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 *****************************************************************************/
33 /*****************************************************************************
35 *****************************************************************************/
36 static int OpenDecoder ( vlc_object_t * );
37 static void CloseDecoder ( vlc_object_t * );
39 static subpicture_t *DecodeBlock ( decoder_t *, block_t ** );
40 static subpicture_t *ParseText ( decoder_t *, block_t * );
41 static char *StripTags ( char * );
42 static char *CreateHtmlSubtitle ( char * );
45 /*****************************************************************************
47 *****************************************************************************/
48 static const char *ppsz_encodings[] = { DEFAULT_NAME, "ASCII", "UTF-8", "",
49 "ISO-8859-1", "CP1252", "MacRoman", "MacIceland","ISO-8859-15", "",
50 "ISO-8859-2", "CP1250", "MacCentralEurope", "MacCroatian", "MacRomania", "",
51 "ISO-8859-5", "CP1251", "MacCyrillic", "MacUkraine", "KOI8-R", "KOI8-U", "KOI8-RU", "",
52 "ISO-8859-6", "CP1256", "MacArabic", "",
53 "ISO-8859-7", "CP1253", "MacGreek", "",
54 "ISO-8859-8", "CP1255", "MacHebrew", "",
55 "ISO-8859-9", "CP1254", "MacTurkish", "",
56 "ISO-8859-13", "CP1257", "",
57 "ISO-2022-JP", "ISO-2022-JP-1", "ISO-2022-JP-2", "EUC-JP", "SHIFT_JIS", "",
58 "ISO-2022-CN", "ISO-2022-CN-EXT", "EUC-CN", "EUC-TW", "BIG5", "BIG5-HKSCS", "",
59 "ISO-2022-KR", "EUC-KR", "",
60 "MacThai", "KOI8-T", "",
61 "ISO-8859-3", "ISO-8859-4", "ISO-8859-10", "ISO-8859-14", "ISO-8859-16", "",
62 "CP850", "CP862", "CP866", "CP874", "CP932", "CP949", "CP950", "CP1133", "CP1258", "",
64 "UTF-7", "UTF-16", "UTF-16BE", "UTF-16LE", "UTF-32", "UTF-32BE", "UTF-32LE",
65 "C99", "JAVA", "UCS-2", "UCS-2BE", "UCS-2LE", "UCS-4", "UCS-4BE", "UCS-4LE", "",
66 "HZ", "GBK", "GB18030", "JOHAB", "ARMSCII-8",
67 "Georgian-Academy", "Georgian-PS", "TIS-620", "MuleLao-1", "VISCII", "TCVN",
68 "HPROMAN8", "NEXTSTEP" };
70 SSA supports charset selection.
71 The following known charsets are used:
73 0 = Ansi - Western European
78 128 = Japanese (Shift JIS)
81 134 = GB2312 Simplified Chinese
82 136 = Big5 Traditional Chinese
89 204 = Russian (Cyrillic)
91 238 = Eastern European
95 static int pi_justification[] = { 0, 1, 2 };
96 static const char *ppsz_justification_text[] = {N_("Center"),N_("Left"),N_("Right")};
98 #define ENCODING_TEXT N_("Subtitles text encoding")
99 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
100 #define ALIGN_TEXT N_("Subtitles justification")
101 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
102 #define AUTODETECT_UTF8_TEXT N_("UTF-8 subtitles autodetection")
103 #define AUTODETECT_UTF8_LONGTEXT N_("This enables automatic detection of " \
104 "UTF-8 encoding within subtitles files.")
105 #define FORMAT_TEXT N_("Formatted Subtitles")
106 #define FORMAT_LONGTEXT N_("Some subtitle formats allow for text formatting. " \
107 "VLC partly implements this, but you can choose to disable all formatting.")
