1 /*****************************************************************************
2 * subsdec.c : text subtitles decoder
3 *****************************************************************************
4 * Copyright (C) 2000-2006 the VideoLAN team
7 * Authors: Gildas Bazin <gbazin@videolan.org>
8 * Samuel Hocevar <sam@zoy.org>
9 * Derk-Jan Hartman <hartman at videolan dot org>
10 * Bernie Purcell <bitmap@videolan.org>
12 * This program is free software; you can redistribute it and/or modify
13 * it under the terms of the GNU General Public License as published by
14 * the Free Software Foundation; either version 2 of the License, or
15 * (at your option) any later version.
17 * This program is distributed in the hope that it will be useful,
18 * but WITHOUT ANY WARRANTY; without even the implied warranty of
19 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20 * GNU General Public License for more details.
22 * You should have received a copy of the GNU General Public License
23 * along with this program; if not, write to the Free Software
24 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
25 *****************************************************************************/
27 /*****************************************************************************
29 *****************************************************************************/
35 #include <vlc_plugin.h>
37 /*****************************************************************************
39 *****************************************************************************/
40 static int OpenDecoder ( vlc_object_t * );
41 static void CloseDecoder ( vlc_object_t * );
43 static subpicture_t *DecodeBlock ( decoder_t *, block_t ** );
44 static subpicture_t *ParseText ( decoder_t *, block_t * );
45 static char *StripTags ( char * );
46 static char *CreateHtmlSubtitle( int *pi_align, char * );
49 /*****************************************************************************
51 *****************************************************************************/
