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 ( char * );
49 /*****************************************************************************
51 *****************************************************************************/
52 static const char *const ppsz_encodings[] = {
53 DEFAULT_NAME, "ASCII", "UTF-8", "",
54 "ISO-8859-1", "CP1252", "MacRoman", "MacIceland","ISO-8859-15", "",
55 "ISO-8859-2", "CP1250", "MacCentralEurope", "MacCroatian", "MacRomania", "",
56 "ISO-8859-5", "CP1251", "MacCyrillic", "MacUkraine", "KOI8-R", "KOI8-U", "KOI8-RU", "",
57 "ISO-8859-6", "CP1256", "MacArabic", "",
58 "ISO-8859-7", "CP1253", "MacGreek", "",
59 "ISO-8859-8", "CP1255", "MacHebrew", "",
60 "ISO-8859-9", "CP1254", "MacTurkish", "",
61 "ISO-8859-13", "CP1257", "",
62 "ISO-2022-JP", "ISO-2022-JP-1", "ISO-2022-JP-2", "EUC-JP", "SHIFT_JIS", "",
63 "ISO-2022-CN", "ISO-2022-CN-EXT", "EUC-CN", "EUC-TW", "BIG5", "BIG5-HKSCS", "",
64 "ISO-2022-KR", "EUC-KR", "",
65 "MacThai", "KOI8-T", "",
66 "ISO-8859-3", "ISO-8859-4", "ISO-8859-10", "ISO-8859-14", "ISO-8859-16", "",
67 "CP850", "CP862", "CP866", "CP874", "CP932", "CP949", "CP950", "CP1133", "CP1258", "",
69 "UTF-7", "UTF-16", "UTF-16BE", "UTF-16LE", "UTF-32", "UTF-32BE", "UTF-32LE",
70 "C99", "JAVA", "UCS-2", "UCS-2BE", "UCS-2LE", "UCS-4", "UCS-4BE", "UCS-4LE", "",
71 "HZ", "GBK", "GB18030", "JOHAB", "ARMSCII-8",
72 "Georgian-Academy", "Georgian-PS", "TIS-620", "MuleLao-1", "VISCII", "TCVN",
73 "HPROMAN8", "NEXTSTEP" };
75 SSA supports charset selection.
76 The following known charsets are used:
78 0 = Ansi - Western European
83 128 = Japanese (Shift JIS)
86 134 = GB2312 Simplified Chinese
87 136 = Big5 Traditional Chinese
94 204 = Russian (Cyrillic)
96 238 = Eastern European
100 static const int pi_justification[] = { 0, 1, 2 };
101 static const char *const ppsz_justification_text[] = {
102 N_("Center"),N_("Left"),N_("Right")};
104 #define ENCODING_TEXT N_("Subtitles text encoding")
105 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
106 #define ALIGN_TEXT N_("Subtitles justification")
107 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
108 #define AUTODETECT_UTF8_TEXT N_("UTF-8 subtitles autodetection")
109 #define AUTODETECT_UTF8_LONGTEXT N_("This enables automatic detection of " \
110 "UTF-8 encoding within subtitles files.")
111 #define FORMAT_TEXT N_("Formatted Subtitles")
112 #define FORMAT_LONGTEXT N_("Some subtitle formats allow for text formatting. " \
113 "VLC partly implements this, but you can choose to disable all formatting.")