111 set_shortname( _("Subtitles"));
112 set_description( _("Text subtitles decoder") );
113 set_capability( "decoder", 50 );
114 set_callbacks( OpenDecoder, CloseDecoder );
115 set_category( CAT_INPUT );
116 set_subcategory( SUBCAT_INPUT_SCODEC );
118 add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
120 change_integer_list( pi_justification, ppsz_justification_text, 0 );
121 add_string( "subsdec-encoding", DEFAULT_NAME, NULL,
122 ENCODING_TEXT, ENCODING_LONGTEXT, VLC_FALSE );
123 change_string_list( ppsz_encodings, 0, 0 );
124 add_bool( "subsdec-autodetect-utf8", VLC_TRUE, NULL,
125 AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, VLC_FALSE );
126 add_bool( "subsdec-formatted", VLC_TRUE, NULL, FORMAT_TEXT, FORMAT_LONGTEXT,
130 /*****************************************************************************
131 * OpenDecoder: probe the decoder and return score
132 *****************************************************************************
133 * Tries to launch a decoder and return score so that the interface is able
135 *****************************************************************************/
136 static int OpenDecoder( vlc_object_t *p_this )
138 decoder_t *p_dec = (decoder_t*)p_this;
139 decoder_sys_t *p_sys;
142 if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','u','b','t') &&
143 p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
148 p_dec->pf_decode_sub = DecodeBlock;
150 /* Allocate the memory needed to store the decoder's structure */
151 if( ( p_dec->p_sys = p_sys =
152 (decoder_sys_t *)calloc(1, sizeof(decoder_sys_t)) ) == NULL )
154 msg_Err( p_dec, "out of memory" );
160 p_sys->iconv_handle = (vlc_iconv_t)-1;
161 p_sys->b_autodetect_utf8 = VLC_FALSE;
162 p_sys->b_ass = VLC_FALSE;
163 p_sys->i_original_height = -1;
164 p_sys->i_original_width = -1;
165 TAB_INIT( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
166 TAB_INIT( p_sys->i_images, p_sys->pp_images );
168 char *psz_charset = NULL;
169 /* First try demux-specified encoding */
170 if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
172 psz_charset = strdup (p_dec->fmt_in.subs.psz_encoding);
173 msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
174 p_dec->fmt_in.subs.psz_encoding ?: "not specified");
177 /* Second, try configured encoding */
178 if (psz_charset == NULL)
180 psz_charset = var_CreateGetNonEmptyString (p_dec, "subsdec-encoding");
181 if ((psz_charset != NULL) && !strcasecmp (psz_charset, DEFAULT_NAME))
187 msg_Dbg (p_dec, "trying configured character encoding: %s",
188 psz_charset ?: "not specified");
191 /* Third, try "local" encoding with optional UTF-8 autodetection */
192 if (psz_charset == NULL)
194 psz_charset = strdup (GetFallbackEncoding ());
195 msg_Dbg (p_dec, "trying default character encoding: %s",
196 psz_charset ?: "not specified");
198 if (var_CreateGetBool (p_dec, "subsdec-autodetect-utf8"))
200 msg_Dbg (p_dec, "using automatic UTF-8 detection");
201 p_sys->b_autodetect_utf8 = VLC_TRUE;
205 if (psz_charset == NULL)
207 psz_charset = strdup ("UTF-8");
208 msg_Dbg (p_dec, "trying hard-coded character encoding: %s",
209 psz_charset ?: "error");
212 if (psz_charset == NULL)
218 if (strcasecmp (psz_charset, "UTF-8") && strcasecmp (psz_charset, "utf8"))
220 p_sys->iconv_handle = vlc_iconv_open ("UTF-8", psz_charset);
221 if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
222 msg_Err (p_dec, "cannot convert from %s: %s", psz_charset,
227 var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
228 var_Get( p_dec, "subsdec-align", &val );
229 p_sys->i_align = val.i_int;
231 if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ')
232 && var_CreateGetBool( p_dec, "subsdec-formatted" ) )
234 if( p_dec->fmt_in.i_extra > 0 )
235 ParseSSAHeader( p_dec );
241 /****************************************************************************
242 * DecodeBlock: the whole thing
243 ****************************************************************************
244 * This function must be fed with complete subtitles units.
