git.sesse.net Git - vlc/blob - 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 #include <assert.h>
  30
  31 /*****************************************************************************
  32  * Local prototypes
  33  *****************************************************************************/
  34 static int  OpenDecoder   ( vlc_object_t * );
  35 static void CloseDecoder  ( vlc_object_t * );
  36
  37 static subpicture_t *DecodeBlock   ( decoder_t *, block_t ** );
  38 static char         *CreatePlainText( char * );
  39 static int           ParseImageAttachments( decoder_t *p_dec );
  40
  41 static subpicture_t        *ParseText     ( decoder_t *, block_t * );
  42 static void                 ParseUSFHeader( decoder_t * );
  43 static subpicture_region_t *ParseUSFString( decoder_t *, char *, subpicture_t * );
  44 static subpicture_region_t *LoadEmbeddedImage( decoder_t *p_dec, subpicture_t *p_spu, const char *psz_filename, int i_transparent_color );
  45
  46 /*****************************************************************************
  47  * Module descriptor.
  48  *****************************************************************************/
  49
  50 vlc_module_begin ()
  51     set_capability( "decoder", 40 )
  52     set_shortname( N_("USFSubs"))
  53     set_description( N_("USF subtitles decoder") )
  54     set_callbacks( OpenDecoder, CloseDecoder )
  55     set_category( CAT_INPUT )
  56     set_subcategory( SUBCAT_INPUT_SCODEC )
  57     /* We inherit subsdec-align and subsdec-formatted from subsdec.c */
  58 vlc_module_end ()
  59
  60 /*****************************************************************************
  61  * OpenDecoder: probe the decoder and return score
  62  *****************************************************************************
  63  * Tries to launch a decoder and return score so that the interface is able
  64  * to chose.
  65  *****************************************************************************/
  66 static int OpenDecoder( vlc_object_t *p_this )
  67 {
  68     decoder_t     *p_dec = (decoder_t*)p_this;
  69     decoder_sys_t *p_sys;
  70
  71     if( p_dec->fmt_in.i_codec != VLC_FOURCC('u','s','f',' ') )
  72         return VLC_EGENERIC;
  73
  74     /* Allocate the memory needed to store the decoder's structure */
  75     if( ( p_dec->p_sys = p_sys = calloc(1, sizeof(decoder_sys_t)) ) == NULL )
  76         return VLC_ENOMEM;
  77
  78     p_dec->pf_decode_sub = DecodeBlock;
  79     p_dec->fmt_out.i_cat = SPU_ES;
  80     p_dec->fmt_out.i_codec = 0;
  81
  82     /* Unused fields of p_sys - not needed for USF decoding */
  83     p_sys->b_ass = false;
  84     p_sys->iconv_handle = (vlc_iconv_t)-1;
  85     p_sys->b_autodetect_utf8 = false;
  86
  87     /* init of p_sys */
  88     p_sys->i_align = 0;
  89     p_sys->i_original_height = 0;
  90     p_sys->i_original_width = 0;
  91     TAB_INIT( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
  92     TAB_INIT( p_sys->i_images, p_sys->pp_images );
  93
  94     /* USF subtitles are mandated to be UTF-8, so don't need vlc_iconv */
  95
  96     p_sys->i_align = var_CreateGetInteger( p_dec, "subsdec-align" );
  97
  98     ParseImageAttachments( p_dec );
  99
 100     if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 101     {
 102         if( p_dec->fmt_in.i_extra > 0 )
 103             ParseUSFHeader( p_dec );
 104     }
 105
 106     return VLC_SUCCESS;
 107 }
 108
 109 /****************************************************************************
 110  * DecodeBlock: the whole thing
 111  ****************************************************************************
 112  * This function must be fed with complete subtitles units.
 113  ****************************************************************************/
 114 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 115 {
 116     subpicture_t *p_spu;
 117     block_t *p_block;
 118
 119     if( !pp_block || *pp_block == NULL )
 120         return NULL;
 121
 122     p_block = *pp_block;
 123
 124     p_spu = ParseText( p_dec, p_block );
 125
 126     block_Release( p_block );
 127     *pp_block = NULL;
 128
 129     return p_spu;
 130 }
 131
 132 /*****************************************************************************
 133  * CloseDecoder: clean up the decoder
 134  *****************************************************************************/
 135 static void CloseDecoder( vlc_object_t *p_this )
 136 {
 137     decoder_t *p_dec = (decoder_t *)p_this;
 138     decoder_sys_t *p_sys = p_dec->p_sys;
 139
 140     if( p_sys->pp_ssa_styles )
 141     {
 142         int i;
 143         for( i = 0; i < p_sys->i_ssa_styles; i++ )
 144         {
 145             if( !p_sys->pp_ssa_styles[i] )
 146                 continue;
 147
 148             free( p_sys->pp_ssa_styles[i]->psz_stylename );
 149             free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
 150             free( p_sys->pp_ssa_styles[i] );
 151         }
 152         TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 153     }
 154     if( p_sys->pp_images )
 155     {
 156         int i;
 157         for( i = 0; i < p_sys->i_images; i++ )
 158         {
 159             if( !p_sys->pp_images[i] )
 160                 continue;
 161
 162             if( p_sys->pp_images[i]->p_pic )
 163                 picture_Release( p_sys->pp_images[i]->p_pic );
 164             free( p_sys->pp_images[i]->psz_filename );
 165
 166             free( p_sys->pp_images[i] );
 167         }
 168         TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
 169     }
 170
 171     free( p_sys );
 172 }
 173
 174 /*****************************************************************************
 175  * ParseText: parse an text subtitle packet and send it to the video output
 176  *****************************************************************************/
 177 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 178 {
 179     decoder_sys_t *p_sys = p_dec->p_sys;
 180     subpicture_t *p_spu = NULL;
 181     char *psz_subtitle = NULL;
 182
 183     /* We cannot display a subpicture with no date */
 184     if( p_block->i_pts == 0 )
 185     {
 186         msg_Warn( p_dec, "subtitle without a date" );
 187         return NULL;
 188     }
 189
 190     /* Check validity of packet data */
