]> git.sesse.net Git - vlc/blob - modules/codec/subtitles/subsusf.c
Use picture helpers.
[vlc] / modules / codec / subtitles / subsusf.c
1 /*****************************************************************************
2  * subsusf.c : USF subtitles decoder
3  *****************************************************************************
4  * Copyright (C) 2000-2006 the VideoLAN team
5  * $Id$
6  *
7  * Authors: Bernie Purcell <bitmap@videolan.org>
8  *
9  * This program is free software; you can redistribute it and/or modify
10  * it under the terms of the GNU General Public License as published by
11  * the Free Software Foundation; either version 2 of the License, or
12  * (at your option) any later version.
13  *
14  * This program is distributed in the hope that it will be useful,
15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
17  * GNU General Public License for more details.
18  *
19  * You should have received a copy of the GNU General Public License
20  * along with this program; if not, write to the Free Software
21  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
22  *****************************************************************************/
23 #ifdef HAVE_CONFIG_H
24 # include "config.h"
25 #endif
26
27 #include "subsdec.h"
28 #include <vlc_plugin.h>
29
30 /*****************************************************************************
31  * Local prototypes
32  *****************************************************************************/
33 static int  OpenDecoder   ( vlc_object_t * );
34 static void CloseDecoder  ( vlc_object_t * );
35
36 static subpicture_t *DecodeBlock   ( decoder_t *, block_t ** );
37 static char         *CreatePlainText( char * );
38 static int           ParseImageAttachments( decoder_t *p_dec );
39
40 static subpicture_t        *ParseText     ( decoder_t *, block_t * );
41 static void                 ParseUSFHeader( decoder_t * );
42 static subpicture_region_t *ParseUSFString( decoder_t *, char *, subpicture_t * );
43 static subpicture_region_t *LoadEmbeddedImage( decoder_t *p_dec, subpicture_t *p_spu, const char *psz_filename, int i_transparent_color );
44
45 /*****************************************************************************
46  * Module descriptor.
47  *****************************************************************************/
48
49 vlc_module_begin();
50     set_capability( "decoder", 40 );
51     set_shortname( N_("USFSubs"));
52     set_description( N_("USF subtitles decoder") );
53     set_callbacks( OpenDecoder, CloseDecoder );
54     set_category( CAT_INPUT );
55     set_subcategory( SUBCAT_INPUT_SCODEC );
56     /* We inherit subsdec-align and subsdec-formatted from subsdec.c */
57 vlc_module_end();
58
59 /*****************************************************************************
60  * OpenDecoder: probe the decoder and return score
61  *****************************************************************************
62  * Tries to launch a decoder and return score so that the interface is able
63  * to chose.
64  *****************************************************************************/
65 static int OpenDecoder( vlc_object_t *p_this )
66 {
67     decoder_t     *p_dec = (decoder_t*)p_this;
68     decoder_sys_t *p_sys;
69     vlc_value_t    val;
70
71     if( p_dec->fmt_in.i_codec != VLC_FOURCC('u','s','f',' ') )
72     {
73         return VLC_EGENERIC;
74     }
75
76     p_dec->pf_decode_sub = DecodeBlock;
77
78     /* Allocate the memory needed to store the decoder's structure */
79     if( ( p_dec->p_sys = p_sys =
80           (decoder_sys_t *)calloc(1, sizeof(decoder_sys_t)) ) == NULL )
81     {
82         return VLC_ENOMEM;
83     }
84
85     /* Unused fields of p_sys - not needed for USF decoding */
86     p_sys->b_ass = false;
87     p_sys->iconv_handle = (vlc_iconv_t)-1;
88     p_sys->b_autodetect_utf8 = false;
89
90     /* init of p_sys */
91     p_sys->i_align = 0;
92     p_sys->i_original_height = -1;
93     p_sys->i_original_width = -1;
94     TAB_INIT( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
95     TAB_INIT( p_sys->i_images, p_sys->pp_images );
96
97     /* USF subtitles are mandated to be UTF-8, so don't need vlc_iconv */
98
99     var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
100     var_Get( p_dec, "subsdec-align", &val );
101     p_sys->i_align = val.i_int;
102
103     ParseImageAttachments( p_dec );
104
105     if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
106     {
107         if( p_dec->fmt_in.i_extra > 0 )
108             ParseUSFHeader( p_dec );
109     }
110
111     return VLC_SUCCESS;
112 }
113
114 /****************************************************************************
115  * DecodeBlock: the whole thing
116  ****************************************************************************
117  * This function must be fed with complete subtitles units.