245 ****************************************************************************/
246 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
248 subpicture_t *p_spu = NULL;
250 if( !pp_block || *pp_block == NULL ) return NULL;
252 p_spu = ParseText( p_dec, *pp_block );
254 block_Release( *pp_block );
260 /*****************************************************************************
261 * CloseDecoder: clean up the decoder
262 *****************************************************************************/
263 static void CloseDecoder( vlc_object_t *p_this )
265 decoder_t *p_dec = (decoder_t *)p_this;
266 decoder_sys_t *p_sys = p_dec->p_sys;
268 if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
269 vlc_iconv_close( p_sys->iconv_handle );
271 if( p_sys->pp_ssa_styles )
274 for( i = 0; i < p_sys->i_ssa_styles; i++ )
276 if( !p_sys->pp_ssa_styles[i] )
279 if( p_sys->pp_ssa_styles[i]->psz_stylename )
280 free( p_sys->pp_ssa_styles[i]->psz_stylename );
281 if( p_sys->pp_ssa_styles[i]->font_style.psz_fontname )
282 free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
283 if( p_sys->pp_ssa_styles[i] )
284 free( p_sys->pp_ssa_styles[i] );
286 TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
288 if( p_sys->pp_images )
291 for( i = 0; i < p_sys->i_images; i++ )
293 if( !p_sys->pp_images[i] )
296 if( p_sys->pp_images[i]->p_pic )
297 p_sys->pp_images[i]->p_pic->pf_release( p_sys->pp_images[i]->p_pic );
298 if( p_sys->pp_images[i]->psz_filename )
299 free( p_sys->pp_images[i]->psz_filename );
301 free( p_sys->pp_images[i] );
303 TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
309 /*****************************************************************************
310 * ParseText: parse an text subtitle packet and send it to the video output
311 *****************************************************************************/
312 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
314 decoder_sys_t *p_sys = p_dec->p_sys;
315 subpicture_t *p_spu = NULL;
316 char *psz_subtitle = NULL;
319 /* We cannot display a subpicture with no date */
320 if( p_block->i_pts == 0 )
322 msg_Warn( p_dec, "subtitle without a date" );
326 /* Check validity of packet data */
327 /* An "empty" line containing only \0 can be used to force
328 and ephemer picture from the screen */
329 if( p_block->i_buffer < 1 )
331 msg_Warn( p_dec, "no subtitle data" );
335 /* Should be resiliant against bad subtitles */
336 psz_subtitle = strndup( (const char *)p_block->p_buffer,
338 if( psz_subtitle == NULL )
341 if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
343 if (EnsureUTF8( psz_subtitle ) == NULL)
345 msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
346 "Try manually setting a character-encoding "
347 "before you open the file.") );
353 if( p_sys->b_autodetect_utf8 )
355 if( IsUTF8( psz_subtitle ) == NULL )
357 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
358 "disabling UTF-8 subtitles autodetection" );
359 p_sys->b_autodetect_utf8 = VLC_FALSE;
363 if( !p_sys->b_autodetect_utf8 )
365 size_t inbytes_left = strlen( psz_subtitle );
366 size_t outbytes_left = 6 * inbytes_left;
367 char *psz_new_subtitle = malloc( outbytes_left + 1 );
368 char *psz_convert_buffer_out = psz_new_subtitle;
369 const char *psz_convert_buffer_in = psz_subtitle;
371 size_t ret = vlc_iconv( p_sys->iconv_handle,
372 &psz_convert_buffer_in, &inbytes_left,
373 &psz_convert_buffer_out, &outbytes_left );
375 *psz_convert_buffer_out++ = '\0';
376 free( psz_subtitle );
378 if( ( ret == (size_t)(-1) ) || inbytes_left )
380 free( psz_new_subtitle );
381 msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
382 "Try manually setting a character-encoding "
383 "before you open the file.") );
387 psz_subtitle = realloc( psz_new_subtitle,
388 psz_convert_buffer_out - psz_new_subtitle );
392 /* Create the subpicture unit */
393 p_spu = p_dec->pf_spu_buffer_new( p_dec );
396 msg_Warn( p_dec, "can't get spu buffer" );
397 if( psz_subtitle ) free( psz_subtitle );
401 p_spu->b_pausable = VLC_TRUE;
403 /* Create a new subpicture region */