 191     /* An "empty" line containing only \0 can be used to force
 192        and ephemer picture from the screen */
 193     if( p_block->i_buffer < 1 )
 194     {
 195         msg_Warn( p_dec, "no subtitle data" );
 196         return NULL;
 197     }
 198
 199     /* Should be resiliant against bad subtitles */
 200     psz_subtitle = strndup( (const char *)p_block->p_buffer,
 201                             p_block->i_buffer );
 202     if( psz_subtitle == NULL )
 203         return NULL;
 204
 205     /* USF Subtitles are mandated to be UTF-8 -- make sure it is */
 206     if (EnsureUTF8( psz_subtitle ) == NULL)
 207     {
 208         msg_Err( p_dec, "USF subtitles must be in UTF-8 format.\n"
 209                  "This stream contains USF subtitles which aren't." );
 210     }
 211
 212     /* Create the subpicture unit */
 213     p_spu = decoder_NewSubpicture( p_dec );
 214     if( !p_spu )
 215     {
 216         msg_Warn( p_dec, "can't get spu buffer" );
 217         free( psz_subtitle );
 218         return NULL;
 219     }
 220
 221     /* Decode USF strings */
 222     p_spu->p_region = ParseUSFString( p_dec, psz_subtitle, p_spu );
 223
 224     p_spu->i_start = p_block->i_pts;
 225     p_spu->i_stop = p_block->i_pts + p_block->i_length;
 226     p_spu->b_ephemer = (p_block->i_length == 0);
 227     p_spu->b_absolute = false;
 228     p_spu->i_original_picture_width = p_sys->i_original_width;
 229     p_spu->i_original_picture_height = p_sys->i_original_height;
 230
 231     free( psz_subtitle );
 232
 233     return p_spu;
 234 }
 235
 236 static char *GrabAttributeValue( const char *psz_attribute,
 237                                  const char *psz_tag_start )
 238 {
 239     if( psz_attribute && psz_tag_start )
 240     {
 241         char *psz_tag_end = strchr( psz_tag_start, '>' );
 242         char *psz_found   = strcasestr( psz_tag_start, psz_attribute );
 243
 244         if( psz_found )
 245         {
 246             psz_found += strlen( psz_attribute );
 247
 248             if(( *(psz_found++) == '=' ) &&
 249                ( *(psz_found++) == '\"' ))
 250             {
 251                 if( psz_found < psz_tag_end )
 252                 {
 253                     int   i_len = strcspn( psz_found, "\"" );
 254                     return strndup( psz_found, i_len );
 255                 }
 256             }
 257         }
 258     }
 259     return NULL;
 260 }
 261
 262 static ssa_style_t *ParseStyle( decoder_sys_t *p_sys, char *psz_subtitle )
 263 {
 264     ssa_style_t *p_style   = NULL;
 265     char        *psz_style = GrabAttributeValue( "style", psz_subtitle );
 266
 267     if( psz_style )
 268     {
 269         int i;
 270
 271         for( i = 0; i < p_sys->i_ssa_styles; i++ )
 272         {
 273             if( !strcmp( p_sys->pp_ssa_styles[i]->psz_stylename, psz_style ) )
 274                 p_style = p_sys->pp_ssa_styles[i];
 275         }
 276         free( psz_style );
 277     }
 278     return p_style;
 279 }
 280
 281 static int ParsePositionAttributeList( char *psz_subtitle, int *i_align,
 282                                        int *i_x, int *i_y )
 283 {
 284     int   i_mask = 0;
 285
 286     char *psz_align    = GrabAttributeValue( "alignment", psz_subtitle );
 287     char *psz_margin_x = GrabAttributeValue( "horizontal-margin", psz_subtitle );
 288     char *psz_margin_y = GrabAttributeValue( "vertical-margin", psz_subtitle );
 289     /* -- UNSUPPORTED
 290     char *psz_relative = GrabAttributeValue( "relative-to", psz_subtitle );
 291     char *psz_rotate_x = GrabAttributeValue( "rotate-x", psz_subtitle );
 292     char *psz_rotate_y = GrabAttributeValue( "rotate-y", psz_subtitle );
 293     char *psz_rotate_z = GrabAttributeValue( "rotate-z", psz_subtitle );
 294     */
 295
 296     *i_align = SUBPICTURE_ALIGN_BOTTOM;
 297     *i_x = 0;
 298     *i_y = 0;
 299
 300     if( psz_align )
 301     {
 302         if( !strcasecmp( "TopLeft", psz_align ) )
 303             *i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_LEFT;
 304         else if( !strcasecmp( "TopCenter", psz_align ) )
 305             *i_align = SUBPICTURE_ALIGN_TOP;
 306         else if( !strcasecmp( "TopRight", psz_align ) )
 307             *i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_RIGHT;
 308         else if( !strcasecmp( "MiddleLeft", psz_align ) )
 309             *i_align = SUBPICTURE_ALIGN_LEFT;
 310         else if( !strcasecmp( "MiddleCenter", psz_align ) )
 311             *i_align = 0;
 312         else if( !strcasecmp( "MiddleRight", psz_align ) )
 313             *i_align = SUBPICTURE_ALIGN_RIGHT;
 314         else if( !strcasecmp( "BottomLeft", psz_align ) )
 315             *i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_LEFT;
 316         else if( !strcasecmp( "BottomCenter", psz_align ) )
 317             *i_align = SUBPICTURE_ALIGN_BOTTOM;
 318         else if( !strcasecmp( "BottomRight", psz_align ) )
 319             *i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_RIGHT;
 320
 321         i_mask |= ATTRIBUTE_ALIGNMENT;
 322         free( psz_align );
 323     }
 324     if( psz_margin_x )
 325     {
 326         *i_x = atoi( psz_margin_x );
 327         if( strchr( psz_margin_x, '%' ) )
 328             i_mask |= ATTRIBUTE_X_PERCENT;
 329         else
 330             i_mask |= ATTRIBUTE_X;
 331
 332         free( psz_margin_x );
 333     }
 334     if( psz_margin_y )
 335     {
 336         *i_y = atoi( psz_margin_y );
 337         if( strchr( psz_margin_y, '%' ) )
 338             i_mask |= ATTRIBUTE_Y_PERCENT;
 339         else
 340             i_mask |= ATTRIBUTE_Y;
 341
 342         free( psz_margin_y );
 343     }
 344     return i_mask;
 345 }
 346
 347 static void SetupPositions( subpicture_region_t *p_region, char *psz_subtitle )
 348 {
 349     int           i_mask = 0;
 350     int           i_align;
 351     int           i_x, i_y;
 352
 353     i_mask = ParsePositionAttributeList( psz_subtitle, &i_align, &i_x, &i_y );
 354
 355     if( i_mask & ATTRIBUTE_ALIGNMENT )
 356         p_region->i_align = i_align;
 357
 358     /* TODO: Setup % based offsets properly, without adversely affecting
 359      *       everything else in vlc. Will address with separate patch, to
 360      *       prevent this one being any more complicated.
