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             //FIXME: Make font_style a pointer and use text_style_* functions
 150             free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
 151             free( p_sys->pp_ssa_styles[i] );
 152         }
 153         TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 154     }
 155     if( p_sys->pp_images )
 156     {
 157         int i;
 158         for( i = 0; i < p_sys->i_images; i++ )
 159         {
 160             if( !p_sys->pp_images[i] )
 161                 continue;
 162
 163             if( p_sys->pp_images[i]->p_pic )
 164                 picture_Release( p_sys->pp_images[i]->p_pic );
 165             free( p_sys->pp_images[i]->psz_filename );
 166
 167             free( p_sys->pp_images[i] );
 168         }
 169         TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
 170     }
 171
 172     free( p_sys );
 173 }
 174
 175 /*****************************************************************************
 176  * ParseText: parse an text subtitle packet and send it to the video output
 177  *****************************************************************************/
 178 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 179 {
 180     decoder_sys_t *p_sys = p_dec->p_sys;
 181     subpicture_t *p_spu = NULL;
 182     char *psz_subtitle = NULL;
 183
 184     /* We cannot display a subpicture with no date */
 185     if( p_block->i_pts == 0 )
 186     {
 187         msg_Warn( p_dec, "subtitle without a date" );
 188         return NULL;
 189     }
 190
 191     /* Check validity of packet data */
 192     /* An "empty" line containing only \0 can be used to force
 193        and ephemer picture from the screen */
 194     if( p_block->i_buffer < 1 )
 195     {
 196         msg_Warn( p_dec, "no subtitle data" );
 197         return NULL;
 198     }
 199
 200     /* Should be resiliant against bad subtitles */
 201     psz_subtitle = strndup( (const char *)p_block->p_buffer,
 202                             p_block->i_buffer );
 203     if( psz_subtitle == NULL )
 204         return NULL;
 205
 206     /* USF Subtitles are mandated to be UTF-8 -- make sure it is */
 207     if (EnsureUTF8( psz_subtitle ) == NULL)
 208     {
 209         msg_Err( p_dec, "USF subtitles must be in UTF-8 format.\n"
 210                  "This stream contains USF subtitles which aren't." );
 211     }
 212
 213     /* Create the subpicture unit */
 214     p_spu = decoder_NewSubpicture( p_dec );
 215     if( !p_spu )
 216     {
 217         msg_Warn( p_dec, "can't get spu buffer" );
 218         free( psz_subtitle );
 219         return NULL;
 220     }
 221
 222     /* Decode USF strings */
 223     p_spu->p_region = ParseUSFString( p_dec, psz_subtitle, p_spu );
 224
 225     p_spu->i_start = p_block->i_pts;
 226     p_spu->i_stop = p_block->i_pts + p_block->i_length;
 227     p_spu->b_ephemer = (p_block->i_length == 0);
 228     p_spu->b_absolute = false;
 229     p_spu->i_original_picture_width = p_sys->i_original_width;
 230     p_spu->i_original_picture_height = p_sys->i_original_height;
 231
 232     free( psz_subtitle );
 233
 234     return p_spu;
 235 }
 236
 237 static char *GrabAttributeValue( const char *psz_attribute,
 238                                  const char *psz_tag_start )
 239 {
 240     if( psz_attribute && psz_tag_start )
 241     {
 242         char *psz_tag_end = strchr( psz_tag_start, '>' );
 243         char *psz_found   = strcasestr( psz_tag_start, psz_attribute );
 244
 245         if( psz_found )
 246         {
 247             psz_found += strlen( psz_attribute );
 248
 249             if(( *(psz_found++) == '=' ) &&
 250                ( *(psz_found++) == '\"' ))
 251             {
 252                 if( psz_found < psz_tag_end )
 253                 {
 254                     int   i_len = strcspn( psz_found, "\"" );
 255                     return strndup( psz_found, i_len );
 256                 }