404 memset( &fmt, 0, sizeof(video_format_t) );
405 fmt.i_chroma = VLC_FOURCC('T','E','X','T');
407 fmt.i_width = fmt.i_height = 0;
408 fmt.i_x_offset = fmt.i_y_offset = 0;
409 p_spu->p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
410 if( !p_spu->p_region )
412 msg_Err( p_dec, "cannot allocate SPU region" );
413 if( psz_subtitle ) free( psz_subtitle );
414 p_dec->pf_spu_buffer_del( p_dec, p_spu );
418 /* Decode and format the subpicture unit */
419 if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
421 /* Normal text subs, easy markup */
422 p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
423 p_spu->i_x = p_sys->i_align ? 20 : 0;
426 /* Remove formatting from string */
428 p_spu->p_region->psz_text = StripTags( psz_subtitle );
429 if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
431 p_spu->p_region->psz_html = CreateHtmlSubtitle( psz_subtitle );
434 p_spu->i_start = p_block->i_pts;
435 p_spu->i_stop = p_block->i_pts + p_block->i_length;
436 p_spu->b_ephemer = (p_block->i_length == 0);
437 p_spu->b_absolute = VLC_FALSE;
441 /* Decode SSA/USF strings */
442 if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
443 ParseSSAString( p_dec, psz_subtitle, p_spu );
445 p_spu->i_start = p_block->i_pts;
446 p_spu->i_stop = p_block->i_pts + p_block->i_length;
447 p_spu->b_ephemer = (p_block->i_length == 0);
448 p_spu->b_absolute = VLC_FALSE;
449 p_spu->i_original_picture_width = p_sys->i_original_width;
450 p_spu->i_original_picture_height = p_sys->i_original_height;
452 if( psz_subtitle ) free( psz_subtitle );
457 char* GotoNextLine( char *psz_text )
459 char *p_newline = psz_text;
461 while( p_newline[0] != '\0' )
463 if( p_newline[0] == '\n' || p_newline[0] == '\r' )
466 while( p_newline[0] == '\n' || p_newline[0] == '\r' )
475 /* Function now handles tags which has attribute values, and tries
476 * to deal with &' commands too. It no longer modifies the string
477 * in place, so that the original text can be reused
479 static char *StripTags( char *psz_subtitle )
481 char *psz_text_start;
484 psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
485 if( !psz_text_start )
488 while( *psz_subtitle )
490 if( *psz_subtitle == '<' )
492 if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
495 psz_subtitle += strcspn( psz_subtitle, ">" );
497 else if( *psz_subtitle == '&' )
499 if( !strncasecmp( psz_subtitle, "<", 4 ))
502 psz_subtitle += strcspn( psz_subtitle, ";" );
504 else if( !strncasecmp( psz_subtitle, ">", 4 ))
507 psz_subtitle += strcspn( psz_subtitle, ";" );
509 else if( !strncasecmp( psz_subtitle, "&", 5 ))
512 psz_subtitle += strcspn( psz_subtitle, ";" );
514 else if( !strncasecmp( psz_subtitle, """, 6 ))
517 psz_subtitle += strcspn( psz_subtitle, ";" );
521 /* Assume it is just a normal ampersand */
527 *psz_text++ = *psz_subtitle;
533 psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
535 return psz_text_start;
538 /* Try to respect any style tags present in the subtitle string. The main
539 * problem here is a lack of adequate specs for the subtitle formats.
540 * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
541 * SAMI has a detailed spec, but extensive rework is needed in the demux
542 * code to prevent all this style information being excised, as it presently
544 * That leaves the others - none of which were (I guess) originally intended
545 * to be carrying style information. Over time people have used them that way.
546 * In the absence of specifications from which to work, the tags supported
547 * have been restricted to the simple set permitted by the USF DTD, ie. :
548 * Basic: <br>, <i>, <b>, <u>
560 * There is also the further restriction that the subtitle be well-formed
561 * as an XML entity, ie. the HTML sentence:
562 * <b><i>Bold and Italics</b></i>
563 * doesn't qualify because the tags aren't nested one inside the other.
564 * <text> tags are automatically added to the output to ensure
566 * If the text doesn't qualify for any reason, a NULL string is
567 * returned, and the rendering engine will fall back to the
568 * plain text version of the subtitle.