 361      */
 362     if( i_mask & ATTRIBUTE_X )
 363         p_region->i_x = i_x;
 364     else if( i_mask & ATTRIBUTE_X_PERCENT )
 365         p_region->i_x = 0;
 366
 367     if( i_mask & ATTRIBUTE_Y )
 368         p_region->i_y = i_y;
 369     else if( i_mask & ATTRIBUTE_Y_PERCENT )
 370         p_region->i_y = 0;
 371 }
 372
 373 static subpicture_region_t *CreateTextRegion( decoder_t *p_dec,
 374                                               subpicture_t *p_spu,
 375                                               char *psz_subtitle,
 376                                               int i_len,
 377                                               int i_sys_align )
 378 {
 379     decoder_sys_t        *p_sys = p_dec->p_sys;
 380     subpicture_region_t  *p_text_region;
 381     video_format_t        fmt;
 382
 383     /* Create a new subpicture region */
 384     memset( &fmt, 0, sizeof(video_format_t) );
 385     fmt.i_chroma = VLC_FOURCC('T','E','X','T');
 386     fmt.i_aspect = 0;
 387     fmt.i_width = fmt.i_height = 0;
 388     fmt.i_x_offset = fmt.i_y_offset = 0;
 389     p_text_region = subpicture_region_New( &fmt );
 390
 391     if( p_text_region != NULL )
 392     {
 393         ssa_style_t  *p_style = NULL;
 394
 395         p_text_region->psz_text = NULL;
 396         p_text_region->psz_html = strndup( psz_subtitle, i_len );
 397         if( ! p_text_region->psz_html )
 398         {
 399             subpicture_region_Delete( p_text_region );
 400             return NULL;
 401         }
 402
 403         p_style = ParseStyle( p_sys, p_text_region->psz_html );
 404         if( !p_style )
 405         {
 406             int i;
 407
 408             for( i = 0; i < p_sys->i_ssa_styles; i++ )
 409             {
 410                 if( !strcasecmp( p_sys->pp_ssa_styles[i]->psz_stylename, "Default" ) )
 411                     p_style = p_sys->pp_ssa_styles[i];
 412             }
 413         }
 414
 415         if( p_style )
 416         {
 417             msg_Dbg( p_dec, "style is: %s", p_style->psz_stylename );
 418
 419             p_text_region->p_style = &p_style->font_style;
 420             p_text_region->i_align = p_style->i_align;
 421
 422             /* TODO: Setup % based offsets properly, without adversely affecting
 423              *       everything else in vlc. Will address with separate patch,
 424              *       to prevent this one being any more complicated.
 425
 426                      * p_style->i_margin_percent_h;
 427                      * p_style->i_margin_percent_v;
 428              */
 429             p_text_region->i_x         = p_style->i_margin_h;
 430             p_text_region->i_y         = p_style->i_margin_v;
 431
 432         }
 433         else
 434         {
 435             p_text_region->i_align = SUBPICTURE_ALIGN_BOTTOM | i_sys_align;
 436             p_text_region->i_x = i_sys_align ? 20 : 0;
 437             p_text_region->i_y = 10;
 438         }
 439         /* Look for position arguments which may override the style-based
 440          * defaults.