52 static const char *const ppsz_encodings[] = {
96 static const char *const ppsz_encoding_names[] = {
98 N_("Universal (UTF-8)"),
99 N_("Universal (UTF-16)"),
100 N_("Universal (big endian UTF-16)"),
101 N_("Universal (little endian UTF-16)"),
102 N_("Universal, Chinese (GB18030)"),
104 /* ISO 8859 and the likes */
106 N_("Western European (Latin-9)"), /* mostly superset of Latin-1 */
107 N_("Western European (Windows-1252)"),
109 N_("Eastern European (Latin-2)"),
110 N_("Eastern European (Windows-1250)"),
112 N_("Esperanto (Latin-3)"),
114 N_("Nordic (Latin-6)"), /* Latin 6 supersedes Latin 4 */
116 N_("Cyrillic (Windows-1251)"), /* ISO 8859-5 is not practically used */
117 N_("Russian (KOI8-R)"),
118 N_("Ukrainian (KOI8-U)"),
120 N_("Arabic (ISO 8859-6)"),
121 N_("Arabic (Windows-1256)"),
123 N_("Greek (ISO 8859-7)"),
124 N_("Greek (Windows-1256)"),
126 N_("Hebrew (ISO 8859-8)"),
127 N_("Hebrew (Windows-1255)"),
129 N_("Turkish (ISO 8859-9)"),
130 N_("Turkish (Windows-1254)"),
133 N_("Thai (TIS 620-2533/ISO 8859-11)"),
134 N_("Thai (Windows-874)"),
136 N_("Baltic (Latin-7)"),
137 N_("Baltic (Windows-1257)"),
138 /* 12 -> /dev/null */
140 N_("Celtic (Latin-8)"),
143 N_("South-Eastern European (Latin-10)"),
145 N_("Simplified Chinese (ISO-2022-CN-EXT)"),
146 N_("Simplified Chinese Unix (EUC-CN)"),
147 N_("Japanese (7-bits JIS/ISO-2022-JP-2)"),
148 N_("Japanese Unix (EUC-JP)"),
149 N_("Japanese (Shift JIS)"),
150 N_("Korean (ISO-2022-KR)"),
151 N_("Korean Unix (EUC-KR)"),
152 N_("Traditional Chinese (Big5)"),
153 N_("Traditional Chinese Unix (EUC-TW)"),
154 N_("Hong-Kong Supplementary (HKSCS)"),
156 N_("Vietnamese (VISCII)"),
157 N_("Vietnamese (Windows-1258)"),
160 SSA supports charset selection.
161 The following known charsets are used:
163 0 = Ansi - Western European
168 128 = Japanese (Shift JIS)
171 134 = GB2312 Simplified Chinese
172 136 = Big5 Traditional Chinese
179 204 = Russian (Cyrillic)
181 238 = Eastern European
185 static const int pi_justification[] = { 0, 1, 2 };
186 static const char *const ppsz_justification_text[] = {
187 N_("Center"),N_("Left"),N_("Right")};
189 #define ENCODING_TEXT N_("Subtitles text encoding")
190 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
191 #define ALIGN_TEXT N_("Subtitles justification")
192 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
193 #define AUTODETECT_UTF8_TEXT N_("UTF-8 subtitles autodetection")
194 #define AUTODETECT_UTF8_LONGTEXT N_("This enables automatic detection of " \
195 "UTF-8 encoding within subtitles files.")
196 #define FORMAT_TEXT N_("Formatted Subtitles")
197 #define FORMAT_LONGTEXT N_("Some subtitle formats allow for text formatting. " \
198 "VLC partly implements this, but you can choose to disable all formatting.")
202 set_shortname( N_("Subtitles"))
203 set_description( N_("Text subtitles decoder") )
204 set_capability( "decoder", 50 )
205 set_callbacks( OpenDecoder, CloseDecoder )
206 set_category( CAT_INPUT )
207 set_subcategory( SUBCAT_INPUT_SCODEC )
209 add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
211 change_integer_list( pi_justification, ppsz_justification_text, NULL )
212 add_string( "subsdec-encoding", "", NULL,
213 ENCODING_TEXT, ENCODING_LONGTEXT, false )
214 change_string_list( ppsz_encodings, ppsz_encoding_names, 0 )
215 add_bool( "subsdec-autodetect-utf8", true, NULL,
216 AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, false )
217 add_bool( "subsdec-formatted", true, NULL, FORMAT_TEXT, FORMAT_LONGTEXT,
221 /*****************************************************************************
222 * OpenDecoder: probe the decoder and return score
223 *****************************************************************************
224 * Tries to launch a decoder and return score so that the interface is able
226 *****************************************************************************/
227 static int OpenDecoder( vlc_object_t *p_this )
229 decoder_t *p_dec = (decoder_t*)p_this;
230 decoder_sys_t *p_sys;
233 switch( p_dec->fmt_in.i_codec )
237 case VLC_CODEC_ITU_T140:
243 p_dec->pf_decode_sub = DecodeBlock;
244 p_dec->fmt_out.i_cat = SPU_ES;
245 p_dec->fmt_out.i_codec = 0;
247 /* Allocate the memory needed to store the decoder's structure */
248 p_dec->p_sys = p_sys = calloc( 1, sizeof( *p_sys ) );
254 p_sys->iconv_handle = (vlc_iconv_t)-1;
255 p_sys->b_autodetect_utf8 = false;
256 p_sys->b_ass = false;
257 p_sys->i_original_height = -1;
258 p_sys->i_original_width = -1;
259 TAB_INIT( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
260 TAB_INIT( p_sys->i_images, p_sys->pp_images );
262 char *psz_charset = NULL;
264 /* First try demux-specified encoding */
265 if( p_dec->fmt_in.i_codec == VLC_CODEC_ITU_T140 )
266 psz_charset = strdup( "UTF-8" ); /* IUT T.140 is always using UTF-8 */
268 if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
270 psz_charset = strdup (p_dec->fmt_in.subs.psz_encoding);
271 msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
272 p_dec->fmt_in.subs.psz_encoding ?
