i); if (!fd) return NULL;

    sub_format=sub_autodetect (fd, &uses_time);
    mpsub_multiplier = (uses_time ? 100.0 : 1.0);
    if (sub_format==SUB_INVALID) {mp_msg(MSGT_SUBREADER,MSGL_WARN,"SUB: Could not determine file format\n");return NULL;}
    srp=sr+sub_format;
    mp_msg(MSGT_SUBREADER, MSGL_V, "SUB: Detected subtitle file format: %s\n", srp->name);

    stream_reset(fd);
    stream_seek(fd,0);

#ifdef CONFIG_ICONV
    sub_utf8_prev=sub_utf8;

	  subcp_close();
          sub_utf8=sub_utf8_prev;
#endif
	    return NULL;
    }

#ifdef CONFIG_SORTSUB
    sub = malloc(sizeof(subtitle));
    //This is to deal with those formats (AQT & Subrip) which define the end of a subtitle
    //as the beginning of the following
    previous_sub_end = 0;
#endif
    while(1){
        if(sub_num>=n_max){
            n_max+=16;
            first=realloc(first,n_max*sizeof(subtitle));
        }

	 {
#ifdef CONFIG_ICONV
          subcp_close();
#endif
    	  if ( first ) free(first);
	  return NULL;
	 }
        // Apply any post processing that needs recoding first
        if ((sub!=ERR) && !sub_no_text_pp && srp->post) srp->post(sub);
#ifdef CONFIG_SORTSUB
	if(!sub_num || (first[sub_num - 1].start <= sub->start)){

	    }
	}
#endif	
        if(sub==ERR) ++sub_errs; else ++sub_num; // Error vs. Valid
    }

    free_stream(fd);

#ifdef CONFIG_ICONV
    subcp_close();
#endif

#if 0
char * strreplace( char * in,char * what,char * whereof )
{
 int i;
 char * tmp;

 if ( ( in == NULL )||( what == NULL )||( whereof == NULL )||( ( tmp=strstr( in,what ) ) == NULL ) ) return NULL;
 for( i=0;i<strlen( whereof );i++ ) tmp[i]=whereof[i];
 if ( strlen( what ) > strlen( whereof ) ) tmp[i]=0;
 return in;
}

	if (*s == 0) break;
	*d++ = ' ';
    }
    *d = 0;
}

static void strcpy_strip_ext(char *d, char *s)
{
    char *tmp = strrchr(s,'.');
    if (!tmp) {
	strcpy(d, s);

    while (*d) {
	*d = tolower(*d);
	d++;
    }
}

static void strcpy_get_ext(char *d, char *s)
{
    char *tmp = strrchr(s,'.');
    if (!tmp) {
	strcpy(d, "");

char** sub_filenames(const char* path, char *fname)
{
    char *f_dir, *f_fname, *f_fname_noext, *f_fname_trim, *tmp, *tmp_sub_id;
    char *tmp_fname_noext, *tmp_fname_trim, *tmp_fname_ext, *tmpresult;

    int len, pos, found, i, j;
    char * sub_exts[] = {  "utf", "utf8", "utf-8", "sub", "srt", "smi", "rt", "txt", "ssa", "aqt", "jss", "js", "ass", NULL};
    subfn *result;
    char **result2;

    int subcnt;

    FILE *f;

    DIR *d;
    struct dirent *de;

    tmpresult = malloc(len);

    result = malloc(sizeof(subfn)*MAX_SUBTITLE_FILES);
    memset(result, 0, sizeof(subfn)*MAX_SUBTITLE_FILES);

    subcnt = 0;

    tmp = strrchr(fname,'/');
#if defined(__MINGW32__) || defined(__CYGWIN__) || defined(__OS2__)
    if(!tmp)tmp = strrchr(fname,'\\');
    if(!tmp)tmp = strrchr(fname,':');
#endif

    // extract filename & dirname from fname
    if (tmp) {
	strcpy(f_fname, tmp+1);
	pos = tmp - fname;
	strncpy(f_dir, fname, pos+1);
	f_dir[pos+1] = 0;
    } else {
	strcpy(f_fname, fname);
	strcpy(f_dir, "./");
    }

    strcpy_strip_ext(f_fname_noext, f_fname);
    strcpy_trim(f_fname_trim, f_fname_noext);

    tmp_sub_id = NULL;
    if (dvdsub_lang && !whiteonly(dvdsub_lang)) {

		    if (strcasecmp(sub_exts[i], tmp_fname_ext) == 0) {
			found = 1;
			break;
		    }
		}