 441          */
 442         SetupPositions( p_text_region, psz_subtitle );
 443
 444         p_text_region->p_next = NULL;
 445     }
 446     return p_text_region;
 447 }
 448
 449 static int ParseImageAttachments( decoder_t *p_dec )
 450 {
 451     decoder_sys_t        *p_sys = p_dec->p_sys;
 452     input_attachment_t  **pp_attachments;
 453     int                   i_attachments_cnt;
 454     int                   k = 0;
 455
 456     if( VLC_SUCCESS != decoder_GetInputAttachments( p_dec, &pp_attachments, &i_attachments_cnt ))
 457         return VLC_EGENERIC;
 458
 459     for( k = 0; k < i_attachments_cnt; k++ )
 460     {
 461         input_attachment_t *p_attach = pp_attachments[k];
 462
 463         vlc_fourcc_t type = image_Mime2Fourcc( p_attach->psz_mime );
 464
 465         if( ( type != 0 ) &&
 466             ( p_attach->i_data > 0 ) &&
 467             ( p_attach->p_data != NULL ) )
 468         {
 469             picture_t         *p_pic = NULL;
 470             image_handler_t   *p_image;
 471
 472             p_image = image_HandlerCreate( p_dec );
 473             if( p_image != NULL )
 474             {
 475                 block_t   *p_block;
 476
 477                 p_block = block_New( p_image->p_parent, p_attach->i_data );
 478
 479                 if( p_block != NULL )
 480                 {
 481                     video_format_t     fmt_in;
 482                     video_format_t     fmt_out;
 483
 484                     memcpy( p_block->p_buffer, p_attach->p_data, p_attach->i_data );
 485
 486                     memset( &fmt_in,  0, sizeof( video_format_t));
 487                     memset( &fmt_out, 0, sizeof( video_format_t));
 488
 489                     fmt_in.i_chroma  = type;
 490                     fmt_out.i_chroma = VLC_FOURCC('Y','U','V','A');
 491
 492                     /* Find a suitable decoder module */
 493                     if( module_exists( "sdl_image" ) )
 494                     {
 495                         /* ffmpeg thinks it can handle bmp properly but it can't (at least
 496                          * not all of them), so use sdl_image if it is available */
 497
 498                         vlc_value_t val;
 499
 500                         var_Create( p_dec, "codec", VLC_VAR_MODULE | VLC_VAR_DOINHERIT );
 501                         val.psz_string = (char*) "sdl_image";
 502                         var_Set( p_dec, "codec", val );
 503                     }
 504
 505                     p_pic = image_Read( p_image, p_block, &fmt_in, &fmt_out );
 506                     var_Destroy( p_dec, "codec" );
 507                 }
 508
 509                 image_HandlerDelete( p_image );
 510             }
 511             if( p_pic )
 512             {
 513                 image_attach_t *p_picture = malloc( sizeof(image_attach_t) );
 514
 515                 if( p_picture )
 516                 {
 517                     p_picture->psz_filename = strdup( p_attach->psz_name );
 518                     p_picture->p_pic = p_pic;
 519
 520                     TAB_APPEND( p_sys->i_images, p_sys->pp_images, p_picture );
 521                 }
 522             }
 523         }
 524         vlc_input_attachment_Delete( pp_attachments[ k ] );
 525     }
 526     free( pp_attachments );
 527
 528     return VLC_SUCCESS;
 529 }
 530
 531 static void ParseUSFHeaderTags( decoder_t *p_dec, xml_reader_t *p_xml_reader )
 532 {
 533     decoder_sys_t *p_sys = p_dec->p_sys;
 534     char *psz_node;
 535     ssa_style_t *p_style = NULL;
 536     int i_style_level = 0;
 537     int i_metadata_level = 0;
 538
 539     while ( xml_ReaderRead( p_xml_reader ) == 1 )
 540     {
 541         switch ( xml_ReaderNodeType( p_xml_reader ) )
 542         {
 543             case XML_READER_TEXT:
 544             case XML_READER_NONE:
 545                 break;
 546             case XML_READER_ENDELEM:
 547                 psz_node = xml_ReaderName( p_xml_reader );
 548
 549                 if( !psz_node )
 550                     break;
 551                 switch (i_style_level)
 552                 {
 553                     case 0:
 554                         if( !strcasecmp( "metadata", psz_node ) && (i_metadata_level == 1) )
 555                         {
 556                             i_metadata_level--;
 557                         }
 558                         break;
 559                     case 1:
 560                         if( !strcasecmp( "styles", psz_node ) )
 561                         {
 562                             i_style_level--;
 563                         }
 564                         break;
 565                     case 2:
 566                         if( !strcasecmp( "style", psz_node ) )
 567                         {
 568                             TAB_APPEND( p_sys->i_ssa_styles, p_sys->pp_ssa_styles, p_style );
 569
 570                             p_style = NULL;
 571                             i_style_level--;
 572                         }
 573                         break;
 574                 }
 575
 576                 free( psz_node );
 577                 break;
 578             case XML_READER_STARTELEM:
 579                 psz_node = xml_ReaderName( p_xml_reader );
 580
 581                 if( !psz_node )
 582                     break;
 583
 584                 if( !strcasecmp( "metadata", psz_node ) && (i_style_level == 0) )
 585                 {
 586                     i_metadata_level++;
 587                 }
 588                 else if( !strcasecmp( "resolution", psz_node ) &&
 589                          ( i_metadata_level == 1) )
 590                 {
 591                     while ( xml_ReaderNextAttr( p_xml_reader ) == VLC_SUCCESS )
 592                     {
 593                         char *psz_name = xml_ReaderName ( p_xml_reader );
 594                         char *psz_value = xml_ReaderValue ( p_xml_reader );
 595
 596                         if( psz_name && psz_value )
 597                         {
 598                             if( !strcasecmp( "x", psz_name ) )
 599                                 p_sys->i_original_width = atoi( psz_value );
 600                             else if( !strcasecmp( "y", psz_name ) )
 601                                 p_sys->i_original_height = atoi( psz_value );
 602                         }
 603                         free( psz_name );
 604                         free( psz_value );
 605                     }
 606                 }
 607                 else if( !strcasecmp( "styles", psz_node ) && (i_style_level == 0) )
 608                 {
 609                     i_style_level++;
 610                 }
 611                 else if( !strcasecmp( "style", psz_node ) && (i_style_level == 1) )
 612                 {
 613                     i_style_level++;
 614
 615                     p_style = calloc( 1, sizeof(ssa_style_t) );
 616                     if( ! p_style )
 617                     {
 618                         free( psz_node );
 619                         return;
 620                     }
 621                     /* All styles are supposed to default to Default, and then
 622                      * one or more settings are over-ridden.