 257             }
 258         }
 259     }
 260     return NULL;
 261 }
 262
 263 static ssa_style_t *ParseStyle( decoder_sys_t *p_sys, char *psz_subtitle )
 264 {
 265     ssa_style_t *p_ssa_style = NULL;
 266     char        *psz_style = GrabAttributeValue( "style", psz_subtitle );
 267
 268     if( psz_style )
 269     {
 270         int i;
 271
 272         for( i = 0; i < p_sys->i_ssa_styles; i++ )
 273         {
 274             if( !strcmp( p_sys->pp_ssa_styles[i]->psz_stylename, psz_style ) )
 275                 p_ssa_style = p_sys->pp_ssa_styles[i];
 276         }
 277         free( psz_style );
 278     }
 279     return p_ssa_style;
 280 }
 281
 282 static int ParsePositionAttributeList( char *psz_subtitle, int *i_align,
 283                                        int *i_x, int *i_y )
 284 {
 285     int   i_mask = 0;
 286
 287     char *psz_align    = GrabAttributeValue( "alignment", psz_subtitle );
 288     char *psz_margin_x = GrabAttributeValue( "horizontal-margin", psz_subtitle );
 289     char *psz_margin_y = GrabAttributeValue( "vertical-margin", psz_subtitle );
 290     /* -- UNSUPPORTED
 291     char *psz_relative = GrabAttributeValue( "relative-to", psz_subtitle );
 292     char *psz_rotate_x = GrabAttributeValue( "rotate-x", psz_subtitle );
 293     char *psz_rotate_y = GrabAttributeValue( "rotate-y", psz_subtitle );
 294     char *psz_rotate_z = GrabAttributeValue( "rotate-z", psz_subtitle );
 295     */
 296
 297     *i_align = SUBPICTURE_ALIGN_BOTTOM;
 298     *i_x = 0;
 299     *i_y = 0;
 300
 301     if( psz_align )
 302     {
 303         if( !strcasecmp( "TopLeft", psz_align ) )
 304             *i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_LEFT;
 305         else if( !strcasecmp( "TopCenter", psz_align ) )
 306             *i_align = SUBPICTURE_ALIGN_TOP;
 307         else if( !strcasecmp( "TopRight", psz_align ) )
 308             *i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_RIGHT;
 309         else if( !strcasecmp( "MiddleLeft", psz_align ) )
 310             *i_align = SUBPICTURE_ALIGN_LEFT;
 311         else if( !strcasecmp( "MiddleCenter", psz_align ) )
 312             *i_align = 0;
 313         else if( !strcasecmp( "MiddleRight", psz_align ) )
 314             *i_align = SUBPICTURE_ALIGN_RIGHT;
 315         else if( !strcasecmp( "BottomLeft", psz_align ) )
 316             *i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_LEFT;
 317         else if( !strcasecmp( "BottomCenter", psz_align ) )
 318             *i_align = SUBPICTURE_ALIGN_BOTTOM;
 319         else if( !strcasecmp( "BottomRight", psz_align ) )
 320             *i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_RIGHT;
 321
 322         i_mask |= ATTRIBUTE_ALIGNMENT;
 323         free( psz_align );
 324     }
 325     if( psz_margin_x )
 326     {
 327         *i_x = atoi( psz_margin_x );
 328         if( strchr( psz_margin_x, '%' ) )
 329             i_mask |= ATTRIBUTE_X_PERCENT;
 330         else
 331             i_mask |= ATTRIBUTE_X;
 332
 333         free( psz_margin_x );
 334     }
 335     if( psz_margin_y )
 336     {
 337         *i_y = atoi( psz_margin_y );
 338         if( strchr( psz_margin_y, '%' ) )
 339             i_mask |= ATTRIBUTE_Y_PERCENT;
 340         else
 341             i_mask |= ATTRIBUTE_Y;
 342
 343         free( psz_margin_y );
 344     }
 345     return i_mask;
 346 }
 347
 348 static void SetupPositions( subpicture_region_t *p_region, char *psz_subtitle )
 349 {
 350     int           i_mask = 0;
 351     int           i_align;
 352     int           i_x, i_y;
 353
 354     i_mask = ParsePositionAttributeList( psz_subtitle, &i_align, &i_x, &i_y );
 355
 356     if( i_mask & ATTRIBUTE_ALIGNMENT )
 357         p_region->i_align = i_align;
 358
 359     /* TODO: Setup % based offsets properly, without adversely affecting
 360      *       everything else in vlc. Will address with separate patch, to
 361      *       prevent this one being any more complicated.
