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