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 *****************************************************************************/
36 /*****************************************************************************
38 *****************************************************************************/
39 static int OpenDecoder ( vlc_object_t * );
40 static void CloseDecoder ( vlc_object_t * );
42 static subpicture_t *DecodeBlock ( decoder_t *, block_t ** );
43 static subpicture_t *ParseText ( decoder_t *, block_t * );
44 static char *StripTags ( char * );
45 static char *CreateHtmlSubtitle ( char * );
48 /*****************************************************************************
50 *****************************************************************************/
51 static const char *ppsz_encodings[] = { DEFAULT_NAME, "ASCII", "UTF-8", "",
52 "ISO-8859-1", "CP1252", "MacRoman", "MacIceland","ISO-8859-15", "",
53 "ISO-8859-2", "CP1250", "MacCentralEurope", "MacCroatian", "MacRomania", "",
54 "ISO-8859-5", "CP1251", "MacCyrillic", "MacUkraine", "KOI8-R", "KOI8-U", "KOI8-RU", "",
55 "ISO-8859-6", "CP1256", "MacArabic", "",
56 "ISO-8859-7", "CP1253", "MacGreek", "",
57 "ISO-8859-8", "CP1255", "MacHebrew", "",
58 "ISO-8859-9", "CP1254", "MacTurkish", "",
59 "ISO-8859-13", "CP1257", "",
60 "ISO-2022-JP", "ISO-2022-JP-1", "ISO-2022-JP-2", "EUC-JP", "SHIFT_JIS", "",
61 "ISO-2022-CN", "ISO-2022-CN-EXT", "EUC-CN", "EUC-TW", "BIG5", "BIG5-HKSCS", "",
62 "ISO-2022-KR", "EUC-KR", "",
63 "MacThai", "KOI8-T", "",
64 "ISO-8859-3", "ISO-8859-4", "ISO-8859-10", "ISO-8859-14", "ISO-8859-16", "",
65 "CP850", "CP862", "CP866", "CP874", "CP932", "CP949", "CP950", "CP1133", "CP1258", "",
67 "UTF-7", "UTF-16", "UTF-16BE", "UTF-16LE", "UTF-32", "UTF-32BE", "UTF-32LE",
68 "C99", "JAVA", "UCS-2", "UCS-2BE", "UCS-2LE", "UCS-4", "UCS-4BE", "UCS-4LE", "",
69 "HZ", "GBK", "GB18030", "JOHAB", "ARMSCII-8",
70 "Georgian-Academy", "Georgian-PS", "TIS-620", "MuleLao-1", "VISCII", "TCVN",
71 "HPROMAN8", "NEXTSTEP" };
73 SSA supports charset selection.
74 The following known charsets are used:
76 0 = Ansi - Western European
81 128 = Japanese (Shift JIS)
84 134 = GB2312 Simplified Chinese
85 136 = Big5 Traditional Chinese
92 204 = Russian (Cyrillic)
94 238 = Eastern European
98 static int pi_justification[] = { 0, 1, 2 };
99 static const char *ppsz_justification_text[] = {N_("Center"),N_("Left"),N_("Right")};
101 #define ENCODING_TEXT N_("Subtitles text encoding")
102 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
103 #define ALIGN_TEXT N_("Subtitles justification")
104 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
105 #define AUTODETECT_UTF8_TEXT N_("UTF-8 subtitles autodetection")
106 #define AUTODETECT_UTF8_LONGTEXT N_("This enables automatic detection of " \
107 "UTF-8 encoding within subtitles files.")
108 #define FORMAT_TEXT N_("Formatted Subtitles")
109 #define FORMAT_LONGTEXT N_("Some subtitle formats allow for text formatting. " \
110 "VLC partly implements this, but you can choose to disable all formatting.")