 623                      * At the moment this only effects styles defined AFTER
 624                      * Default in the XML
 625                      */
 626                     int i;
 627                     for( i = 0; i < p_sys->i_ssa_styles; i++ )
 628                     {
 629                         if( !strcasecmp( p_sys->pp_ssa_styles[i]->psz_stylename, "Default" ) )
 630                         {
 631                             ssa_style_t *p_default_style = p_sys->pp_ssa_styles[i];
 632
 633                             memcpy( p_style, p_default_style, sizeof( ssa_style_t ) );
 634                             p_style->font_style.psz_fontname = strdup( p_style->font_style.psz_fontname );
 635                             p_style->psz_stylename = NULL;
 636                         }
 637                     }
 638
 639                     while ( xml_ReaderNextAttr( p_xml_reader ) == VLC_SUCCESS )
 640                     {
 641                         char *psz_name = xml_ReaderName ( p_xml_reader );
 642                         char *psz_value = xml_ReaderValue ( p_xml_reader );
 643
 644                         if( psz_name && psz_value )
 645                         {
 646                             if( !strcasecmp( "name", psz_name ) )
 647                                 p_style->psz_stylename = strdup( psz_value);
 648                         }
 649                         free( psz_name );
 650                         free( psz_value );
 651                     }
 652                 }
 653                 else if( !strcasecmp( "fontstyle", psz_node ) && (i_style_level == 2) )
 654                 {
 655                     while ( xml_ReaderNextAttr( p_xml_reader ) == VLC_SUCCESS )
 656                     {
 657                         char *psz_name = xml_ReaderName ( p_xml_reader );
 658                         char *psz_value = xml_ReaderValue ( p_xml_reader );
 659
 660                         if( psz_name && psz_value )
 661                         {
 662                             if( !strcasecmp( "face", psz_name ) )
 663                             {
 664                                 free( p_style->font_style.psz_fontname );
 665                                 p_style->font_style.psz_fontname = strdup( psz_value );
 666                             }
 667                             else if( !strcasecmp( "size", psz_name ) )
 668                             {
 669                                 if( ( *psz_value == '+' ) || ( *psz_value == '-' ) )
 670                                 {
 671                                     int i_value = atoi( psz_value );
 672
 673                                     if( ( i_value >= -5 ) && ( i_value <= 5 ) )
 674                                         p_style->font_style.i_font_size  +=
 675                                             ( i_value * p_style->font_style.i_font_size ) / 10;
 676                                     else if( i_value < -5 )
 677                                         p_style->font_style.i_font_size  = - i_value;
 678                                     else if( i_value > 5 )
 679                                         p_style->font_style.i_font_size  = i_value;
 680                                 }
 681                                 else
 682                                     p_style->font_style.i_font_size  = atoi( psz_value );
 683                             }
 684                             else if( !strcasecmp( "italic", psz_name ) )
 685                             {
 686                                 if( !strcasecmp( "yes", psz_value ))
 687                                     p_style->font_style.i_style_flags |= STYLE_ITALIC;
 688                                 else
 689                                     p_style->font_style.i_style_flags &= ~STYLE_ITALIC;
 690                             }
 691                             else if( !strcasecmp( "weight", psz_name ) )
 692                             {
 693                                 if( !strcasecmp( "bold", psz_value ))
 694                                     p_style->font_style.i_style_flags |= STYLE_BOLD;
 695                                 else
 696                                     p_style->font_style.i_style_flags &= ~STYLE_BOLD;
 697                             }
 698                             else if( !strcasecmp( "underline", psz_name ) )
 699                             {
 700                                 if( !strcasecmp( "yes", psz_value ))
 701                                     p_style->font_style.i_style_flags |= STYLE_UNDERLINE;
 702                                 else
 703                                     p_style->font_style.i_style_flags &= ~STYLE_UNDERLINE;
 704                             }
 705                             else if( !strcasecmp( "color", psz_name ) )
 706                             {
 707                                 if( *psz_value == '#' )
 708                                 {
 709                                     unsigned long col = strtol(psz_value+1, NULL, 16);
 710                                     p_style->font_style.i_font_color = (col & 0x00ffffff);
 711                                     p_style->font_style.i_font_alpha = (col >> 24) & 0xff;
 712                                 }
 713                             }
 714                             else if( !strcasecmp( "outline-color", psz_name ) )
 715                             {
 716                                 if( *psz_value == '#' )
 717                                 {
 718                                     unsigned long col = strtol(psz_value+1, NULL, 16);
 719                                     p_style->font_style.i_outline_color = (col & 0x00ffffff);
 720                                     p_style->font_style.i_outline_alpha = (col >> 24) & 0xff;
 721                                 }
 722                             }
 723                             else if( !strcasecmp( "outline-level", psz_name ) )
 724                             {
 725                                 p_style->font_style.i_outline_width = atoi( psz_value );
 726                             }
 727                             else if( !strcasecmp( "shadow-color", psz_name ) )
 728                             {
 729                                 if( *psz_value == '#' )
 730                                 {
 731                                     unsigned long col = strtol(psz_value+1, NULL, 16);
 732                                     p_style->font_style.i_shadow_color = (col & 0x00ffffff);
 733                                     p_style->font_style.i_shadow_alpha = (col >> 24) & 0xff;
 734                                 }
 735                             }
 736                             else if( !strcasecmp( "shadow-level", psz_name ) )
 737                             {
 738                                 p_style->font_style.i_shadow_width = atoi( psz_value );
 739                             }
 740                             else if( !strcasecmp( "back-color", psz_name ) )
 741                             {
 742                                 if( *psz_value == '#' )
 743                                 {
 744                                     unsigned long col = strtol(psz_value+1, NULL, 16);
 745                                     p_style->font_style.i_karaoke_background_color = (col & 0x00ffffff);
 746                                     p_style->font_style.i_karaoke_background_alpha = (col >> 24) & 0xff;
 747                                 }
 748                             }
 749                             else if( !strcasecmp( "spacing", psz_name ) )
 750                             {
 751                                 p_style->font_style.i_spacing = atoi( psz_value );
 752                             }
 753                         }
 754                         free( psz_name );
 755                         free( psz_value );
 756                     }
 757                 }
 758                 else if( !strcasecmp( "position", psz_node ) && (i_style_level == 2) )
 759                 {
 760                     while ( xml_ReaderNextAttr( p_xml_reader ) == VLC_SUCCESS )
 761                     {
 762                         char *psz_name = xml_ReaderName ( p_xml_reader );
 763                         char *psz_value = xml_ReaderValue ( p_xml_reader );
 764
 765                         if( psz_name && psz_value )
 766                         {
 767                             if( !strcasecmp( "alignment", psz_name ) )