273 p_dec->fmt_in.subs.psz_encoding : "not specified");
276 /* Second, try configured encoding */
277 if (psz_charset == NULL)
279 psz_charset = var_CreateGetNonEmptyString (p_dec, "subsdec-encoding");
280 msg_Dbg (p_dec, "trying configured character encoding: %s",
281 psz_charset ? psz_charset : "not specified");
284 /* Third, try "local" encoding with optional UTF-8 autodetection */
285 if (psz_charset == NULL)
287 psz_charset = strdup (GetFallbackEncoding ());
288 msg_Dbg (p_dec, "trying default character encoding: %s",
289 psz_charset ? psz_charset : "not specified");
291 if (var_CreateGetBool (p_dec, "subsdec-autodetect-utf8"))
293 msg_Dbg (p_dec, "using automatic UTF-8 detection");
294 p_sys->b_autodetect_utf8 = true;
298 /* Forth, don't do character decoding, i.e. assume UTF-8 */
299 if (psz_charset == NULL)
301 psz_charset = strdup ("UTF-8");
302 msg_Dbg (p_dec, "using UTF-8 character encoding" );
305 if ((psz_charset != NULL)
306 && strcasecmp (psz_charset, "UTF-8")
307 && strcasecmp (psz_charset, "utf8"))
309 p_sys->iconv_handle = vlc_iconv_open ("UTF-8", psz_charset);
310 if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
311 msg_Err (p_dec, "cannot convert from %s: %m", psz_charset);
315 var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
316 var_Get( p_dec, "subsdec-align", &val );
317 p_sys->i_align = val.i_int;
319 if( p_dec->fmt_in.i_codec == VLC_CODEC_SSA
320 && var_CreateGetBool( p_dec, "subsdec-formatted" ) )
322 if( p_dec->fmt_in.i_extra > 0 )
323 ParseSSAHeader( p_dec );
329 /****************************************************************************
330 * DecodeBlock: the whole thing
331 ****************************************************************************
332 * This function must be fed with complete subtitles units.
333 ****************************************************************************/
334 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
339 if( !pp_block || *pp_block == NULL )
343 if( p_block->i_flags & (BLOCK_FLAG_DISCONTINUITY|BLOCK_FLAG_CORRUPTED) )
345 block_Release( p_block );
349 p_spu = ParseText( p_dec, p_block );
351 block_Release( p_block );
357 /*****************************************************************************
358 * CloseDecoder: clean up the decoder
359 *****************************************************************************/
360 static void CloseDecoder( vlc_object_t *p_this )
362 decoder_t *p_dec = (decoder_t *)p_this;
363 decoder_sys_t *p_sys = p_dec->p_sys;
365 if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
366 vlc_iconv_close( p_sys->iconv_handle );
368 if( p_sys->pp_ssa_styles )
371 for( i = 0; i < p_sys->i_ssa_styles; i++ )
373 if( !p_sys->pp_ssa_styles[i] )
376 free( p_sys->pp_ssa_styles[i]->psz_stylename );
377 free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
378 free( p_sys->pp_ssa_styles[i] );
380 TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
382 if( p_sys->pp_images )
385 for( i = 0; i < p_sys->i_images; i++ )
387 if( !p_sys->pp_images[i] )
390 if( p_sys->pp_images[i]->p_pic )
391 picture_Release( p_sys->pp_images[i]->p_pic );
392 free( p_sys->pp_images[i]->psz_filename );
394 free( p_sys->pp_images[i] );
396 TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
402 /*****************************************************************************