114 set_shortname( _("Subtitles"));
115 set_description( _("Text subtitles decoder") );
116 set_capability( "decoder", 50 );
117 set_callbacks( OpenDecoder, CloseDecoder );
118 set_category( CAT_INPUT );
119 set_subcategory( SUBCAT_INPUT_SCODEC );
121 add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
123 change_integer_list( pi_justification, ppsz_justification_text, 0 );
124 add_string( "subsdec-encoding", DEFAULT_NAME, NULL,
125 ENCODING_TEXT, ENCODING_LONGTEXT, false );
126 change_string_list( ppsz_encodings, 0, 0 );
127 add_bool( "subsdec-autodetect-utf8", true, NULL,
128 AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, false );
129 add_bool( "subsdec-formatted", true, NULL, FORMAT_TEXT, FORMAT_LONGTEXT,
133 /*****************************************************************************
134 * OpenDecoder: probe the decoder and return score
135 *****************************************************************************
136 * Tries to launch a decoder and return score so that the interface is able
138 *****************************************************************************/
139 static int OpenDecoder( vlc_object_t *p_this )
141 decoder_t *p_dec = (decoder_t*)p_this;
142 decoder_sys_t *p_sys;
145 switch( p_dec->fmt_in.i_codec )
147 case VLC_FOURCC('s','u','b','t'):
148 case VLC_FOURCC('s','s','a',' '):
149 case VLC_FOURCC('t','1','4','0'):
155 p_dec->pf_decode_sub = DecodeBlock;
157 /* Allocate the memory needed to store the decoder's structure */
158 p_dec->p_sys = p_sys = malloc( sizeof( *p_sys ) );
161 msg_Err( p_dec, "out of memory" );
166 memset( p_sys, 0, sizeof( *p_sys ) );
168 p_sys->iconv_handle = (vlc_iconv_t)-1;
169 p_sys->b_autodetect_utf8 = false;
170 p_sys->b_ass = false;
171 p_sys->i_original_height = -1;
172 p_sys->i_original_width = -1;
173 TAB_INIT( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
174 TAB_INIT( p_sys->i_images, p_sys->pp_images );
176 char *psz_charset = NULL;
178 /* First try demux-specified encoding */
179 if( p_dec->fmt_in.i_codec == VLC_FOURCC('t','1','4','0') )
180 psz_charset = strdup( "UTF-8" ); /* IUT T.140 is always using UTF-8 */
182 if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
184 psz_charset = strdup (p_dec->fmt_in.subs.psz_encoding);
185 msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
186 p_dec->fmt_in.subs.psz_encoding ?: "not specified");
189 /* Second, try configured encoding */
190 if (psz_charset == NULL)
192 psz_charset = var_CreateGetNonEmptyString (p_dec, "subsdec-encoding");
193 if ((psz_charset != NULL) && !strcasecmp (psz_charset, DEFAULT_NAME))
199 msg_Dbg (p_dec, "trying configured character encoding: %s",
200 psz_charset ?: "not specified");
203 /* Third, try "local" encoding with optional UTF-8 autodetection */
204 if (psz_charset == NULL)
206 psz_charset = strdup (GetFallbackEncoding ());
207 msg_Dbg (p_dec, "trying default character encoding: %s",
208 psz_charset ?: "not specified");
210 if (var_CreateGetBool (p_dec, "subsdec-autodetect-utf8"))
212 msg_Dbg (p_dec, "using automatic UTF-8 detection");
213 p_sys->b_autodetect_utf8 = true;
217 /* Forth, don't do character decoding, i.e. assume UTF-8 */
218 if (psz_charset == NULL)
220 psz_charset = strdup ("UTF-8");
221 msg_Dbg (p_dec, "using UTF-8 character encoding" );
224 if ((psz_charset != NULL)
225 && strcasecmp (psz_charset, "UTF-8")
226 && strcasecmp (psz_charset, "utf8"))
228 p_sys->iconv_handle = vlc_iconv_open ("UTF-8", psz_charset);
229 if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
230 msg_Err (p_dec, "cannot convert from %s: %m", psz_charset);
234 var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
235 var_Get( p_dec, "subsdec-align", &val );
236 p_sys->i_align = val.i_int;
238 if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ')
239 && var_CreateGetBool( p_dec, "subsdec-formatted" ) )
241 if( p_dec->fmt_in.i_extra > 0 )
242 ParseSSAHeader( p_dec );
248 /****************************************************************************
249 * DecodeBlock: the whole thing
250 ****************************************************************************
251 * This function must be fed with complete subtitles units.