117 set_shortname( N_("Subtitles"));
118 set_description( N_("Text subtitles decoder") );
119 set_capability( "decoder", 50 );
120 set_callbacks( OpenDecoder, CloseDecoder );
121 set_category( CAT_INPUT );
122 set_subcategory( SUBCAT_INPUT_SCODEC );
124 add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
126 change_integer_list( pi_justification, ppsz_justification_text, 0 );
127 add_string( "subsdec-encoding", DEFAULT_NAME, NULL,
128 ENCODING_TEXT, ENCODING_LONGTEXT, false );
129 change_string_list( ppsz_encodings, 0, 0 );
130 add_bool( "subsdec-autodetect-utf8", true, NULL,
131 AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, false );
132 add_bool( "subsdec-formatted", true, NULL, FORMAT_TEXT, FORMAT_LONGTEXT,
136 /*****************************************************************************
137 * OpenDecoder: probe the decoder and return score
138 *****************************************************************************
139 * Tries to launch a decoder and return score so that the interface is able
141 *****************************************************************************/
142 static int OpenDecoder( vlc_object_t *p_this )
144 decoder_t *p_dec = (decoder_t*)p_this;
145 decoder_sys_t *p_sys;
148 switch( p_dec->fmt_in.i_codec )
150 case VLC_FOURCC('s','u','b','t'):
151 case VLC_FOURCC('s','s','a',' '):
152 case VLC_FOURCC('t','1','4','0'):
158 p_dec->pf_decode_sub = DecodeBlock;
160 /* Allocate the memory needed to store the decoder's structure */
161 p_dec->p_sys = p_sys = malloc( sizeof( *p_sys ) );
164 msg_Err( p_dec, "out of memory" );
169 memset( p_sys, 0, sizeof( *p_sys ) );
171 p_sys->iconv_handle = (vlc_iconv_t)-1;
172 p_sys->b_autodetect_utf8 = false;
173 p_sys->b_ass = false;
174 p_sys->i_original_height = -1;
175 p_sys->i_original_width = -1;
176 TAB_INIT( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
177 TAB_INIT( p_sys->i_images, p_sys->pp_images );
179 char *psz_charset = NULL;
181 /* First try demux-specified encoding */
182 if( p_dec->fmt_in.i_codec == VLC_FOURCC('t','1','4','0') )
183 psz_charset = strdup( "UTF-8" ); /* IUT T.140 is always using UTF-8 */
185 if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
187 psz_charset = strdup (p_dec->fmt_in.subs.psz_encoding);
188 msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
189 p_dec->fmt_in.subs.psz_encoding ?: "not specified");
192 /* Second, try configured encoding */
193 if (psz_charset == NULL)
195 psz_charset = var_CreateGetNonEmptyString (p_dec, "subsdec-encoding");
196 if ((psz_charset != NULL) && !strcasecmp (psz_charset, DEFAULT_NAME))
202 msg_Dbg (p_dec, "trying configured character encoding: %s",
203 psz_charset ?: "not specified");
206 /* Third, try "local" encoding with optional UTF-8 autodetection */
207 if (psz_charset == NULL)
209 psz_charset = strdup (GetFallbackEncoding ());
210 msg_Dbg (p_dec, "trying default character encoding: %s",
211 psz_charset ?: "not specified");
213 if (var_CreateGetBool (p_dec, "subsdec-autodetect-utf8"))
215 msg_Dbg (p_dec, "using automatic UTF-8 detection");
216 p_sys->b_autodetect_utf8 = true;
220 /* Forth, don't do character decoding, i.e. assume UTF-8 */
221 if (psz_charset == NULL)
223 psz_charset = strdup ("UTF-8");
224 msg_Dbg (p_dec, "using UTF-8 character encoding" );
227 if ((psz_charset != NULL)
228 && strcasecmp (psz_charset, "UTF-8")
229 && strcasecmp (psz_charset, "utf8"))
231 p_sys->iconv_handle = vlc_iconv_open ("UTF-8", psz_charset);
232 if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
233 msg_Err (p_dec, "cannot convert from %s: %m", psz_charset);
237 var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
238 var_Get( p_dec, "subsdec-align", &val );
239 p_sys->i_align = val.i_int;
241 if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ')
242 && var_CreateGetBool( p_dec, "subsdec-formatted" ) )
244 if( p_dec->fmt_in.i_extra > 0 )
245 ParseSSAHeader( p_dec );
251 /****************************************************************************
252 * DecodeBlock: the whole thing
253 ****************************************************************************
254 * This function must be fed with complete subtitles units.