 768                             {
 769                                 if( !strcasecmp( "TopLeft", psz_value ) )
 770                                     p_style->i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_LEFT;
 771                                 else if( !strcasecmp( "TopCenter", psz_value ) )
 772                                     p_style->i_align = SUBPICTURE_ALIGN_TOP;
 773                                 else if( !strcasecmp( "TopRight", psz_value ) )
 774                                     p_style->i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_RIGHT;
 775                                 else if( !strcasecmp( "MiddleLeft", psz_value ) )
 776                                     p_style->i_align = SUBPICTURE_ALIGN_LEFT;
 777                                 else if( !strcasecmp( "MiddleCenter", psz_value ) )
 778                                     p_style->i_align = 0;
 779                                 else if( !strcasecmp( "MiddleRight", psz_value ) )
 780                                     p_style->i_align = SUBPICTURE_ALIGN_RIGHT;
 781                                 else if( !strcasecmp( "BottomLeft", psz_value ) )
 782                                     p_style->i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_LEFT;
 783                                 else if( !strcasecmp( "BottomCenter", psz_value ) )
 784                                     p_style->i_align = SUBPICTURE_ALIGN_BOTTOM;
 785                                 else if( !strcasecmp( "BottomRight", psz_value ) )
 786                                     p_style->i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_RIGHT;
 787                             }
 788                             else if( !strcasecmp( "horizontal-margin", psz_name ) )
 789                             {
 790                                 if( strchr( psz_value, '%' ) )
 791                                 {
 792                                     p_style->i_margin_h = 0;
 793                                     p_style->i_margin_percent_h = atoi( psz_value );
 794                                 }
 795                                 else
 796                                 {
 797                                     p_style->i_margin_h = atoi( psz_value );
 798                                     p_style->i_margin_percent_h = 0;
 799                                 }
 800                             }
 801                             else if( !strcasecmp( "vertical-margin", psz_name ) )
 802                             {
 803                                 if( strchr( psz_value, '%' ) )
 804                                 {
 805                                     p_style->i_margin_v = 0;
 806                                     p_style->i_margin_percent_v = atoi( psz_value );
 807                                 }
 808                                 else
 809                                 {
 810                                     p_style->i_margin_v = atoi( psz_value );
 811                                     p_style->i_margin_percent_v = 0;
 812                                 }
 813                             }
 814                         }
 815                         free( psz_name );
 816                         free( psz_value );
 817                     }
 818                 }
 819
 820                 free( psz_node );
 821                 break;
 822         }
 823     }
 824     free( p_style );
 825 }
 826
 827
 828
 829 static subpicture_region_t *ParseUSFString( decoder_t *p_dec,
 830                                             char *psz_subtitle,
 831                                             subpicture_t *p_spu_in )
 832 {
 833     decoder_sys_t        *p_sys = p_dec->p_sys;
 834     subpicture_t         *p_spu = p_spu_in;
 835     subpicture_region_t  *p_region_first = NULL;
 836     subpicture_region_t  *p_region_upto  = p_region_first;
 837
 838     while( *psz_subtitle )
 839     {
 840         if( *psz_subtitle == '<' )
 841         {
 842             char *psz_end = NULL;
 843
 844             if(( !strncasecmp( psz_subtitle, "<text ", 6 )) ||
 845                ( !strncasecmp( psz_subtitle, "<text>", 6 )))
 846             {
 847                 psz_end = strcasestr( psz_subtitle, "</text>" );
 848
 849                 if( psz_end )
 850                 {
 851                     subpicture_region_t  *p_text_region;
 852
 853                     psz_end += strcspn( psz_end, ">" ) + 1;
 854
 855                     p_text_region = CreateTextRegion( p_dec,
 856                                                       p_spu,
 857                                                       psz_subtitle,
 858                                                       psz_end - psz_subtitle,
 859                                                       p_sys->i_align );
 860
 861                     if( p_text_region )
 862                     {
 863                         p_text_region->psz_text = CreatePlainText( p_text_region->psz_html );
 864
 865                         if( ! var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 866                         {
 867                             free( p_text_region->psz_html );
 868                             p_text_region->psz_html = NULL;
 869                         }
 870                     }
 871
 872                     if( !p_region_first )
 873                     {
 874                         p_region_first = p_region_upto = p_text_region;
 875                     }
 876                     else if( p_text_region )
 877                     {
 878                         p_region_upto->p_next = p_text_region;
 879                         p_region_upto = p_region_upto->p_next;
 880                     }
 881                 }
 882             }
 883             else if(( !strncasecmp( psz_subtitle, "<karaoke ", 9 )) ||
 884                     ( !strncasecmp( psz_subtitle, "<karaoke>", 9 )))
 885             {
 886                 psz_end = strcasestr( psz_subtitle, "</karaoke>" );
 887
 888                 if( psz_end )
 889                 {
 890                     subpicture_region_t  *p_text_region;
 891
 892                     psz_end += strcspn( psz_end, ">" ) + 1;
 893
 894                     p_text_region = CreateTextRegion( p_dec,
 895                                                       p_spu,
 896                                                       psz_subtitle,
 897                                                       psz_end - psz_subtitle,
 898                                                       p_sys->i_align );
 899
 900                     if( p_text_region )
 901                     {
 902                         if( ! var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 903                         {
 904                             free( p_text_region->psz_html );
 905                             p_text_region->psz_html = NULL;
 906                         }
 907                     }
 908                     if( !p_region_first )
 909                     {
 910                         p_region_first = p_region_upto = p_text_region;
 911                     }
 912                     else if( p_text_region )
 913                     {
 914                         p_region_upto->p_next = p_text_region;
 915                         p_region_upto = p_region_upto->p_next;
 916                     }
 917                 }
 918             }
 919             else if(( !strncasecmp( psz_subtitle, "<image ", 7 )) ||
 920                     ( !strncasecmp( psz_subtitle, "<image>", 7 )))
 921             {
 922                 subpicture_region_t *p_image_region = NULL;
 923
 924                 char *psz_end = strcasestr( psz_subtitle, "</image>" );
 925                 char *psz_content = strchr( psz_subtitle, '>' );
 926                 int   i_transparent = -1;
 927
 928                 /* If a colorkey parameter is specified, then we have to map
 929                  * that index in the picture through as transparent (it is
 930                  * required by the USF spec but is also recommended that if the
 931                  * creator really wants a transparent colour that they use a
 932                  * type like PNG that properly supports it; this goes doubly
 933                  * for VLC because the pictures are stored internally in YUV
 934                  * and the resulting colour-matching may not produce the
 935                  * desired results.)