252 ****************************************************************************/
253 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
255 subpicture_t *p_spu = NULL;
257 if( !pp_block || *pp_block == NULL ) return NULL;
259 p_spu = ParseText( p_dec, *pp_block );
261 block_Release( *pp_block );
267 /*****************************************************************************
268 * CloseDecoder: clean up the decoder
269 *****************************************************************************/
270 static void CloseDecoder( vlc_object_t *p_this )
272 decoder_t *p_dec = (decoder_t *)p_this;
273 decoder_sys_t *p_sys = p_dec->p_sys;
275 if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
276 vlc_iconv_close( p_sys->iconv_handle );
278 if( p_sys->pp_ssa_styles )
281 for( i = 0; i < p_sys->i_ssa_styles; i++ )
283 if( !p_sys->pp_ssa_styles[i] )
286 free( p_sys->pp_ssa_styles[i]->psz_stylename );
287 free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
288 free( p_sys->pp_ssa_styles[i] );
290 TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
292 if( p_sys->pp_images )
295 for( i = 0; i < p_sys->i_images; i++ )
297 if( !p_sys->pp_images[i] )
300 if( p_sys->pp_images[i]->p_pic )
301 p_sys->pp_images[i]->p_pic->pf_release( p_sys->pp_images[i]->p_pic );
302 free( p_sys->pp_images[i]->psz_filename );
304 free( p_sys->pp_images[i] );
306 TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
312 /*****************************************************************************
313 * ParseText: parse an text subtitle packet and send it to the video output
314 *****************************************************************************/
315 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
317 decoder_sys_t *p_sys = p_dec->p_sys;
318 subpicture_t *p_spu = NULL;
319 char *psz_subtitle = NULL;
322 /* We cannot display a subpicture with no date */
323 if( p_block->i_pts == 0 )
325 msg_Warn( p_dec, "subtitle without a date" );
329 /* Check validity of packet data */
330 /* An "empty" line containing only \0 can be used to force
331 and ephemer picture from the screen */
332 if( p_block->i_buffer < 1 )
334 msg_Warn( p_dec, "no subtitle data" );
338 /* Should be resiliant against bad subtitles */
339 psz_subtitle = strndup( (const char *)p_block->p_buffer,
341 if( psz_subtitle == NULL )
344 if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
346 if (EnsureUTF8( psz_subtitle ) == NULL)
348 msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
349 "Try manually setting a character-encoding "
350 "before you open the file.") );
356 if( p_sys->b_autodetect_utf8 )
358 if( IsUTF8( psz_subtitle ) == NULL )
360 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
361 "disabling UTF-8 subtitles autodetection" );
362 p_sys->b_autodetect_utf8 = false;
366 if( !p_sys->b_autodetect_utf8 )
368 size_t inbytes_left = strlen( psz_subtitle );
369 size_t outbytes_left = 6 * inbytes_left;
370 char *psz_new_subtitle = malloc( outbytes_left + 1 );
371 char *psz_convert_buffer_out = psz_new_subtitle;
372 const char *psz_convert_buffer_in = psz_subtitle;
374 size_t ret = vlc_iconv( p_sys->iconv_handle,
375 &psz_convert_buffer_in, &inbytes_left,
376 &psz_convert_buffer_out, &outbytes_left );
378 *psz_convert_buffer_out++ = '\0';
379 free( psz_subtitle );
381 if( ( ret == (size_t)(-1) ) || inbytes_left )
383 free( psz_new_subtitle );
384 msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
385 "Try manually setting a character-encoding "
386 "before you open the file.") );
390 psz_subtitle = realloc( psz_new_subtitle,
391 psz_convert_buffer_out - psz_new_subtitle );
395 /* Create the subpicture unit */
396 p_spu = p_dec->pf_spu_buffer_new( p_dec );
399 msg_Warn( p_dec, "can't get spu buffer" );
400 free( psz_subtitle );
404 p_spu->b_pausable = true;
406 /* Create a new subpicture region */
407 memset( &fmt, 0, sizeof(video_format_t) );
408 fmt.i_chroma = VLC_FOURCC('T','E','X','T');