 362      */
 363     if( i_mask & ATTRIBUTE_X )
 364         p_region->i_x = i_x;
 365     else if( i_mask & ATTRIBUTE_X_PERCENT )
 366         p_region->i_x = 0;
 367
 368     if( i_mask & ATTRIBUTE_Y )
 369         p_region->i_y = i_y;
 370     else if( i_mask & ATTRIBUTE_Y_PERCENT )
 371         p_region->i_y = 0;
 372 }
 373
 374 static subpicture_region_t *CreateTextRegion( decoder_t *p_dec,
 375                                               subpicture_t *p_spu,
 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_FOURCC('T','E','X','T');
 387     fmt.i_aspect = 0;
 388     fmt.i_width = fmt.i_height = 0;
 389     fmt.i_x_offset = fmt.i_y_offset = 0;
 390     p_text_region = subpicture_region_New( &fmt );
 391
 392     if( p_text_region != NULL )
 393     {
 394         ssa_style_t  *p_ssa_style = NULL;
 395
 396         p_text_region->psz_text = NULL;
 397         p_text_region->psz_html = strndup( psz_subtitle, i_len );
 398         if( ! p_text_region->psz_html )
 399         {
 400             subpicture_region_Delete( p_text_region );
 401             return NULL;
 402         }
 403
 404         p_ssa_style = ParseStyle( p_sys, p_text_region->psz_html );
 405         if( !p_ssa_style )
 406         {
 407             int i;
 408
 409             for( i = 0; i < p_sys->i_ssa_styles; i++ )
 410             {
 411                 if( !strcasecmp( p_sys->pp_ssa_styles[i]->psz_stylename, "Default" ) )
 412                     p_ssa_style = p_sys->pp_ssa_styles[i];
 413             }
 414         }
 415
 416         if( p_ssa_style )
 417         {
 418             msg_Dbg( p_dec, "style is: %s", p_ssa_style->psz_stylename );
 419
 420             p_text_region->p_style = text_style_Duplicate( &p_ssa_style->font_style );
 421             p_text_region->i_align = p_ssa_style->i_align;
 422
 423             /* TODO: Setup % based offsets properly, without adversely affecting
 424              *       everything else in vlc. Will address with separate patch,
 425              *       to prevent this one being any more complicated.
 426
 427                      * p_ssa_style->i_margin_percent_h;
 428                      * p_ssa_style->i_margin_percent_v;
 429              */
 430             p_text_region->i_x         = p_ssa_style->i_margin_h;
 431             p_text_region->i_y         = p_ssa_style->i_margin_v;
 432
 433         }
 434         else
 435         {
 436             p_text_region->i_align = SUBPICTURE_ALIGN_BOTTOM | i_sys_align;
 437             p_text_region->i_x = i_sys_align ? 20 : 0;
 438             p_text_region->i_y = 10;
 439         }
 440         /* Look for position arguments which may override the style-based
 441          * defaults.