255 ****************************************************************************/
256 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
258 subpicture_t *p_spu = NULL;
260 if( !pp_block || *pp_block == NULL ) return NULL;
262 p_spu = ParseText( p_dec, *pp_block );
264 block_Release( *pp_block );
270 /*****************************************************************************
271 * CloseDecoder: clean up the decoder
272 *****************************************************************************/
273 static void CloseDecoder( vlc_object_t *p_this )
275 decoder_t *p_dec = (decoder_t *)p_this;
276 decoder_sys_t *p_sys = p_dec->p_sys;
278 if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
279 vlc_iconv_close( p_sys->iconv_handle );
281 if( p_sys->pp_ssa_styles )
284 for( i = 0; i < p_sys->i_ssa_styles; i++ )
286 if( !p_sys->pp_ssa_styles[i] )
289 free( p_sys->pp_ssa_styles[i]->psz_stylename );
290 free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
291 free( p_sys->pp_ssa_styles[i] );
293 TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
295 if( p_sys->pp_images )
298 for( i = 0; i < p_sys->i_images; i++ )
300 if( !p_sys->pp_images[i] )
303 if( p_sys->pp_images[i]->p_pic )
304 p_sys->pp_images[i]->p_pic->pf_release( p_sys->pp_images[i]->p_pic );
305 free( p_sys->pp_images[i]->psz_filename );
307 free( p_sys->pp_images[i] );
309 TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
315 /*****************************************************************************
316 * ParseText: parse an text subtitle packet and send it to the video output
317 *****************************************************************************/
318 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
320 decoder_sys_t *p_sys = p_dec->p_sys;
321 subpicture_t *p_spu = NULL;
322 char *psz_subtitle = NULL;
325 /* We cannot display a subpicture with no date */
326 if( p_block->i_pts == 0 )
328 msg_Warn( p_dec, "subtitle without a date" );
332 /* Check validity of packet data */
333 /* An "empty" line containing only \0 can be used to force
334 and ephemer picture from the screen */
335 if( p_block->i_buffer < 1 )
337 msg_Warn( p_dec, "no subtitle data" );
341 /* Should be resiliant against bad subtitles */
342 psz_subtitle = strndup( (const char *)p_block->p_buffer,
344 if( psz_subtitle == NULL )
347 if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
349 if (EnsureUTF8( psz_subtitle ) == NULL)
351 msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
352 "Try manually setting a character-encoding "
353 "before you open the file.") );
359 if( p_sys->b_autodetect_utf8 )
361 if( IsUTF8( psz_subtitle ) == NULL )
363 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
364 "disabling UTF-8 subtitles autodetection" );
365 p_sys->b_autodetect_utf8 = false;
369 if( !p_sys->b_autodetect_utf8 )
371 size_t inbytes_left = strlen( psz_subtitle );
372 size_t outbytes_left = 6 * inbytes_left;
373 char *psz_new_subtitle = malloc( outbytes_left + 1 );
374 char *psz_convert_buffer_out = psz_new_subtitle;
375 const char *psz_convert_buffer_in = psz_subtitle;
377 size_t ret = vlc_iconv( p_sys->iconv_handle,
378 &psz_convert_buffer_in, &inbytes_left,
379 &psz_convert_buffer_out, &outbytes_left );
381 *psz_convert_buffer_out++ = '\0';
382 free( psz_subtitle );
384 if( ( ret == (size_t)(-1) ) || inbytes_left )
386 free( psz_new_subtitle );
387 msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
388 "Try manually setting a character-encoding "
389 "before you open the file.") );
393 psz_subtitle = realloc( psz_new_subtitle,
394 psz_convert_buffer_out - psz_new_subtitle );
398 /* Create the subpicture unit */
399 p_spu = p_dec->pf_spu_buffer_new( p_dec );
402 msg_Warn( p_dec, "can't get spu buffer" );
403 free( psz_subtitle );
407 p_spu->b_pausable = true;
409 /* Create a new subpicture region */
410 memset( &fmt, 0, sizeof(video_format_t) );
411 fmt.i_chroma = VLC_FOURCC('T','E','X','T');