410 fmt.i_width = fmt.i_height = 0;
411 fmt.i_x_offset = fmt.i_y_offset = 0;
412 p_spu->p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
413 if( !p_spu->p_region )
415 msg_Err( p_dec, "cannot allocate SPU region" );
416 free( psz_subtitle );
417 p_dec->pf_spu_buffer_del( p_dec, p_spu );
421 /* Decode and format the subpicture unit */
422 if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
424 /* Normal text subs, easy markup */
425 p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
426 p_spu->i_x = p_sys->i_align ? 20 : 0;
429 /* Remove formatting from string */
431 p_spu->p_region->psz_text = StripTags( psz_subtitle );
432 if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
434 p_spu->p_region->psz_html = CreateHtmlSubtitle( psz_subtitle );
437 p_spu->i_start = p_block->i_pts;
438 p_spu->i_stop = p_block->i_pts + p_block->i_length;
439 p_spu->b_ephemer = (p_block->i_length == 0);
440 p_spu->b_absolute = false;
444 /* Decode SSA/USF strings */
445 if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
446 ParseSSAString( p_dec, psz_subtitle, p_spu );
448 p_spu->i_start = p_block->i_pts;
449 p_spu->i_stop = p_block->i_pts + p_block->i_length;
450 p_spu->b_ephemer = (p_block->i_length == 0);
451 p_spu->b_absolute = false;
452 p_spu->i_original_picture_width = p_sys->i_original_width;
453 p_spu->i_original_picture_height = p_sys->i_original_height;
455 free( psz_subtitle );
460 char* GotoNextLine( char *psz_text )
462 char *p_newline = psz_text;
464 while( p_newline[0] != '\0' )
466 if( p_newline[0] == '\n' || p_newline[0] == '\r' )
469 while( p_newline[0] == '\n' || p_newline[0] == '\r' )
478 /* Function now handles tags with attribute values, and tries
479 * to deal with &' commands too. It no longer modifies the string
480 * in place, so that the original text can be reused
482 static char *StripTags( char *psz_subtitle )
484 char *psz_text_start;
487 psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
488 if( !psz_text_start )
491 while( *psz_subtitle )
493 if( *psz_subtitle == '<' )
495 if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
498 psz_subtitle += strcspn( psz_subtitle, ">" );
500 else if( *psz_subtitle == '&' )
502 if( !strncasecmp( psz_subtitle, "<", 4 ))
505 psz_subtitle += strcspn( psz_subtitle, ";" );
507 else if( !strncasecmp( psz_subtitle, ">", 4 ))
510 psz_subtitle += strcspn( psz_subtitle, ";" );
512 else if( !strncasecmp( psz_subtitle, "&", 5 ))
515 psz_subtitle += strcspn( psz_subtitle, ";" );
517 else if( !strncasecmp( psz_subtitle, """, 6 ))
520 psz_subtitle += strcspn( psz_subtitle, ";" );
524 /* Assume it is just a normal ampersand */
530 *psz_text++ = *psz_subtitle;
536 psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
538 return psz_text_start;
541 /* Try to respect any style tags present in the subtitle string. The main
542 * problem here is a lack of adequate specs for the subtitle formats.
543 * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
544 * SAMI has a detailed spec, but extensive rework is needed in the demux
545 * code to prevent all this style information being excised, as it presently
547 * That leaves the others - none of which were (I guess) originally intended
548 * to be carrying style information. Over time people have used them that way.
549 * In the absence of specifications from which to work, the tags supported
550 * have been restricted to the simple set permitted by the USF DTD, ie. :
551 * Basic: <br>, <i>, <b>, <u>
563 * There is also the further restriction that the subtitle be well-formed
564 * as an XML entity, ie. the HTML sentence:
565 * <b><i>Bold and Italics</b></i>
566 * doesn't qualify because the tags aren't nested one inside the other.
567 * <text> tags are automatically added to the output to ensure
569 * If the text doesn't qualify for any reason, a NULL string is
570 * returned, and the rendering engine will fall back to the
571 * plain text version of the subtitle.