 442          */
 443         SetupPositions( p_text_region, psz_subtitle );
 444
 445         p_text_region->p_next = NULL;
 446     }
 447     return p_text_region;
 448 }
 449
 450 static int ParseImageAttachments( decoder_t *p_dec )
 451 {
 452     decoder_sys_t        *p_sys = p_dec->p_sys;
 453     input_attachment_t  **pp_attachments;
 454     int                   i_attachments_cnt;
 455     int                   k = 0;
 456
 457     if( VLC_SUCCESS != decoder_GetInputAttachments( p_dec, &pp_attachments, &i_attachments_cnt ))
 458         return VLC_EGENERIC;
 459
 460     for( k = 0; k < i_attachments_cnt; k++ )
 461     {
 462         input_attachment_t *p_attach = pp_attachments[k];
 463
 464         vlc_fourcc_t type = image_Mime2Fourcc( p_attach->psz_mime );
 465
 466         if( ( type != 0 ) &&
 467             ( p_attach->i_data > 0 ) &&
 468             ( p_attach->p_data != NULL ) )
 469         {
 470             picture_t         *p_pic = NULL;
 471             image_handler_t   *p_image;
 472
 473             p_image = image_HandlerCreate( p_dec );
 474             if( p_image != NULL )
 475             {
 476                 block_t   *p_block;
 477
 478                 p_block = block_New( p_image->p_parent, p_attach->i_data );
 479
 480                 if( p_block != NULL )
 481                 {
 482                     video_format_t     fmt_in;
 483                     video_format_t     fmt_out;
 484
 485                     memcpy( p_block->p_buffer, p_attach->p_data, p_attach->i_data );
 486
 487                     memset( &fmt_in,  0, sizeof( video_format_t));
 488                     memset( &fmt_out, 0, sizeof( video_format_t));
 489
 490                     fmt_in.i_chroma  = type;
 491                     fmt_out.i_chroma = VLC_FOURCC('Y','U','V','A');
 492
 493                     /* Find a suitable decoder module */
 494                     if( module_exists( "sdl_image" ) )
 495                     {
 496                         /* ffmpeg thinks it can handle bmp properly but it can't (at least
 497                          * not all of them), so use sdl_image if it is available */
 498
 499                         vlc_value_t val;
 500
 501                         var_Create( p_dec, "codec", VLC_VAR_MODULE | VLC_VAR_DOINHERIT );
 502                         val.psz_string = (char*) "sdl_image";
 503                         var_Set( p_dec, "codec", val );
 504                     }
 505
 506                     p_pic = image_Read( p_image, p_block, &fmt_in, &fmt_out );
 507                     var_Destroy( p_dec, "codec" );
 508                 }
 509
 510                 image_HandlerDelete( p_image );
 511             }
 512             if( p_pic )
 513             {
 514                 image_attach_t *p_picture = malloc( sizeof(image_attach_t) );
 515
 516                 if( p_picture )
 517                 {
 518                     p_picture->psz_filename = strdup( p_attach->psz_name );
 519                     p_picture->p_pic = p_pic;
 520
 521                     TAB_APPEND( p_sys->i_images, p_sys->pp_images, p_picture );
 522                 }
 523             }
 524         }
 525         vlc_input_attachment_Delete( pp_attachments[ k ] );
 526     }
 527     free( pp_attachments );
 528
 529     return VLC_SUCCESS;
 530 }
 531
 532 static void ParseUSFHeaderTags( decoder_t *p_dec, xml_reader_t *p_xml_reader )
 533 {
 534     decoder_sys_t *p_sys = p_dec->p_sys;
 535     char *psz_node;
 536     ssa_style_t *p_ssa_style = NULL;
 537     int i_style_level = 0;
 538     int i_metadata_level = 0;
 539
 540     while ( xml_ReaderRead( p_xml_reader ) == 1 )
 541     {
 542         switch ( xml_ReaderNodeType( p_xml_reader ) )
 543         {
 544             case XML_READER_TEXT:
 545             case XML_READER_NONE:
 546                 break;
 547             case XML_READER_ENDELEM:
 548                 psz_node = xml_ReaderName( p_xml_reader );
 549
 550                 if( !psz_node )
 551                     break;
 552                 switch (i_style_level)
 553                 {
 554                     case 0:
 555                         if( !strcasecmp( "metadata", psz_node ) && (i_metadata_level == 1) )
 556                         {
 557                             i_metadata_level--;
 558                         }
 559                         break;
 560                     case 1:
 561                         if( !strcasecmp( "styles", psz_node ) )
 562                         {