403 * ParseText: parse an text subtitle packet and send it to the video output
404 *****************************************************************************/
405 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
407 decoder_sys_t *p_sys = p_dec->p_sys;
408 subpicture_t *p_spu = NULL;
409 char *psz_subtitle = NULL;
412 /* We cannot display a subpicture with no date */
413 if( p_block->i_pts == 0 )
415 msg_Warn( p_dec, "subtitle without a date" );
419 /* Check validity of packet data */
420 /* An "empty" line containing only \0 can be used to force
421 and ephemer picture from the screen */
422 if( p_block->i_buffer < 1 )
424 msg_Warn( p_dec, "no subtitle data" );
428 /* Should be resiliant against bad subtitles */
429 psz_subtitle = strndup( (const char *)p_block->p_buffer,
431 if( psz_subtitle == NULL )
434 if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
436 if (EnsureUTF8( psz_subtitle ) == NULL)
438 msg_Err( p_dec, "failed to convert subtitle encoding.\n"
439 "Try manually setting a character-encoding "
440 "before you open the file." );
446 if( p_sys->b_autodetect_utf8 )
448 if( IsUTF8( psz_subtitle ) == NULL )
450 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
451 "disabling UTF-8 subtitles autodetection" );
452 p_sys->b_autodetect_utf8 = false;
456 if( !p_sys->b_autodetect_utf8 )
458 size_t inbytes_left = strlen( psz_subtitle );
459 size_t outbytes_left = 6 * inbytes_left;
460 char *psz_new_subtitle = malloc( outbytes_left + 1 );
461 char *psz_convert_buffer_out = psz_new_subtitle;
462 const char *psz_convert_buffer_in = psz_subtitle;
464 size_t ret = vlc_iconv( p_sys->iconv_handle,
465 &psz_convert_buffer_in, &inbytes_left,
466 &psz_convert_buffer_out, &outbytes_left );
468 *psz_convert_buffer_out++ = '\0';
469 free( psz_subtitle );
471 if( ( ret == (size_t)(-1) ) || inbytes_left )
473 free( psz_new_subtitle );
474 msg_Err( p_dec, "failed to convert subtitle encoding.\n"
475 "Try manually setting a character-encoding "
476 "before you open the file." );
480 psz_subtitle = realloc( psz_new_subtitle,
481 psz_convert_buffer_out - psz_new_subtitle );
485 /* Create the subpicture unit */
486 p_spu = decoder_NewSubpicture( p_dec );
489 msg_Warn( p_dec, "can't get spu buffer" );
490 free( psz_subtitle );
494 /* Create a new subpicture region */
495 memset( &fmt, 0, sizeof(video_format_t) );
496 fmt.i_chroma = VLC_CODEC_TEXT;
498 fmt.i_width = fmt.i_height = 0;
499 fmt.i_x_offset = fmt.i_y_offset = 0;
500 p_spu->p_region = subpicture_region_New( &fmt );
501 if( !p_spu->p_region )
503 msg_Err( p_dec, "cannot allocate SPU region" );
504 free( psz_subtitle );
505 decoder_DeleteSubpicture( p_dec, p_spu );
509 /* Decode and format the subpicture unit */
510 if( p_dec->fmt_in.i_codec != VLC_CODEC_SSA )
512 /* Normal text subs, easy markup */
513 p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
514 p_spu->p_region->i_x = p_sys->i_align ? 20 : 0;
515 p_spu->p_region->i_y = 10;
517 /* Remove formatting from string */
519 p_spu->p_region->psz_text = StripTags( psz_subtitle );
520 if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
522 p_spu->p_region->psz_html = CreateHtmlSubtitle( &p_spu->p_region->i_align, psz_subtitle );
525 p_spu->i_start = p_block->i_pts;
526 p_spu->i_stop = p_block->i_pts + p_block->i_length;
527 p_spu->b_ephemer = (p_block->i_length == 0);
528 p_spu->b_absolute = false;