118  ****************************************************************************/
119 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
120 {
121     subpicture_t *p_spu = NULL;
122
123     if( !pp_block || *pp_block == NULL ) return NULL;
124
125     p_spu = ParseText( p_dec, *pp_block );
126
127     block_Release( *pp_block );
128     *pp_block = NULL;
129
130     return p_spu;
131 }
132
133 /*****************************************************************************
134  * CloseDecoder: clean up the decoder
135  *****************************************************************************/
136 static void CloseDecoder( vlc_object_t *p_this )
137 {
138     decoder_t *p_dec = (decoder_t *)p_this;
139     decoder_sys_t *p_sys = p_dec->p_sys;
140
141     if( p_sys->pp_ssa_styles )
142     {
143         int i;
144         for( i = 0; i < p_sys->i_ssa_styles; i++ )
145         {
146             if( !p_sys->pp_ssa_styles[i] )
147                 continue;
148
149             free( p_sys->pp_ssa_styles[i]->psz_stylename );
150             free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
151             free( p_sys->pp_ssa_styles[i] );
152         }
153         TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
154     }
155     if( p_sys->pp_images )
156     {
157         int i;
158         for( i = 0; i < p_sys->i_images; i++ )
159         {
160             if( !p_sys->pp_images[i] )
161                 continue;
162
163             if( p_sys->pp_images[i]->p_pic )
164                 picture_Release( p_sys->pp_images[i] );
165             free( p_sys->pp_images[i]->psz_filename );
166
167             free( p_sys->pp_images[i] );
168         }
169         TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
170     }
171
172     free( p_sys );
173 }
174
175 /*****************************************************************************
176  * ParseText: parse an text subtitle packet and send it to the video output
177  *****************************************************************************/
178 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
179 {
180     decoder_sys_t *p_sys = p_dec->p_sys;
181     subpicture_t *p_spu = NULL;
182     char *psz_subtitle = NULL;
183
184     /* We cannot display a subpicture with no date */
185     if( p_block->i_pts == 0 )
186     {
187         msg_Warn( p_dec, "subtitle without a date" );
188         return NULL;
189     }
190
191     /* Check validity of packet data */
192     /* An "empty" line containing only \0 can be used to force
193        and ephemer picture from the screen */
194     if( p_block->i_buffer < 1 )
195     {
196         msg_Warn( p_dec, "no subtitle data" );
197         return NULL;
198     }
199
200     /* Should be resiliant against bad subtitles */
201     psz_subtitle = strndup( (const char *)p_block->p_buffer,
202                             p_block->i_buffer );
203     if( psz_subtitle == NULL )
204         return NULL;
205
206     /* USF Subtitles are mandated to be UTF-8 -- make sure it is */
207     if (EnsureUTF8( psz_subtitle ) == NULL)
208     {
209         msg_Err( p_dec, _("USF subtitles must be in UTF-8 format.\n"
210                  "This stream contains USF subtitles which aren't.") );
211     }
212
213     /* Create the subpicture unit */
214     p_spu = p_dec->pf_spu_buffer_new( p_dec );
215     if( !p_spu )
216     {
217         msg_Warn( p_dec, "can't get spu buffer" );
218         free( psz_subtitle );
219         return NULL;
220     }
221
222     p_spu->b_pausable = true;
223
224     /* Decode USF strings */
225     p_spu->p_region = ParseUSFString( p_dec, psz_subtitle, p_spu );
226
227     p_spu->i_start = p_block->i_pts;
228     p_spu->i_stop = p_block->i_pts + p_block->i_length;
229     p_spu->b_ephemer = (p_block->i_length == 0);
230     p_spu->b_absolute = false;
231     p_spu->i_original_picture_width = p_sys->i_original_width;
232     p_spu->i_original_picture_height = p_sys->i_original_height;
233
234     free( psz_subtitle );
235
236     return p_spu;
237 }
238
239 static char *GrabAttributeValue( const char *psz_attribute,
240                                  const char *psz_tag_start )
241 {
242     if( psz_attribute && psz_tag_start )
243     {
244         char *psz_tag_end = strchr( psz_tag_start, '>' );
245         char *psz_found   = strcasestr( psz_tag_start, psz_attribute );
246
247         if( psz_found )
248         {
249             psz_found += strlen( psz_attribute );
250
251             if(( *(psz_found++) == '=' ) &&
252                ( *(psz_found++) == '\"' ))
253             {
254                 if( psz_found < psz_tag_end )
255                 {
256                     int   i_len = strcspn( psz_found, "\"" );
257                     return strndup( psz_found, i_len );
258                 }
259             }
260         }
261     }
262     return NULL;
263 }
264
265 static ssa_style_t *ParseStyle( decoder_sys_t *p_sys, char *psz_subtitle )
266 {
267     ssa_style_t *p_style   = NULL;
268     char        *psz_style = GrabAttributeValue( "style", psz_subtitle );
269
270     if( psz_style )
271     {
272         int i;
273
274         for( i = 0; i < p_sys->i_ssa_styles; i++ )
275         {
276             if( !strcmp( p_sys->pp_ssa_styles[i]->psz_stylename, psz_style ) )
277                 p_style = p_sys->pp_ssa_styles[i];
278         }
279         free( psz_style );
280     }
281     return p_style;
282 }
283
284 static int ParsePositionAttributeList( char *psz_subtitle, int *i_align,
285                                        int *i_x, int *i_y )
286 {
287     int   i_mask = 0;
288
289     char *psz_align    = GrabAttributeValue( "alignment", psz_subtitle );
290     char *psz_margin_x = GrabAttributeValue( "horizontal-margin", psz_subtitle );
291     char *psz_margin_y = GrabAttributeValue( "vertical-margin", psz_subtitle );
292     /* -- UNSUPPORTED
293     char *psz_relative = GrabAttributeValue( "relative-to", psz_subtitle );
294     char *psz_rotate_x = GrabAttributeValue( "rotate-x", psz_subtitle );
295     char *psz_rotate_y = GrabAttributeValue( "rotate-y", psz_subtitle );
296     char *psz_rotate_z = GrabAttributeValue( "rotate-z", psz_subtitle );
297     */
298
299     *i_align = SUBPICTURE_ALIGN_BOTTOM;
300     *i_x = 0;
301     *i_y = 0;
302
303     if( psz_align )
304     {
305         if( !strcasecmp( "TopLeft", psz_align ) )
306             *i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_LEFT;
307         else if( !strcasecmp( "TopCenter", psz_align ) )
308             *i_align = SUBPICTURE_ALIGN_TOP;
309         else if( !strcasecmp( "TopRight", psz_align ) )
310             *i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_RIGHT;
311         else if( !strcasecmp( "MiddleLeft", psz_align ) )
312             *i_align = SUBPICTURE_ALIGN_LEFT;
313         else if( !strcasecmp( "MiddleCenter", psz_align ) )
314             *i_align = 0;
315         else if( !strcasecmp( "MiddleRight", psz_align ) )
316             *i_align = SUBPICTURE_ALIGN_RIGHT;
317         else if( !strcasecmp( "BottomLeft", psz_align ) )
318             *i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_LEFT;
319         else if( !strcasecmp( "BottomCenter", psz_align ) )
320             *i_align = SUBPICTURE_ALIGN_BOTTOM;
321         else if( !strcasecmp( "BottomRight", psz_align ) )
322             *i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_RIGHT;
323
324         i_mask |= ATTRIBUTE_ALIGNMENT;
325         free( psz_align );
326     }
327     if( psz_margin_x )
328     {
329         *i_x = atoi( psz_margin_x );
330         if( strchr( psz_margin_x, '%' ) )
331             i_mask |= ATTRIBUTE_X_PERCENT;
332         else
333             i_mask |= ATTRIBUTE_X;
334
335         free( psz_margin_x );
336     }
337     if( psz_margin_y )
338     {
339         *i_y = atoi( psz_margin_y );
340         if( strchr( psz_margin_y, '%' ) )
341             i_mask |= ATTRIBUTE_Y_PERCENT;
342         else
343             i_mask |= ATTRIBUTE_Y;
344
345         free( psz_margin_y );
346     }
347     return i_mask;
348 }
349
350 static void SetupPositions( subpicture_region_t *p_region, char *psz_subtitle )
351 {
352     int           i_mask = 0;
353     int           i_align;
354     int           i_x, i_y;
355
356     i_mask = ParsePositionAttributeList( psz_subtitle, &i_align, &i_x, &i_y );
357
358     if( i_mask & ATTRIBUTE_ALIGNMENT )
359         p_region->i_align = i_align;
360
361     /* TODO: Setup % based offsets properly, without adversely affecting
362      *       everything else in vlc. Will address with separate patch, to
363      *       prevent this one being any more complicated.