570 static char *CreateHtmlSubtitle( char *psz_subtitle )
572 char psz_tagStack[ 100 ];
573 size_t i_buf_size = strlen( psz_subtitle ) + 100;
574 char *psz_html_start = malloc( i_buf_size );
576 psz_tagStack[ 0 ] = '\0';
578 if( psz_html_start != NULL )
580 char *psz_html = psz_html_start;
582 strcpy( psz_html, "<text>" );
585 while( *psz_subtitle )
587 if( *psz_subtitle == '\n' )
589 strcpy( psz_html, "<br/>" );
593 else if( *psz_subtitle == '<' )
595 if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
597 strcpy( psz_html, "<br/>" );
601 else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
603 strcpy( psz_html, "<b>" );
604 strcat( psz_tagStack, "b" );
608 else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
610 strcpy( psz_html, "<i>" );
611 strcat( psz_tagStack, "i" );
615 else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
617 strcpy( psz_html, "<u>" );
618 strcat( psz_tagStack, "u" );
622 else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
624 const char *psz_attribs[] = { "face=\"", "family=\"", "size=\"",
625 "color=\"", "outline-color=\"", "shadow-color=\"",
626 "outline-level=\"", "shadow-level=\"", "back-color=\"",
629 strcpy( psz_html, "<font " );
630 strcat( psz_tagStack, "f" );
634 while( *psz_subtitle != '>' )
638 for( k=0; psz_attribs[ k ]; k++ )
640 int i_len = strlen( psz_attribs[ k ] );
642 if( !strncasecmp( psz_subtitle, psz_attribs[ k ], i_len ))
644 i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
646 strncpy( psz_html, psz_subtitle, i_len );
648 psz_subtitle += i_len;
652 if( psz_attribs[ k ] == NULL )
654 /* Jump over unrecognised tag */
655 int i_len = strcspn( psz_subtitle, "\"" ) + 1;
657 i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
658 psz_subtitle += i_len;
660 while (*psz_subtitle == ' ')
661 *psz_html++ = *psz_subtitle++;
663 *psz_html++ = *psz_subtitle++;
665 else if( !strncmp( psz_subtitle, "</", 2 ))
667 vlc_bool_t b_match = VLC_FALSE;
668 int i_len = strlen( psz_tagStack ) - 1;
669 char *psz_lastTag = NULL;
673 psz_lastTag = psz_tagStack + i_len;
676 switch( *psz_lastTag )
679 b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
683 b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
687 b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
691 b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
698 /* Not well formed -- kill everything */
699 free( psz_html_start );
700 psz_html_start = NULL;
704 strncpy( psz_html, psz_subtitle, i_len );
706 psz_subtitle += i_len;
710 psz_subtitle += strcspn( psz_subtitle, ">" );
713 else if( *psz_subtitle == '&' )
715 if( !strncasecmp( psz_subtitle, "<", 4 ))
717 strcpy( psz_html, "<" );
721 else if( !strncasecmp( psz_subtitle, ">", 4 ))
723 strcpy( psz_html, ">" );
727 else if( !strncasecmp( psz_subtitle, "&", 5 ))
729 strcpy( psz_html, "&" );
735 strcpy( psz_html, "&" );
742 *psz_html = *psz_subtitle;
743 if( psz_html > psz_html_start )
745 /* Check for double whitespace */
746 if((( *psz_html == ' ' ) ||
747 ( *psz_html == '\t' )) &&
748 (( *(psz_html-1) == ' ' ) ||
749 ( *(psz_html-1) == '\t' )))
751 strcpy( psz_html, NO_BREAKING_SPACE );
752 psz_html += strlen( NO_BREAKING_SPACE ) - 1;
759 if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 10 )
761 int i_len = psz_html - psz_html_start;
764 psz_html_start = realloc( psz_html_start, i_buf_size );
765 psz_html = psz_html_start + i_len;
769 strcpy( psz_html, "</text>" );
772 if( psz_tagStack[ 0 ] != '\0' )
774 /* Not well formed -- kill everything */
775 free( psz_html_start );
776 psz_html_start = NULL;
778 else if( psz_html_start )
780 /* Shrink the memory requirements */
781 psz_html_start = realloc( psz_html_start, psz_html - psz_html_start + 1 );
784 return psz_html_start;