532 /* Decode SSA/USF strings */
533 ParseSSAString( p_dec, psz_subtitle, p_spu );
535 p_spu->i_start = p_block->i_pts;
536 p_spu->i_stop = p_block->i_pts + p_block->i_length;
537 p_spu->b_ephemer = (p_block->i_length == 0);
538 p_spu->b_absolute = false;
539 p_spu->i_original_picture_width = p_sys->i_original_width;
540 p_spu->i_original_picture_height = p_sys->i_original_height;
542 free( psz_subtitle );
547 char* GotoNextLine( char *psz_text )
549 char *p_newline = psz_text;
551 while( p_newline[0] != '\0' )
553 if( p_newline[0] == '\n' || p_newline[0] == '\r' )
556 while( p_newline[0] == '\n' || p_newline[0] == '\r' )
565 /* Function now handles tags with attribute values, and tries
566 * to deal with &' commands too. It no longer modifies the string
567 * in place, so that the original text can be reused
569 static char *StripTags( char *psz_subtitle )
571 char *psz_text_start;
574 psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
575 if( !psz_text_start )
578 while( *psz_subtitle )
580 if( *psz_subtitle == '<' )
582 if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
585 psz_subtitle += strcspn( psz_subtitle, ">" );
587 else if( *psz_subtitle == '&' )
589 if( !strncasecmp( psz_subtitle, "<", 4 ))
592 psz_subtitle += strcspn( psz_subtitle, ";" );
594 else if( !strncasecmp( psz_subtitle, ">", 4 ))
597 psz_subtitle += strcspn( psz_subtitle, ";" );
599 else if( !strncasecmp( psz_subtitle, "&", 5 ))
602 psz_subtitle += strcspn( psz_subtitle, ";" );
604 else if( !strncasecmp( psz_subtitle, """, 6 ))
607 psz_subtitle += strcspn( psz_subtitle, ";" );
611 /* Assume it is just a normal ampersand */
617 *psz_text++ = *psz_subtitle;
623 psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
625 return psz_text_start;
628 /* Try to respect any style tags present in the subtitle string. The main
629 * problem here is a lack of adequate specs for the subtitle formats.
630 * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
631 * SAMI has a detailed spec, but extensive rework is needed in the demux
632 * code to prevent all this style information being excised, as it presently
634 * That leaves the others - none of which were (I guess) originally intended
635 * to be carrying style information. Over time people have used them that way.
636 * In the absence of specifications from which to work, the tags supported
637 * have been restricted to the simple set permitted by the USF DTD, ie. :
638 * Basic: <br>, <i>, <b>, <u>
650 * There is also the further restriction that the subtitle be well-formed
651 * as an XML entity, ie. the HTML sentence:
652 * <b><i>Bold and Italics</b></i>
653 * doesn't qualify because the tags aren't nested one inside the other.
654 * <text> tags are automatically added to the output to ensure
656 * If the text doesn't qualify for any reason, a NULL string is
657 * returned, and the rendering engine will fall back to the
658 * plain text version of the subtitle.
660 static void HtmlNPut( char **ppsz_html, const char *psz_text, int i_max )
662 const int i_len = strlen(psz_text);
664 strncpy( *ppsz_html, psz_text, i_max );
665 *ppsz_html += __MIN(i_max,i_len);
668 static void HtmlPut( char **ppsz_html, const char *psz_text )
670 strcpy( *ppsz_html, psz_text );
671 *ppsz_html += strlen(psz_text);
673 static void HtmlCopy( char **ppsz_html, char **ppsz_subtitle, const char *psz_text )