364      */
365     if( i_mask & ATTRIBUTE_X )
366         p_region->i_x = i_x;
367     else if( i_mask & ATTRIBUTE_X_PERCENT )
368         p_region->i_x = 0;
369
370     if( i_mask & ATTRIBUTE_Y )
371         p_region->i_y = i_y;
372     else if( i_mask & ATTRIBUTE_Y_PERCENT )
373         p_region->i_y = 0;
374 }
375
376 static subpicture_region_t *CreateTextRegion( decoder_t *p_dec,
377                                               subpicture_t *p_spu,
378                                               char *psz_subtitle,
379                                               int i_len,
380                                               int i_sys_align )
381 {
382     decoder_sys_t        *p_sys = p_dec->p_sys;
383     subpicture_region_t  *p_text_region;
384     video_format_t        fmt;
385
386     /* Create a new subpicture region */
387     memset( &fmt, 0, sizeof(video_format_t) );
388     fmt.i_chroma = VLC_FOURCC('T','E','X','T');
389     fmt.i_aspect = 0;
390     fmt.i_width = fmt.i_height = 0;
391     fmt.i_x_offset = fmt.i_y_offset = 0;
392     p_text_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
393
394     if( p_text_region != NULL )
395     {
396         ssa_style_t  *p_style = NULL;
397
398         p_text_region->psz_text = NULL;
399         p_text_region->psz_html = strndup( psz_subtitle, i_len );
400         if( ! p_text_region->psz_html )
401         {
402             p_spu->pf_destroy_region( VLC_OBJECT(p_dec), p_text_region );
403             return NULL;
404         }
405
406         p_style = ParseStyle( p_sys, p_text_region->psz_html );
407         if( !p_style )
408         {
409             int i;
410
411             for( i = 0; i < p_sys->i_ssa_styles; i++ )
412             {
413                 if( !strcasecmp( p_sys->pp_ssa_styles[i]->psz_stylename, "Default" ) )
414                     p_style = p_sys->pp_ssa_styles[i];
415             }
416         }
417
418         if( p_style )
419         {
420             msg_Dbg( p_dec, "style is: %s", p_style->psz_stylename );
421
422             p_text_region->p_style = &p_style->font_style;
423             p_text_region->i_align = p_style->i_align;
424
425             /* TODO: Setup % based offsets properly, without adversely affecting
426              *       everything else in vlc. Will address with separate patch,
427              *       to prevent this one being any more complicated.
428
429                      * p_style->i_margin_percent_h;
430                      * p_style->i_margin_percent_v;
431              */
432             p_text_region->i_x         = p_style->i_margin_h;
433             p_text_region->i_y         = p_style->i_margin_v;
434
435         }
436         else
437         {
438             p_text_region->i_align = SUBPICTURE_ALIGN_BOTTOM | i_sys_align;
439             p_text_region->i_x = i_sys_align ? 20 : 0;
440             p_text_region->i_y = 10;
441         }
442         /* Look for position arguments which may override the style-based
443          * defaults.