573 static char *CreateHtmlSubtitle( char *psz_subtitle )
575 char *psz_tag = malloc( ( strlen( psz_subtitle ) / 3 ) + 1 );
576 if( !psz_tag ) return NULL;
577 size_t i_buf_size = strlen( psz_subtitle ) + 100;
578 char *psz_html_start = malloc( i_buf_size );
582 if( psz_html_start != NULL )
584 char *psz_html = psz_html_start;
586 strcpy( psz_html, "<text>" );
589 while( *psz_subtitle )
591 if( *psz_subtitle == '\n' )
593 strcpy( psz_html, "<br/>" );
597 else if( *psz_subtitle == '<' )
599 if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
601 strcpy( psz_html, "<br/>" );
605 else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
607 strcpy( psz_html, "<b>" );
608 strcat( psz_tag, "b" );
612 else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
614 strcpy( psz_html, "<i>" );
615 strcat( psz_tag, "i" );
619 else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
621 strcpy( psz_html, "<u>" );
622 strcat( psz_tag, "u" );
626 else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
628 const char *psz_attribs[] = { "face=\"", "family=\"", "size=\"",
629 "color=\"", "outline-color=\"", "shadow-color=\"",
630 "outline-level=\"", "shadow-level=\"", "back-color=\"",
633 strcpy( psz_html, "<font " );
634 strcat( psz_tag, "f" );
638 while( *psz_subtitle != '>' )
642 for( k=0; psz_attribs[ k ]; k++ )
644 int i_len = strlen( psz_attribs[ k ] );
646 if( !strncasecmp( psz_subtitle, psz_attribs[ k ], i_len ))
648 i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
650 strncpy( psz_html, psz_subtitle, i_len );
652 psz_subtitle += i_len;
656 if( psz_attribs[ k ] == NULL )
658 /* Jump over unrecognised tag */
659 int i_len = strcspn( psz_subtitle, "\"" ) + 1;
661 i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
662 psz_subtitle += i_len;
664 while (*psz_subtitle == ' ')
665 *psz_html++ = *psz_subtitle++;
667 *psz_html++ = *psz_subtitle++;
669 else if( !strncmp( psz_subtitle, "</", 2 ))
671 bool b_match = false;
672 int i_len = strlen( psz_tag ) - 1;
673 char *psz_lastTag = NULL;
677 psz_lastTag = psz_tag + i_len;
680 switch( *psz_lastTag )
683 b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
687 b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
691 b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
695 b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
702 /* Not well formed -- kill everything */
703 free( psz_html_start );
704 psz_html_start = NULL;
708 strncpy( psz_html, psz_subtitle, i_len );
710 psz_subtitle += i_len;
714 psz_subtitle += strcspn( psz_subtitle, ">" );
717 else if( *psz_subtitle == '&' )
719 if( !strncasecmp( psz_subtitle, "<", 4 ))
721 strcpy( psz_html, "<" );
725 else if( !strncasecmp( psz_subtitle, ">", 4 ))
727 strcpy( psz_html, ">" );
731 else if( !strncasecmp( psz_subtitle, "&", 5 ))
733 strcpy( psz_html, "&" );
739 strcpy( psz_html, "&" );
746 *psz_html = *psz_subtitle;
747 if( psz_html > psz_html_start )
749 /* Check for double whitespace */
750 if((( *psz_html == ' ' ) ||
751 ( *psz_html == '\t' )) &&
752 (( *(psz_html-1) == ' ' ) ||
753 ( *(psz_html-1) == '\t' )))
755 strcpy( psz_html, NO_BREAKING_SPACE );
756 psz_html += strlen( NO_BREAKING_SPACE ) - 1;
763 if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 10 )
765 int i_len = psz_html - psz_html_start;
768 psz_html_start = realloc( psz_html_start, i_buf_size );
769 psz_html = psz_html_start + i_len;
773 strcpy( psz_html, "</text>" );
776 if( psz_tag[ 0 ] != '\0' )
778 /* Not well formed -- kill everything */
779 free( psz_html_start );
780 psz_html_start = NULL;
782 else if( psz_html_start )
784 /* Shrink the memory requirements */
785 psz_html_start = realloc( psz_html_start, psz_html - psz_html_start + 1 );
789 return psz_html_start;