413 fmt.i_width = fmt.i_height = 0;
414 fmt.i_x_offset = fmt.i_y_offset = 0;
415 p_spu->p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
416 if( !p_spu->p_region )
418 msg_Err( p_dec, "cannot allocate SPU region" );
419 free( psz_subtitle );
420 p_dec->pf_spu_buffer_del( p_dec, p_spu );
424 /* Decode and format the subpicture unit */
425 if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
427 /* Normal text subs, easy markup */
428 p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
429 p_spu->i_x = p_sys->i_align ? 20 : 0;
432 /* Remove formatting from string */
434 p_spu->p_region->psz_text = StripTags( psz_subtitle );
435 if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
437 p_spu->p_region->psz_html = CreateHtmlSubtitle( psz_subtitle );
440 p_spu->i_start = p_block->i_pts;
441 p_spu->i_stop = p_block->i_pts + p_block->i_length;
442 p_spu->b_ephemer = (p_block->i_length == 0);
443 p_spu->b_absolute = false;
447 /* Decode SSA/USF strings */
448 if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
449 ParseSSAString( p_dec, psz_subtitle, p_spu );
451 p_spu->i_start = p_block->i_pts;
452 p_spu->i_stop = p_block->i_pts + p_block->i_length;
453 p_spu->b_ephemer = (p_block->i_length == 0);
454 p_spu->b_absolute = false;
455 p_spu->i_original_picture_width = p_sys->i_original_width;
456 p_spu->i_original_picture_height = p_sys->i_original_height;
458 free( psz_subtitle );
463 char* GotoNextLine( char *psz_text )
465 char *p_newline = psz_text;
467 while( p_newline[0] != '\0' )
469 if( p_newline[0] == '\n' || p_newline[0] == '\r' )
472 while( p_newline[0] == '\n' || p_newline[0] == '\r' )
481 /* Function now handles tags with attribute values, and tries
482 * to deal with &' commands too. It no longer modifies the string
483 * in place, so that the original text can be reused
485 static char *StripTags( char *psz_subtitle )
487 char *psz_text_start;
490 psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
491 if( !psz_text_start )
494 while( *psz_subtitle )
496 if( *psz_subtitle == '<' )
498 if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
501 psz_subtitle += strcspn( psz_subtitle, ">" );
503 else if( *psz_subtitle == '&' )
505 if( !strncasecmp( psz_subtitle, "<", 4 ))
508 psz_subtitle += strcspn( psz_subtitle, ";" );
510 else if( !strncasecmp( psz_subtitle, ">", 4 ))
513 psz_subtitle += strcspn( psz_subtitle, ";" );
515 else if( !strncasecmp( psz_subtitle, "&", 5 ))
518 psz_subtitle += strcspn( psz_subtitle, ";" );
520 else if( !strncasecmp( psz_subtitle, """, 6 ))
523 psz_subtitle += strcspn( psz_subtitle, ";" );
527 /* Assume it is just a normal ampersand */
533 *psz_text++ = *psz_subtitle;
539 psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
541 return psz_text_start;
544 /* Try to respect any style tags present in the subtitle string. The main
545 * problem here is a lack of adequate specs for the subtitle formats.
546 * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
547 * SAMI has a detailed spec, but extensive rework is needed in the demux
548 * code to prevent all this style information being excised, as it presently
550 * That leaves the others - none of which were (I guess) originally intended
551 * to be carrying style information. Over time people have used them that way.
552 * In the absence of specifications from which to work, the tags supported
553 * have been restricted to the simple set permitted by the USF DTD, ie. :
554 * Basic: <br>, <i>, <b>, <u>
566 * There is also the further restriction that the subtitle be well-formed
567 * as an XML entity, ie. the HTML sentence:
568 * <b><i>Bold and Italics</b></i>
569 * doesn't qualify because the tags aren't nested one inside the other.
570 * <text> tags are automatically added to the output to ensure
572 * If the text doesn't qualify for any reason, a NULL string is
573 * returned, and the rendering engine will fall back to the
574 * plain text version of the subtitle.