444          */
445         SetupPositions( p_text_region, psz_subtitle );
446
447         p_text_region->p_next = NULL;
448     }
449     return p_text_region;
450 }
451
452 static int ParseImageAttachments( decoder_t *p_dec )
453 {
454     decoder_sys_t        *p_sys = p_dec->p_sys;
455     input_attachment_t  **pp_attachments;
456     int                   i_attachments_cnt;
457     int                   k = 0;
458
459     if( VLC_SUCCESS != decoder_GetInputAttachments( p_dec, &pp_attachments, &i_attachments_cnt ))
460         return VLC_EGENERIC;
461
462     for( k = 0; k < i_attachments_cnt; k++ )
463     {
464         input_attachment_t *p_attach = pp_attachments[k];
465
466         vlc_fourcc_t  type  = 0;
467
468         if( ( !strcmp( p_attach->psz_mime, "image/bmp" ) )      || /* BMP */
469             ( !strcmp( p_attach->psz_mime, "image/x-bmp" ) )    ||
470             ( !strcmp( p_attach->psz_mime, "image/x-bitmap" ) ) ||
471             ( !strcmp( p_attach->psz_mime, "image/x-ms-bmp" ) ) )
472         {
473              type = VLC_FOURCC('b','m','p',' ');
474         }
475         else if( ( !strcmp( p_attach->psz_mime, "image/x-portable-anymap" ) )  || /* PNM */
476                  ( !strcmp( p_attach->psz_mime, "image/x-portable-bitmap" ) )  || /* PBM */
477                  ( !strcmp( p_attach->psz_mime, "image/x-portable-graymap" ) ) || /* PGM */
478                  ( !strcmp( p_attach->psz_mime, "image/x-portable-pixmap" ) ) )   /* PPM */
479         {
480             type = VLC_FOURCC('p','n','m',' ');
481         }
482         else if ( !strcmp( p_attach->psz_mime, "image/gif" ) )         /* GIF */
483             type = VLC_FOURCC('g','i','f',' ');
484         else if ( !strcmp( p_attach->psz_mime, "image/jpeg" ) )        /* JPG, JPEG */
485             type = VLC_FOURCC('j','p','e','g');
486         else if ( !strcmp( p_attach->psz_mime, "image/pcx" ) )         /* PCX */
487             type = VLC_FOURCC('p','c','x',' ');
488         else if ( !strcmp( p_attach->psz_mime, "image/png" ) )         /* PNG */
489             type = VLC_FOURCC('p','n','g',' ');
490         else if ( !strcmp( p_attach->psz_mime, "image/tiff" ) )        /* TIF, TIFF */
491             type = VLC_FOURCC('t','i','f','f');
492         else if ( !strcmp( p_attach->psz_mime, "image/x-tga" ) )       /* TGA */
493             type = VLC_FOURCC('t','g','a',' ');
494         else if ( !strcmp( p_attach->psz_mime, "image/x-xpixmap") )    /* XPM */
495             type = VLC_FOURCC('x','p','m',' ');
496
497         if( ( type != 0 ) &&
498             ( p_attach->i_data > 0 ) &&
499             ( p_attach->p_data != NULL ) )
500         {
501             picture_t         *p_pic = NULL;
502             image_handler_t   *p_image;
503
504             p_image = image_HandlerCreate( p_dec );
505             if( p_image != NULL )
506             {
507                 block_t   *p_block;
508
509                 p_block = block_New( p_image->p_parent, p_attach->i_data );
510
511                 if( p_block != NULL )
512                 {
513                     video_format_t     fmt_in;
514                     video_format_t     fmt_out;
515
516                     memcpy( p_block->p_buffer, p_attach->p_data, p_attach->i_data );
517
518                     memset( &fmt_in,  0, sizeof( video_format_t));
519                     memset( &fmt_out, 0, sizeof( video_format_t));
520
521                     fmt_in.i_chroma  = type;
522                     fmt_out.i_chroma = VLC_FOURCC('Y','U','V','A');
523
524                     /* Find a suitable decoder module */
525                     if( module_Exists( p_dec, "sdl_image" ) )
526                     {
527                         /* ffmpeg thinks it can handle bmp properly but it can't (at least
528                          * not all of them), so use sdl_image if it is available */
529
530                         vlc_value_t val;
531
532                         var_Create( p_dec, "codec", VLC_VAR_MODULE | VLC_VAR_DOINHERIT );
533                         val.psz_string = (char*) "sdl_image";
534                         var_Set( p_dec, "codec", val );
535                     }
536
537                     p_pic = image_Read( p_image, p_block, &fmt_in, &fmt_out );
538                     var_Destroy( p_dec, "codec" );
539                 }
540
541                 image_HandlerDelete( p_image );
542             }
543             if( p_pic )
544             {
545                 image_attach_t *p_picture = malloc( sizeof(image_attach_t) );
546
547                 if( p_picture )
548                 {
549                     p_picture->psz_filename = strdup( p_attach->psz_name );
550                     p_picture->p_pic = p_pic;
551
552                     TAB_APPEND( p_sys->i_images, p_sys->pp_images, p_picture );
553                 }
554             }
555         }
556         vlc_input_attachment_Delete( pp_attachments[ k ] );
557     }
558     free( pp_attachments );
559
560     return VLC_SUCCESS;
561 }
562
563 static void ParseUSFHeaderTags( decoder_t *p_dec, xml_reader_t *p_xml_reader )
564 {
565     decoder_sys_t *p_sys = p_dec->p_sys;
566     char *psz_node;
567     ssa_style_t *p_style = NULL;
568     int i_style_level = 0;
569     int i_metadata_level = 0;
570
571     while ( xml_ReaderRead( p_xml_reader ) == 1 )
572     {
573         switch ( xml_ReaderNodeType( p_xml_reader ) )
574         {
575             case XML_READER_TEXT:
576             case XML_READER_NONE:
577                 break;
578             case XML_READER_ENDELEM:
579                 psz_node = xml_ReaderName( p_xml_reader );
580
581                 if( !psz_node )
582                     break;
583                 switch (i_style_level)
584                 {
585                     case 0:
586                         if( !strcasecmp( "metadata", psz_node ) && (i_metadata_level == 1) )
587                         {
588                             i_metadata_level--;
589                         }
590                         break;
591                     case 1:
592                         if( !strcasecmp( "styles", psz_node ) )
593                         {
594                             i_style_level--;
595                         }
596                         break;
597                     case 2:
598                         if( !strcasecmp( "style", psz_node ) )
599                         {
600                             TAB_APPEND( p_sys->i_ssa_styles, p_sys->pp_ssa_styles, p_style );
601
602                             p_style = NULL;
603                             i_style_level--;
604                         }
605                         break;
606                 }
607
608                 free( psz_node );
609                 break;
610             case XML_READER_STARTELEM:
611                 psz_node = xml_ReaderName( p_xml_reader );
612
613                 if( !psz_node )
614                     break;
615
616                 if( !strcasecmp( "metadata", psz_node ) && (i_style_level == 0) )
617                 {
618                     i_metadata_level++;
619                 }
620                 else if( !strcasecmp( "resolution", psz_node ) &&
621                          ( i_metadata_level == 1) )
622                 {
623                     while ( xml_ReaderNextAttr( p_xml_reader ) == VLC_SUCCESS )
624                     {
625                         char *psz_name = xml_ReaderName ( p_xml_reader );
626                         char *psz_value = xml_ReaderValue ( p_xml_reader );
627
628                         if( psz_name && psz_value )
629                         {
630                             if( !strcasecmp( "x", psz_name ) )
631                                 p_sys->i_original_width = atoi( psz_value );
632                             else if( !strcasecmp( "y", psz_name ) )
633                                 p_sys->i_original_height = atoi( psz_value );
634                         }
635                         free( psz_name );
636                         free( psz_value );
637                     }
638                 }
639                 else if( !strcasecmp( "styles", psz_node ) && (i_style_level == 0) )
640                 {
641                     i_style_level++;
642                 }
643                 else if( !strcasecmp( "style", psz_node ) && (i_style_level == 1) )
644                 {
645                     i_style_level++;
646
647                     p_style = calloc( 1, sizeof(ssa_style_t) );
648                     if( ! p_style )
649                     {
650                         free( psz_node );
651                         break;
652                     }
653                     /* All styles are supposed to default to Default, and then
654                      * one or more settings are over-ridden.