675 HtmlPut( ppsz_html, psz_text );
676 *ppsz_subtitle += strlen(psz_text);
679 static char *CreateHtmlSubtitle( int *pi_align, char *psz_subtitle )
682 char *psz_tag = malloc( ( strlen( psz_subtitle ) / 3 ) + 1 );
688 size_t i_buf_size = strlen( psz_subtitle ) + 100;
689 char *psz_html_start = malloc( i_buf_size );
690 char *psz_html = psz_html_start;
691 if( psz_html_start == NULL )
698 bool b_has_align = false;
700 HtmlPut( &psz_html, "<text>" );
703 while( *psz_subtitle )
705 if( *psz_subtitle == '\n' )
707 HtmlPut( &psz_html, "<br/>" );
710 else if( *psz_subtitle == '<' )
712 if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
714 HtmlCopy( &psz_html, &psz_subtitle, "<br/>" );
716 else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
718 HtmlCopy( &psz_html, &psz_subtitle, "<b>" );
719 strcat( psz_tag, "b" );
721 else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
723 HtmlCopy( &psz_html, &psz_subtitle, "<i>" );
724 strcat( psz_tag, "i" );
726 else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
728 HtmlCopy( &psz_html, &psz_subtitle, "<u>" );
729 strcat( psz_tag, "u" );
731 else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
733 const char *psz_attribs[] = { "face=", "family=", "size=",
734 "color=", "outline-color=", "shadow-color=",
735 "outline-level=", "shadow-level=", "back-color=",
738 HtmlCopy( &psz_html, &psz_subtitle, "<font " );
739 strcat( psz_tag, "f" );
741 while( *psz_subtitle != '>' )
745 for( k=0; psz_attribs[ k ]; k++ )
747 int i_len = strlen( psz_attribs[ k ] );
749 if( !strncasecmp( psz_subtitle, psz_attribs[k], i_len ) )
752 HtmlPut( &psz_html, psz_attribs[k] );
753 psz_subtitle += i_len;
756 if( *psz_subtitle == '"' )
759 i_len = strcspn( psz_subtitle, "\"" );
763 i_len = strcspn( psz_subtitle, " \t>" );
765 HtmlPut( &psz_html, "\"" );
766 if( !strcmp( psz_attribs[ k ], "color=" ) && *psz_subtitle >= '0' && *psz_subtitle <= '9' )
767 HtmlPut( &psz_html, "#" );
768 HtmlNPut( &psz_html, psz_subtitle, i_len );
769 HtmlPut( &psz_html, "\"" );
771 psz_subtitle += i_len;
772 if( *psz_subtitle == '\"' )
777 if( psz_attribs[ k ] == NULL )
779 /* Jump over unrecognised tag */
780 int i_len = strcspn( psz_subtitle, "\"" ) + 1;
782 i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
783 psz_subtitle += i_len;
785 while (*psz_subtitle == ' ')
786 *psz_html++ = *psz_subtitle++;
788 *psz_html++ = *psz_subtitle++;
790 else if( !strncmp( psz_subtitle, "</", 2 ))
792 bool b_match = false;
793 bool b_ignore = false;
794 int i_len = strlen( psz_tag ) - 1;
795 char *psz_lastTag = NULL;
799 psz_lastTag = psz_tag + i_len;
802 switch( *psz_lastTag )
805 b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
809 b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
813 b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
817 b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
821 i_len = strcspn( psz_subtitle, ">" );
822 b_match = psz_subtitle[i_len] == '>';
831 /* Not well formed -- kill everything */
832 free( psz_html_start );
833 psz_html_start = NULL;
838 HtmlNPut( &psz_html, psz_subtitle, i_len );
840 psz_subtitle += i_len;
842 else if( ( psz_subtitle[1] < 'a' || psz_subtitle[1] > 'z' ) &&
843 ( psz_subtitle[1] < 'A' || psz_subtitle[1] > 'Z' ) )
845 /* We have a single < */
846 HtmlPut( &psz_html, "<" );