 936                  */
 937                 char *psz_tmp = GrabAttributeValue( "colorkey", psz_subtitle );
 938                 if( psz_tmp )
 939                 {
 940                     if( *psz_tmp == '#' )
 941                         i_transparent = strtol( psz_tmp + 1, NULL, 16 ) & 0x00ffffff;
 942                     free( psz_tmp );
 943                 }
 944                 if( psz_content && ( psz_content < psz_end ) )
 945                 {
 946                     char *psz_filename = strndup( &psz_content[1], psz_end - &psz_content[1] );
 947                     if( psz_filename )
 948                     {
 949                         p_image_region = LoadEmbeddedImage( p_dec, p_spu,
 950                                             psz_filename, i_transparent );
 951                         free( psz_filename );
 952                     }
 953                 }
 954
 955                 if( psz_end ) psz_end += strcspn( psz_end, ">" ) + 1;
 956
 957                 if( p_image_region )
 958                 {
 959                     SetupPositions( p_image_region, psz_subtitle );
 960
 961                     p_image_region->p_next   = NULL;
 962                     p_image_region->psz_text = NULL;
 963                     p_image_region->psz_html = NULL;
 964
 965                 }
 966                 if( !p_region_first )
 967                 {
 968                     p_region_first = p_region_upto = p_image_region;
 969                 }
 970                 else if( p_image_region )
 971                 {
 972                     p_region_upto->p_next = p_image_region;
 973                     p_region_upto = p_region_upto->p_next;
 974                 }
 975             }
 976             if( psz_end )
 977                 psz_subtitle = psz_end - 1;
 978
 979             psz_subtitle += strcspn( psz_subtitle, ">" );
 980         }
 981
 982         psz_subtitle++;
 983     }
 984
 985     return p_region_first;
 986 }
 987
 988 /*****************************************************************************
 989  * ParseUSFHeader: Retrieve global formatting information etc
 990  *****************************************************************************/
 991 static void ParseUSFHeader( decoder_t *p_dec )
 992 {
 993     stream_t      *p_sub = NULL;
 994     xml_t         *p_xml = NULL;
 995     xml_reader_t  *p_xml_reader = NULL;
 996
 997     p_sub = stream_MemoryNew( VLC_OBJECT(p_dec),
 998                               p_dec->fmt_in.p_extra,
 999                               p_dec->fmt_in.i_extra,
1000                               true );
1001     if( !p_sub )
1002         return;
1003
1004     p_xml = xml_Create( p_dec );
1005     if( p_xml )
1006     {
1007         p_xml_reader = xml_ReaderCreate( p_xml, p_sub );
1008         if( p_xml_reader )
1009         {
1010             /* Look for Root Node */
1011             if( xml_ReaderRead( p_xml_reader ) == 1 )
1012             {
1013                 char *psz_node = xml_ReaderName( p_xml_reader );
1014
1015                 if( !strcasecmp( "usfsubtitles", psz_node ) )
1016                     ParseUSFHeaderTags( p_dec, p_xml_reader );
1017
1018                 free( psz_node );
1019             }
1020
1021             xml_ReaderDelete( p_xml, p_xml_reader );
1022         }
1023         xml_Delete( p_xml );
1024     }
1025     stream_Delete( p_sub );
1026 }
1027
1028 /* Function now handles tags which has attribute values, and tries
1029  * to deal with &' commands too. It no longer modifies the string
1030  * in place, so that the original text can be reused
1031  */
1032 static char *StripTags( char *psz_subtitle )
1033 {
1034     char *psz_text_start;
1035     char *psz_text;
1036
1037     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
1038     if( !psz_text_start )
1039         return NULL;
1040
1041     while( *psz_subtitle )
1042     {
1043         /* Mask out any pre-existing LFs in the subtitle */
1044         if( *psz_subtitle == '\n' )
1045             *psz_subtitle = ' ';
1046
1047         if( *psz_subtitle == '<' )
1048         {
1049             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
1050                 *psz_text++ = '\n';
1051
1052             psz_subtitle += strcspn( psz_subtitle, ">" );
1053         }
1054         else if( *psz_subtitle == '&' )
1055         {
1056             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
1057             {
1058                 *psz_text++ = '<';
1059                 psz_subtitle += strcspn( psz_subtitle, ";" );
1060             }
1061             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
1062             {
1063                 *psz_text++ = '>';
1064                 psz_subtitle += strcspn( psz_subtitle, ";" );
1065             }
1066             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
1067             {
1068                 *psz_text++ = '&';
1069                 psz_subtitle += strcspn( psz_subtitle, ";" );
1070             }
1071             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
1072             {
1073                 *psz_text++ = '\"';
1074                 psz_subtitle += strcspn( psz_subtitle, ";" );
1075             }
1076             else