655                      * At the moment this only effects styles defined AFTER
656                      * Default in the XML
657                      */
658                     int i;
659                     for( i = 0; i < p_sys->i_ssa_styles; i++ )
660                     {
661                         if( !strcasecmp( p_sys->pp_ssa_styles[i]->psz_stylename, "Default" ) )
662                         {
663                             ssa_style_t *p_default_style = p_sys->pp_ssa_styles[i];
664
665                             memcpy( p_style, p_default_style, sizeof( ssa_style_t ) );
666                             p_style->font_style.psz_fontname = strdup( p_style->font_style.psz_fontname );
667                             p_style->psz_stylename = NULL;
668                         }
669                     }
670
671                     while ( xml_ReaderNextAttr( p_xml_reader ) == VLC_SUCCESS )
672                     {
673                         char *psz_name = xml_ReaderName ( p_xml_reader );
674                         char *psz_value = xml_ReaderValue ( p_xml_reader );
675
676                         if( psz_name && psz_value )
677                         {
678                             if( !strcasecmp( "name", psz_name ) )
679                                 p_style->psz_stylename = strdup( psz_value);
680                         }
681                         free( psz_name );
682                         free( psz_value );
683                     }
684                 }
685                 else if( !strcasecmp( "fontstyle", psz_node ) && (i_style_level == 2) )
686                 {
687                     while ( xml_ReaderNextAttr( p_xml_reader ) == VLC_SUCCESS )
688                     {
689                         char *psz_name = xml_ReaderName ( p_xml_reader );
690                         char *psz_value = xml_ReaderValue ( p_xml_reader );
691
692                         if( psz_name && psz_value )
693                         {
694                             if( !strcasecmp( "face", psz_name ) )
695                             {
696                                 free( p_style->font_style.psz_fontname );
697                                 p_style->font_style.psz_fontname = strdup( psz_value );
698                             }
699                             else if( !strcasecmp( "size", psz_name ) )
700                             {
701                                 if( ( *psz_value == '+' ) || ( *psz_value == '-' ) )
702                                 {
703                                     int i_value = atoi( psz_value );
704
705                                     if( ( i_value >= -5 ) && ( i_value <= 5 ) )
706                                         p_style->font_style.i_font_size  +=
707                                             ( i_value * p_style->font_style.i_font_size ) / 10;
708                                     else if( i_value < -5 )
709                                         p_style->font_style.i_font_size  = - i_value;
710                                     else if( i_value > 5 )
711                                         p_style->font_style.i_font_size  = i_value;
712                                 }
713                                 else
714                                     p_style->font_style.i_font_size  = atoi( psz_value );
715                             }
716                             else if( !strcasecmp( "italic", psz_name ) )
717                             {
718                                 if( !strcasecmp( "yes", psz_value ))
719                                     p_style->font_style.i_style_flags |= STYLE_ITALIC;
720                                 else
721                                     p_style->font_style.i_style_flags &= ~STYLE_ITALIC;
722                             }
723                             else if( !strcasecmp( "weight", psz_name ) )
724                             {
725                                 if( !strcasecmp( "bold", psz_value ))
726                                     p_style->font_style.i_style_flags |= STYLE_BOLD;
727                                 else
728                                     p_style->font_style.i_style_flags &= ~STYLE_BOLD;
729                             }
730                             else if( !strcasecmp( "underline", psz_name ) )
731                             {
732                                 if( !strcasecmp( "yes", psz_value ))
733                                     p_style->font_style.i_style_flags |= STYLE_UNDERLINE;
734                                 else
735                                     p_style->font_style.i_style_flags &= ~STYLE_UNDERLINE;
736                             }
737                             else if( !strcasecmp( "color", psz_name ) )
738                             {
739                                 if( *psz_value == '#' )
740                                 {
741                                     unsigned long col = strtol(psz_value+1, NULL, 16);
742                                     p_style->font_style.i_font_color = (col & 0x00ffffff);
743                                     p_style->font_style.i_font_alpha = (col >> 24) & 0xff;
744                                 }
745                             }
746                             else if( !strcasecmp( "outline-color", psz_name ) )
747                             {
748                                 if( *psz_value == '#' )
749                                 {
750                                     unsigned long col = strtol(psz_value+1, NULL, 16);
751                                     p_style->font_style.i_outline_color = (col & 0x00ffffff);
752                                     p_style->font_style.i_outline_alpha = (col >> 24) & 0xff;
753                                 }
754                             }
755                             else if( !strcasecmp( "outline-level", psz_name ) )
756                             {
757                                 p_style->font_style.i_outline_width = atoi( psz_value );
758                             }
759                             else if( !strcasecmp( "shadow-color", psz_name ) )
760                             {
761                                 if( *psz_value == '#' )
762                                 {
763                                     unsigned long col = strtol(psz_value+1, NULL, 16);
764                                     p_style->font_style.i_shadow_color = (col & 0x00ffffff);
765                                     p_style->font_style.i_shadow_alpha = (col >> 24) & 0xff;
766                                 }
767                             }
768                             else if( !strcasecmp( "shadow-level", psz_name ) )
769                             {
770                                 p_style->font_style.i_shadow_width = atoi( psz_value );
771                             }
772                             else if( !strcasecmp( "back-color", psz_name ) )
773                             {
774                                 if( *psz_value == '#' )
775                                 {
776                                     unsigned long col = strtol(psz_value+1, NULL, 16);
777                                     p_style->font_style.i_karaoke_background_color = (col & 0x00ffffff);
778                                     p_style->font_style.i_karaoke_background_alpha = (col >> 24) & 0xff;
779                                 }
780                             }
781                             else if( !strcasecmp( "spacing", psz_name ) )
782                             {
783                                 p_style->font_style.i_spacing = atoi( psz_value );
784                             }
785                         }
786                         free( psz_name );
787                         free( psz_value );
788                     }
789                 }
790                 else if( !strcasecmp( "position", psz_node ) && (i_style_level == 2) )
791                 {
792                     while ( xml_ReaderNextAttr( p_xml_reader ) == VLC_SUCCESS )
793                     {
794                         char *psz_name = xml_ReaderName ( p_xml_reader );
795                         char *psz_value = xml_ReaderValue ( p_xml_reader );
796
797                         if( psz_name && psz_value )
798                         {