851 /* We have an unknown tag or a single < */
853 /* Search for the next tag or end of tag or end of string */
854 char *psz_stop = psz_subtitle + 1 + strcspn( &psz_subtitle[1], "<>" );
855 char *psz_closing = strstr( psz_subtitle, "/>" );
857 if( psz_closing && psz_closing < psz_stop )
859 /* We have a self closed tag, remove it */
860 psz_subtitle = &psz_closing[2];
862 else if( *psz_stop == '>' )
866 snprintf( psz_match, sizeof(psz_match), "</%s", &psz_subtitle[1] );
867 psz_match[strcspn( psz_match, " \t>" )] = '\0';
869 if( strstr( psz_subtitle, psz_match ) )
871 /* We have the closing tag, ignore it TODO */
872 psz_subtitle = &psz_stop[1];
873 strcat( psz_tag, "I" );
877 int i_len = psz_stop + 1 - psz_subtitle;
879 /* Copy the whole data */
880 for( ; i_len > 0; i_len--, psz_subtitle++ )
882 if( *psz_subtitle == '<' )
883 HtmlPut( &psz_html, "<" );
884 else if( *psz_subtitle == '>' )
885 HtmlPut( &psz_html, ">" );
887 *psz_html++ = *psz_subtitle;
893 /* We have a single < */
894 HtmlPut( &psz_html, "<" );
899 else if( *psz_subtitle == '&' )
901 if( !strncasecmp( psz_subtitle, "<", 4 ))
903 HtmlCopy( &psz_html, &psz_subtitle, "<" );
905 else if( !strncasecmp( psz_subtitle, ">", 4 ))
907 HtmlCopy( &psz_html, &psz_subtitle, ">" );
909 else if( !strncasecmp( psz_subtitle, "&", 5 ))
911 HtmlCopy( &psz_html, &psz_subtitle, "&" );
915 HtmlPut( &psz_html, "&" );
919 else if( *psz_subtitle == '>' )
921 HtmlPut( &psz_html, ">" );
924 else if( psz_subtitle[0] == '{' && psz_subtitle[1] == '\\' &&
925 strchr( psz_subtitle, '}' ) )
927 /* Check for forced alignment */
929 !strncmp( psz_subtitle, "{\\an", 4 ) && psz_subtitle[4] >= '1' && psz_subtitle[4] <= '9' && psz_subtitle[5] == '}' )
931 static const int pi_vertical[3] = { SUBPICTURE_ALIGN_BOTTOM, 0, SUBPICTURE_ALIGN_TOP };
932 static const int pi_horizontal[3] = { SUBPICTURE_ALIGN_LEFT, 0, SUBPICTURE_ALIGN_RIGHT };
933 const int i_id = psz_subtitle[4] - '1';
936 *pi_align = pi_vertical[i_id/3] | pi_horizontal[i_id%3];
938 /* TODO fr -> rotation */
940 /* Hide {\stupidity} */
941 psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
945 *psz_html = *psz_subtitle;
946 if( psz_html > psz_html_start )
948 /* Check for double whitespace */
949 if( ( *psz_html == ' ' || *psz_html == '\t' ) &&
950 ( *(psz_html-1) == ' ' || *(psz_html-1) == '\t' ) )
952 HtmlPut( &psz_html, NO_BREAKING_SPACE );
960 if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 50 )
962 const int i_len = psz_html - psz_html_start;
965 char *psz_new = realloc( psz_html_start, i_buf_size );
968 psz_html_start = psz_new;
969 psz_html = &psz_new[i_len];
974 static const char *psz_text_close = "</text>";
975 static const char *psz_tag_long = "/font>";
977 /* Realloc for closing tags and shrink memory */
978 const size_t i_length = (size_t)( psz_html - psz_html_start );
980 const size_t i_size = i_length + strlen(psz_tag_long) * strlen(psz_tag) + strlen(psz_text_close) + 1;
981 char *psz_new = realloc( psz_html_start, i_size );
984 psz_html_start = psz_new;
985 psz_html = &psz_new[i_length];
987 /* Close not well formed subtitle */
991 char *psz_last = &psz_tag[strlen(psz_tag)-1];
995 HtmlPut( &psz_html, "</b>" );
998 HtmlPut( &psz_html, "</i>" );
1001 HtmlPut( &psz_html, "</u>" );
1004 HtmlPut( &psz_html, "/font>" );
1012 HtmlPut( &psz_html, psz_text_close );
1017 return psz_html_start;