1077             {
1078                 /* Assume it is just a normal ampersand */
1079                 *psz_text++ = '&';
1080             }
1081         }
1082         else
1083         {
1084             *psz_text++ = *psz_subtitle;
1085         }
1086
1087         psz_subtitle++;
1088     }
1089     *psz_text = '\0';
1090     psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
1091
1092     return psz_text_start;
1093 }
1094
1095 /* Turn a HTML subtitle, turn into a plain-text version,
1096  *  complete with sensible whitespace compaction
1097  */
1098
1099 static char *CreatePlainText( char *psz_subtitle )
1100 {
1101     char *psz_text = StripTags( psz_subtitle );
1102     char *s;
1103
1104     if( !psz_text )
1105         return NULL;
1106
1107     s = strpbrk( psz_text, "\t\r\n " );
1108     while( s )
1109     {
1110         int   k;
1111         char  spc = ' ';
1112         int   i_whitespace = strspn( s, "\t\r\n " );
1113
1114         /* Favour '\n' over other whitespaces - if one of these
1115          * occurs in the whitespace use a '\n' as our value,
1116          * otherwise just use a ' '
1117          */
1118         for( k = 0; k < i_whitespace; k++ )
1119             if( s[k] == '\n' ) spc = '\n';
1120
1121         if( i_whitespace > 1 )
1122         {
1123             memmove( &s[1],
1124                      &s[i_whitespace],
1125                      strlen( s ) - i_whitespace + 1 );
1126         }
1127         *s++ = spc;
1128
1129         s = strpbrk( s, "\t\r\n " );
1130     }
1131     return psz_text;
1132 }
1133
1134 /****************************************************************************
1135  * download and resize image located at psz_url
1136  ***************************************************************************/
1137 static subpicture_region_t *LoadEmbeddedImage( decoder_t *p_dec,
1138                                                subpicture_t *p_spu,
1139                                                const char *psz_filename,
1140                                                int i_transparent_color )
1141 {
1142     decoder_sys_t         *p_sys = p_dec->p_sys;
1143     subpicture_region_t   *p_region;
1144     video_format_t         fmt_out;
1145     int                    k;
1146     picture_t             *p_pic = NULL;
1147
1148     for( k = 0; k < p_sys->i_images; k++ )
1149     {
1150         if( p_sys->pp_images &&
1151             !strcmp( p_sys->pp_images[k]->psz_filename, psz_filename ) )
1152         {
1153             p_pic = p_sys->pp_images[k]->p_pic;
1154             break;
1155         }
1156     }
1157
1158     if( !p_pic )
1159     {
1160         msg_Err( p_dec, "Unable to read image %s", psz_filename );
1161         return NULL;
1162     }
1163
1164     /* Display the feed's image */
1165     memset( &fmt_out, 0, sizeof( video_format_t));
1166
1167     fmt_out.i_chroma = VLC_FOURCC('Y','U','V','A');
1168     fmt_out.i_aspect = VOUT_ASPECT_FACTOR;
1169     fmt_out.i_sar_num = fmt_out.i_sar_den = 1;
1170     fmt_out.i_width =
1171         fmt_out.i_visible_width = p_pic->format.i_visible_width;
1172     fmt_out.i_height =
1173         fmt_out.i_visible_height = p_pic->format.i_visible_height;
1174
1175     p_region = subpicture_region_New( &fmt_out );
1176     if( !p_region )
1177     {
1178         msg_Err( p_dec, "cannot allocate SPU region" );
1179         return NULL;
1180     }
1181     assert( p_pic->format.i_chroma == VLC_FOURCC('Y','U','V','A') );
1182     /* FIXME the copy is probably not needed anymore */
1183     picture_CopyPixels( p_region->p_picture, p_pic );
1184
1185     /* This isn't the best way to do this - if you really want transparency, then
1186      * you're much better off using an image type that supports it like PNG. The
1187      * spec requires this support though.
1188      */
1189     if( i_transparent_color > 0 )
1190     {
1191         int i_r = ( i_transparent_color >> 16 ) & 0xff;
1192         int i_g = ( i_transparent_color >>  8 ) & 0xff;
1193         int i_b = ( i_transparent_color       ) & 0xff;
1194
1195         /* FIXME it cannot work as the yuv conversion code will probably NOT match
1196          * this one  */
1197         int i_y = ( ( (  66 * i_r + 129 * i_g +  25 * i_b + 128 ) >> 8 ) + 16 );
1198         int i_u =   ( ( -38 * i_r -  74 * i_g + 112 * i_b + 128 ) >> 8 ) + 128 ;
1199         int i_v =   ( ( 112 * i_r -  94 * i_g -  18 * i_b + 128 ) >> 8 ) + 128 ;
1200
1201         assert( p_region->fmt.i_chroma == VLC_FOURCC('Y','U','V','A') );
1202         for( unsigned int y = 0; y < p_region->fmt.i_height; y++ )
1203         {
1204             for( unsigned int x = 0; x < p_region->fmt.i_width; x++ )
1205             {
1206                 if( p_region->p_picture->Y_PIXELS[y*p_region->p_picture->Y_PITCH + x] != i_y ||
1207                     p_region->p_picture->U_PIXELS[y*p_region->p_picture->U_PITCH + x] != i_u ||
1208                     p_region->p_picture->V_PIXELS[y*p_region->p_picture->V_PITCH + x] != i_v )
1209                     continue;
1210                 p_region->p_picture->A_PIXELS[y*p_region->p_picture->A_PITCH + x] = 0;
1211
1212             }
1213         }
1214     }
1215     return p_region;
1216 }