799                             if( !strcasecmp( "alignment", psz_name ) )
800                             {
801                                 if( !strcasecmp( "TopLeft", psz_value ) )
802                                     p_style->i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_LEFT;
803                                 else if( !strcasecmp( "TopCenter", psz_value ) )
804                                     p_style->i_align = SUBPICTURE_ALIGN_TOP;
805                                 else if( !strcasecmp( "TopRight", psz_value ) )
806                                     p_style->i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_RIGHT;
807                                 else if( !strcasecmp( "MiddleLeft", psz_value ) )
808                                     p_style->i_align = SUBPICTURE_ALIGN_LEFT;
809                                 else if( !strcasecmp( "MiddleCenter", psz_value ) )
810                                     p_style->i_align = 0;
811                                 else if( !strcasecmp( "MiddleRight", psz_value ) )
812                                     p_style->i_align = SUBPICTURE_ALIGN_RIGHT;
813                                 else if( !strcasecmp( "BottomLeft", psz_value ) )
814                                     p_style->i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_LEFT;
815                                 else if( !strcasecmp( "BottomCenter", psz_value ) )
816                                     p_style->i_align = SUBPICTURE_ALIGN_BOTTOM;
817                                 else if( !strcasecmp( "BottomRight", psz_value ) )
818                                     p_style->i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_RIGHT;
819                             }
820                             else if( !strcasecmp( "horizontal-margin", psz_name ) )
821                             {
822                                 if( strchr( psz_value, '%' ) )
823                                 {
824                                     p_style->i_margin_h = 0;
825                                     p_style->i_margin_percent_h = atoi( psz_value );
826                                 }
827                                 else
828                                 {
829                                     p_style->i_margin_h = atoi( psz_value );
830                                     p_style->i_margin_percent_h = 0;
831                                 }
832                             }
833                             else if( !strcasecmp( "vertical-margin", psz_name ) )
834                             {
835                                 if( strchr( psz_value, '%' ) )
836                                 {
837                                     p_style->i_margin_v = 0;
838                                     p_style->i_margin_percent_v = atoi( psz_value );
839                                 }
840                                 else
841                                 {
842                                     p_style->i_margin_v = atoi( psz_value );
843                                     p_style->i_margin_percent_v = 0;
844                                 }
845                             }
846                         }
847                         free( psz_name );
848                         free( psz_value );
849                     }
850                 }
851
852                 free( psz_node );
853                 break;
854         }
855     }
856     free( p_style );
857 }
858
859
860
861 static subpicture_region_t *ParseUSFString( decoder_t *p_dec,
862                                             char *psz_subtitle,
863                                             subpicture_t *p_spu_in )
864 {
865     decoder_sys_t        *p_sys = p_dec->p_sys;
866     subpicture_t         *p_spu = p_spu_in;
867     subpicture_region_t  *p_region_first = NULL;
868     subpicture_region_t  *p_region_upto  = p_region_first;
869
870     while( *psz_subtitle )
871     {
872         if( *psz_subtitle == '<' )
873         {
874             char *psz_end = NULL;
875
876             if(( !strncasecmp( psz_subtitle, "<text ", 6 )) ||
877                ( !strncasecmp( psz_subtitle, "<text>", 6 )))
878             {
879                 psz_end = strcasestr( psz_subtitle, "</text>" );
880
881                 if( psz_end )
882                 {
883                     subpicture_region_t  *p_text_region;
884
885                     psz_end += strcspn( psz_end, ">" ) + 1;
886
887                     p_text_region = CreateTextRegion( p_dec,
888                                                       p_spu,
889                                                       psz_subtitle,
890                                                       psz_end - psz_subtitle,
891                                                       p_sys->i_align );
892
893                     if( p_text_region )
894                     {
895                         p_text_region->psz_text = CreatePlainText( p_text_region->psz_html );
896
897                         if( ! var_CreateGetBool( p_dec, "subsdec-formatted" ) )
898                         {
899                             free( p_text_region->psz_html );
900                             p_text_region->psz_html = NULL;
901                         }
902                     }
903
904                     if( !p_region_first )
905                     {
906                         p_region_first = p_region_upto = p_text_region;
907                     }
908                     else if( p_text_region )
909                     {
910                         p_region_upto->p_next = p_text_region;
911                         p_region_upto = p_region_upto->p_next;
912                     }
913                 }
914             }
915             else if(( !strncasecmp( psz_subtitle, "<karaoke ", 9 )) ||
916                     ( !strncasecmp( psz_subtitle, "<karaoke>", 9 )))
917             {
918                 psz_end = strcasestr( psz_subtitle, "</karaoke>" );
919
920                 if( psz_end )
921                 {
922                     subpicture_region_t  *p_text_region;
923
924                     psz_end += strcspn( psz_end, ">" ) + 1;
925
926                     p_text_region = CreateTextRegion( p_dec,
927                                                       p_spu,
928                                                       psz_subtitle,
929                                                       psz_end - psz_subtitle,
930                                                       p_sys->i_align );
931
932                     if( p_text_region )
933                     {
934                         if( ! var_CreateGetBool( p_dec, "subsdec-formatted" ) )
935                         {
936                             free( p_text_region->psz_html );
937                             p_text_region->psz_html = NULL;
938                         }
939                     }
940                     if( !p_region_first )
941                     {
942                         p_region_first = p_region_upto = p_text_region;
943                     }
944                     else if( p_text_region )
945                     {
946                         p_region_upto->p_next = p_text_region;
947                         p_region_upto = p_region_upto->p_next;
948                     }
949                 }
950             }
951             else if(( !strncasecmp( psz_subtitle, "<image ", 7 )) ||
952                     ( !strncasecmp( psz_subtitle, "<image>", 7 )))
953             {
954                 subpicture_region_t *p_image_region = NULL;
955
956                 char *psz_end = strcasestr( psz_subtitle, "</image>" );
957                 char *psz_content = strchr( psz_subtitle, '>' );
958                 int   i_transparent = -1;
959
960                 /* If a colorkey parameter is specified, then we have to map
961                  * that index in the picture through as transparent (it is
962                  * required by the USF spec but is also recommended that if the
963                  * creator really wants a transparent colour that they use a
964                  * type like PNG that properly supports it; this goes doubly
965                  * for VLC because the pictures are stored internally in YUV
966                  * and the resulting colour-matching may not produce the
967                  * desired results.)