576 static char *CreateHtmlSubtitle( char *psz_subtitle )
578 char *psz_tag = malloc( ( strlen( psz_subtitle ) / 3 ) + 1 );
579 if( !psz_tag ) return NULL;
580 size_t i_buf_size = strlen( psz_subtitle ) + 100;
581 char *psz_html_start = malloc( i_buf_size );
585 if( psz_html_start == NULL )
591 char *psz_html = psz_html_start;
593 strcpy( psz_html, "<text>" );
596 while( *psz_subtitle )
598 if( *psz_subtitle == '\n' )
600 strcpy( psz_html, "<br/>" );
604 else if( *psz_subtitle == '<' )
606 if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
608 strcpy( psz_html, "<br/>" );
612 else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
614 strcpy( psz_html, "<b>" );
615 strcat( psz_tag, "b" );
619 else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
621 strcpy( psz_html, "<i>" );
622 strcat( psz_tag, "i" );
626 else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
628 strcpy( psz_html, "<u>" );
629 strcat( psz_tag, "u" );
633 else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
635 const char *psz_attribs[] = { "face=\"", "family=\"", "size=\"",
636 "color=\"", "outline-color=\"", "shadow-color=\"",
637 "outline-level=\"", "shadow-level=\"", "back-color=\"",
640 strcpy( psz_html, "<font " );
641 strcat( psz_tag, "f" );
645 while( *psz_subtitle != '>' )
649 for( k=0; psz_attribs[ k ]; k++ )
651 int i_len = strlen( psz_attribs[ k ] );
653 if( !strncasecmp( psz_subtitle, psz_attribs[k], i_len ))
655 i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
657 strncpy( psz_html, psz_subtitle, i_len );
659 psz_subtitle += i_len;
663 if( psz_attribs[ k ] == NULL )
665 /* Jump over unrecognised tag */
666 int i_len = strcspn( psz_subtitle, "\"" ) + 1;
668 i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
669 psz_subtitle += i_len;
671 while (*psz_subtitle == ' ')
672 *psz_html++ = *psz_subtitle++;
674 *psz_html++ = *psz_subtitle++;
676 else if( !strncmp( psz_subtitle, "</", 2 ))
678 bool b_match = false;
679 int i_len = strlen( psz_tag ) - 1;
680 char *psz_lastTag = NULL;
684 psz_lastTag = psz_tag + i_len;
687 switch( *psz_lastTag )
690 b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
694 b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
698 b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
702 b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
709 /* Not well formed -- kill everything */
710 free( psz_html_start );
711 psz_html_start = NULL;
715 strncpy( psz_html, psz_subtitle, i_len );
717 psz_subtitle += i_len;
721 psz_subtitle += strcspn( psz_subtitle, ">" );
724 else if( *psz_subtitle == '&' )
726 if( !strncasecmp( psz_subtitle, "<", 4 ))
728 strcpy( psz_html, "<" );
732 else if( !strncasecmp( psz_subtitle, ">", 4 ))
734 strcpy( psz_html, ">" );
738 else if( !strncasecmp( psz_subtitle, "&", 5 ))
740 strcpy( psz_html, "&" );
746 strcpy( psz_html, "&" );
753 *psz_html = *psz_subtitle;
754 if( psz_html > psz_html_start )
756 /* Check for double whitespace */
757 if((( *psz_html == ' ' ) ||
758 ( *psz_html == '\t' )) &&
759 (( *(psz_html-1) == ' ' ) ||
760 ( *(psz_html-1) == '\t' )))
762 strcpy( psz_html, NO_BREAKING_SPACE );
763 psz_html += strlen( NO_BREAKING_SPACE ) - 1;
770 if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 10 )
772 int i_len = psz_html - psz_html_start;
775 psz_html_start = realloc( psz_html_start, i_buf_size );
776 psz_html = psz_html_start + i_len;
780 strcpy( psz_html, "</text>" );
783 if( psz_tag[ 0 ] != '\0' )
785 /* Not well formed -- kill everything */
786 free( psz_html_start );
787 psz_html_start = NULL;
789 else if( psz_html_start )
791 /* Shrink the memory requirements */
792 psz_html_start = realloc( psz_html_start, psz_html - psz_html_start + 1 );
795 return psz_html_start;