		// we have a (likely) subtitle file
		if (found) {
		    int prio = 0;
		    if (!prio && tmp_sub_id)
		    {
			sprintf(tmpresult, "%s %s", f_fname_trim, tmp_sub_id);
			mp_msg(MSGT_SUBREADER,MSGL_INFO,"dvdsublang...%s\n", tmpresult);
			if (strcmp(tmp_fname_trim, tmpresult) == 0 && sub_match_fuzziness >= 1) {
			    // matches the movie name + lang extension
			    prio = 5;
			}
		    }
		    if (!prio && strcmp(tmp_fname_trim, f_fname_trim) == 0) {
			// matches the movie name
			prio = 4;
		    }

		}
		if (subcnt >= MAX_SUBTITLE_FILES) break;
	    }
	    closedir(d);
	}

    }

    if (tmp_sub_id) free(tmp_sub_id);

    free(f_dir);
    free(f_fname);
    free(f_fname_noext);
    free(f_fname_trim);

    for (i = 0; i < subcnt; i++) {
	result2[i] = result[i].fname;
    }
    result2[subcnt] = NULL;

    free(result);

    return result2;
}

	    mp_msg(MSGT_SUBREADER,MSGL_INFO,"\t\t%d: %s%s", i,egysub->text[i], i==egysub->lines-1?"":" \n ");
	}
	mp_msg(MSGT_SUBREADER,MSGL_INFO,"\n");
    }

    mp_msg(MSGT_SUBREADER,MSGL_INFO,"Subtitle format %s time.\n",
                                  subd->sub_uses_time ? "uses":"doesn't use");
    mp_msg(MSGT_SUBREADER,MSGL_INFO,"Read %i subtitles, %i errors.\n", subd->sub_num, subd->sub_errs);
}

void dump_srt(sub_data* subd, float fps){

    if (!subd->sub_uses_time && sub_fps == 0)
	sub_fps = fps;
    fd=fopen("dumpsub.srt","w");
    if(!fd)
    {
	perror("dump_srt: fopen");
	return;
    }
    for(i=0; i < subd->sub_num; i++)
    {

			b=((egysub->end-egysub->start)/100.0);
			if ( (float)((int)a) == a)
			fprintf (fd, "%.0f",a);
			else
			fprintf (fd, "%.2f",a);

			if ( (float)((int)b) == b)
			fprintf (fd, " %.0f\n",b);
			else
			fprintf (fd, " %.2f\n",b);
		} else {

	    end = end * sub_fps / fps;
	}
	start -= delay;
	end -= delay;
	fprintf(fd, "{%d}{%d}", start, end);
	for (j = 0; j < subs[i].lines; ++j)
	    fprintf(fd, "%s%s", j ? "|" : "", subs[i].text[j]);
	fprintf(fd, "\n");
    }
    fclose(fd);
    mp_msg(MSGT_SUBREADER,MSGL_INFO,"SUB: Subtitles dumped in \'dumpsub.sub\'.\n");

    if (!subd->sub_uses_time && sub_fps == 0)
	sub_fps = fps;
    fd=fopen("dumpsub.jss","w");
    if(!fd)
    {
	perror("dump_jacosub: fopen");
	return;
    }
    fprintf(fd, "#TIMERES %d\n", (subd->sub_uses_time) ? 100 : (int)sub_fps);
    for(i=0; i < subd->sub_num; i++)
    {
        onesub=subs+i;    //=&subs[i];

	temp=onesub->start;

    if (!subd->sub_uses_time && sub_fps == 0)
	sub_fps = fps;
    fd=fopen("dumpsub.smi","w");
    if(!fd)
    {
	perror("dump_jacosub: fopen");
	return;
    }
    fprintf(fd, "<SAMI>\n"
		"<HEAD>\n"

}

void sub_free( sub_data * subd )
{
 int i;

    if ( !subd ) return;

    if (subd->subtitles) {
	for (i=0; i < subd->subtitles->lines; i++) free( subd->subtitles->text[i] );
	free( subd->subtitles );
    }
    if (subd->filename) free( subd->filename );

      }
      double_newline = 1;
      buf[pos] = 0;
      sub->lines++;
      pos = 0;
      buf = malloc(MAX_SUBLINE + 1);
      sub->text[sub->lines] = buf;
      sub->endpts[sub->lines] = endpts;
    }
  }
  buf[pos] = 0;