968                  */
969                 char *psz_tmp = GrabAttributeValue( "colorkey", psz_subtitle );
970                 if( psz_tmp )
971                 {
972                     if( *psz_tmp == '#' )
973                         i_transparent = strtol( psz_tmp + 1, NULL, 16 ) & 0x00ffffff;
974                     free( psz_tmp );
975                 }
976                 if( psz_content && ( psz_content < psz_end ) )
977                 {
978                     char *psz_filename = strndup( &psz_content[1], psz_end - &psz_content[1] );
979                     if( psz_filename )
980                     {
981                         p_image_region = LoadEmbeddedImage( p_dec, p_spu,
982                                             psz_filename, i_transparent );
983                         free( psz_filename );
984                     }
985                 }
986
987                 if( psz_end ) psz_end += strcspn( psz_end, ">" ) + 1;
988
989                 if( p_image_region )
990                 {
991                     SetupPositions( p_image_region, psz_subtitle );
992
993                     p_image_region->p_next   = NULL;
994                     p_image_region->psz_text = NULL;
995                     p_image_region->psz_html = NULL;
996
997                 }
998                 if( !p_region_first )
999                 {
1000                     p_region_first = p_region_upto = p_image_region;
1001                 }
1002                 else if( p_image_region )
1003                 {
1004                     p_region_upto->p_next = p_image_region;
1005                     p_region_upto = p_region_upto->p_next;
1006                 }
1007             }
1008             if( psz_end )
1009                 psz_subtitle = psz_end - 1;
1010
1011             psz_subtitle += strcspn( psz_subtitle, ">" );
1012         }
1013
1014         psz_subtitle++;
1015     }
1016
1017     return p_region_first;
1018 }
1019
1020 /*****************************************************************************
1021  * ParseUSFHeader: Retrieve global formatting information etc
1022  *****************************************************************************/
1023 static void ParseUSFHeader( decoder_t *p_dec )
1024 {
1025     stream_t      *p_sub = NULL;
1026     xml_t         *p_xml = NULL;
1027     xml_reader_t  *p_xml_reader = NULL;
1028
1029     p_sub = stream_MemoryNew( VLC_OBJECT(p_dec),
1030                               p_dec->fmt_in.p_extra,
1031                               p_dec->fmt_in.i_extra,
1032                               true );
1033     if( !p_sub )
1034         return;
1035
1036     p_xml = xml_Create( p_dec );
1037     if( p_xml )
1038     {
1039         p_xml_reader = xml_ReaderCreate( p_xml, p_sub );
1040         if( p_xml_reader )
1041         {
1042             /* Look for Root Node */
1043             if( xml_ReaderRead( p_xml_reader ) == 1 )
1044             {
1045                 char *psz_node = xml_ReaderName( p_xml_reader );
1046
1047                 if( !strcasecmp( "usfsubtitles", psz_node ) )
1048                     ParseUSFHeaderTags( p_dec, p_xml_reader );
1049
1050                 free( psz_node );
1051             }
1052
1053             xml_ReaderDelete( p_xml, p_xml_reader );
1054         }
1055         xml_Delete( p_xml );
1056     }
1057     stream_Delete( p_sub );
1058 }
1059
1060 /* Function now handles tags which has attribute values, and tries
1061  * to deal with &' commands too. It no longer modifies the string
1062  * in place, so that the original text can be reused
1063  */
1064 static char *StripTags( char *psz_subtitle )
1065 {
1066     char *psz_text_start;
1067     char *psz_text;
1068
1069     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
1070     if( !psz_text_start )
1071         return NULL;
1072
1073     while( *psz_subtitle )
1074     {
1075         /* Mask out any pre-existing LFs in the subtitle */
1076         if( *psz_subtitle == '\n' )
1077             *psz_subtitle = ' ';
1078
1079         if( *psz_subtitle == '<' )
1080         {
1081             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
1082                 *psz_text++ = '\n';
1083
1084             psz_subtitle += strcspn( psz_subtitle, ">" );
1085         }
1086         else if( *psz_subtitle == '&' )
1087         {
1088             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
1089             {
1090                 *psz_text++ = '<';
1091                 psz_subtitle += strcspn( psz_subtitle, ";" );
1092             }
1093             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
1094             {
1095                 *psz_text++ = '>';
1096                 psz_subtitle += strcspn( psz_subtitle, ";" );
1097             }
1098             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
1099             {
1100                 *psz_text++ = '&';
1101                 psz_subtitle += strcspn( psz_subtitle, ";" );
1102             }
1103             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
1104             {
1105                 *psz_text++ = '\"';
1106                 psz_subtitle += strcspn( psz_subtitle, ";" );
1107             }
1108             else
1109             {
1110                 /* Assume it is just a normal ampersand */
1111                 *psz_text++ = '&';
1112             }
1113         }
1114         else
1115         {
1116             *psz_text++ = *psz_subtitle;
1117         }
1118
1119         psz_subtitle++;
1120     }
1121     *psz_text = '\0';
1122     psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
1123
1124     return psz_text_start;
1125 }
1126
1127 /* Turn a HTML subtitle, turn into a plain-text version,
1128  *  complete with sensible whitespace compaction
1129  */
1130
1131 static char *CreatePlainText( char *psz_subtitle )
1132 {
1133     char *psz_text = StripTags( psz_subtitle );
1134     char *s;
1135
1136     if( !psz_text )
1137         return NULL;
1138
1139     s = strpbrk( psz_text, "\t\r\n " );
1140     while( s )
1141     {
1142         int   k;
1143         char  spc = ' ';
1144         int   i_whitespace = strspn( s, "\t\r\n " );
1145
1146         /* Favour '\n' over other whitespaces - if one of these
1147          * occurs in the whitespace use a '\n' as our value,
1148          * otherwise just use a ' '
1149          */
1150         for( k = 0; k < i_whitespace; k++ )
1151             if( s[k] == '\n' ) spc = '\n';
1152
1153         if( i_whitespace > 1 )
1154         {
1155             memmove( &s[1],
1156                      &s[i_whitespace],
1157                      strlen( s ) - i_whitespace + 1 );
1158         }
1159         *s++ = spc;
1160
1161         s = strpbrk( s, "\t\r\n " );
1162     }
1163     return psz_text;
1164 }
1165
1166 /****************************************************************************
1167  * download and resize image located at psz_url
1168  ***************************************************************************/
1169 static subpicture_region_t *LoadEmbeddedImage( decoder_t *p_dec,
1170                                                subpicture_t *p_spu,
1171                                                const char *psz_filename,
1172                                                int i_transparent_color )
1173 {
1174     decoder_sys_t         *p_sys = p_dec->p_sys;
1175     subpicture_region_t   *p_region;
1176     video_format_t         fmt_out;
1177     int                    k;
1178     picture_t             *p_pic = NULL;
1179
1180     for( k = 0; k < p_sys->i_images; k++ )
1181     {
1182         if( p_sys->pp_images &&
1183             !strcmp( p_sys->pp_images[k]->psz_filename, psz_filename ) )
1184         {
1185             p_pic = p_sys->pp_images[k]->p_pic;
1186             break;
1187         }
1188     }
1189
1190     if( !p_pic )
1191     {
1192         msg_Err( p_dec, "Unable to read image %s", psz_filename );
1193         return NULL;
1194     }
1195
1196     /* Display the feed's image */
1197     memset( &fmt_out, 0, sizeof( video_format_t));
1198
1199     fmt_out.i_chroma = VLC_FOURCC('Y','U','V','A');
1200     fmt_out.i_aspect = VOUT_ASPECT_FACTOR;
1201     fmt_out.i_sar_num = fmt_out.i_sar_den = 1;
1202     fmt_out.i_width =
1203         fmt_out.i_visible_width = p_pic->p[Y_PLANE].i_visible_pitch;
1204     fmt_out.i_height =
1205         fmt_out.i_visible_height = p_pic->p[Y_PLANE].i_visible_lines;
1206
1207     p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt_out );
1208     if( !p_region )
1209     {
1210         msg_Err( p_dec, "cannot allocate SPU region" );
1211         return NULL;
1212     }
1213     vout_CopyPicture( p_dec, &p_region->picture, p_pic );
1214
1215     /* This isn't the best way to do this - if you really want transparency, then
1216      * you're much better off using an image type that supports it like PNG. The
1217      * spec requires this support though.
1218      */
1219     if( i_transparent_color > 0 )
1220     {
1221         uint8_t i_r = ( i_transparent_color >> 16 ) & 0xff;
1222         uint8_t i_g = ( i_transparent_color >>  8 ) & 0xff;
1223         uint8_t i_b = ( i_transparent_color       ) & 0xff;
1224         uint8_t i_y = ( ( (  66 * i_r + 129 * i_g +  25 * i_b + 128 ) >> 8 ) + 16 );
1225         uint8_t i_u =   ( ( -38 * i_r -  74 * i_g + 112 * i_b + 128 ) >> 8 ) + 128 ;
1226         uint8_t i_v =   ( ( 112 * i_r -  94 * i_g -  18 * i_b + 128 ) >> 8 ) + 128 ;
1227
1228         if( ( p_region->picture.Y_PITCH == p_region->picture.U_PITCH ) &&
1229             ( p_region->picture.Y_PITCH == p_region->picture.V_PITCH ) &&
1230             ( p_region->picture.Y_PITCH == p_region->picture.A_PITCH ) )
1231         {
1232             int i_lines = p_region->picture.p[ Y_PLANE ].i_lines;
1233             if( i_lines > p_region->picture.p[ U_PLANE ].i_lines )
1234                 i_lines = p_region->picture.p[ U_PLANE ].i_lines;
1235             if( i_lines > p_region->picture.p[ V_PLANE ].i_lines )
1236                 i_lines = p_region->picture.p[ V_PLANE ].i_lines;
1237             if( i_lines > p_region->picture.p[ A_PLANE ].i_lines )
1238                 i_lines = p_region->picture.p[ A_PLANE ].i_lines;
1239
1240             int   i;
1241
1242             for( i = 0; i < p_region->picture.A_PITCH * i_lines; i++ )
1243             {
1244                 if(( p_region->picture.Y_PIXELS[ i ] == i_y ) &&
1245                    ( p_region->picture.U_PIXELS[ i ] == i_u ) &&
1246                    ( p_region->picture.V_PIXELS[ i ] == i_v ) )
1247                 {
1248                     p_region->picture.A_PIXELS[ i ] = 1;
1249                 }
1250             }
1251         }
1252     }
1253     return p_region;
1254 }