 563                             i_style_level--;
 564                         }
 565                         break;
 566                     case 2:
 567                         if( !strcasecmp( "style", psz_node ) )
 568                         {
 569                             TAB_APPEND( p_sys->i_ssa_styles, p_sys->pp_ssa_styles, p_ssa_style );
 570
 571                             p_ssa_style = NULL;
 572                             i_style_level--;
 573                         }
 574                         break;
 575                 }
 576
 577                 free( psz_node );
 578                 break;
 579             case XML_READER_STARTELEM:
 580                 psz_node = xml_ReaderName( p_xml_reader );
 581
 582                 if( !psz_node )
 583                     break;
 584
 585                 if( !strcasecmp( "metadata", psz_node ) && (i_style_level == 0) )
 586                 {
 587                     i_metadata_level++;
 588                 }
 589                 else if( !strcasecmp( "resolution", psz_node ) &&
 590                          ( i_metadata_level == 1) )
 591                 {
 592                     while ( xml_ReaderNextAttr( p_xml_reader ) == VLC_SUCCESS )
 593                     {
 594                         char *psz_name = xml_ReaderName ( p_xml_reader );
 595                         char *psz_value = xml_ReaderValue ( p_xml_reader );
 596
 597                         if( psz_name && psz_value )
 598                         {
 599                             if( !strcasecmp( "x", psz_name ) )
 600                                 p_sys->i_original_width = atoi( psz_value );
 601                             else if( !strcasecmp( "y", psz_name ) )
 602                                 p_sys->i_original_height = atoi( psz_value );
 603                         }
 604                         free( psz_name );
 605                         free( psz_value );
 606                     }
 607                 }
 608                 else if( !strcasecmp( "styles", psz_node ) && (i_style_level == 0) )
 609                 {
 610                     i_style_level++;
 611                 }
 612                 else if( !strcasecmp( "style", psz_node ) && (i_style_level == 1) )
 613                 {
 614                     i_style_level++;
 615
 616                     p_ssa_style = calloc( 1, sizeof(ssa_style_t) );
 617                     if( !p_ssa_style )
 618                     {
 619                         free( psz_node );
 620                         return;
 621                     }
 622                     /* All styles are supposed to default to Default, and then
 623                      * one or more settings are over-ridden.
 624                      * At the moment this only effects styles defined AFTER
 625                      * Default in the XML
 626                      */
 627                     int i;
 628                     for( i = 0; i < p_sys->i_ssa_styles; i++ )
 629                     {
 630                         if( !strcasecmp( p_sys->pp_ssa_styles[i]->psz_stylename, "Default" ) )
 631                         {
 632                             ssa_style_t *p_default_style = p_sys->pp_ssa_styles[i];
 633
 634                             memcpy( p_ssa_style, p_default_style, sizeof( ssa_style_t ) );
 635                             //FIXME: Make font_style a pointer. Actually we double copy some data here,
 636                             //   we use text_style_Copy to avoid copying psz_fontname, though .
 637                             text_style_Copy( &p_ssa_style->font_style, &p_default_style->font_style );
 638                             p_ssa_style->psz_stylename = NULL;
 639                         }
 640                     }
 641
 642                     while ( xml_ReaderNextAttr( p_xml_reader ) == VLC_SUCCESS )
 643                     {
 644                         char *psz_name = xml_ReaderName ( p_xml_reader );
 645                         char *psz_value = xml_ReaderValue ( p_xml_reader );
 646
 647                         if( psz_name && psz_value )
 648                         {
 649                             if( !strcasecmp( "name", psz_name ) )
 650                                 p_ssa_style->psz_stylename = strdup( psz_value );
 651                         }
 652                         free( psz_name );
 653                         free( psz_value );
 654                     }
 655                 }
 656                 else if( !strcasecmp( "fontstyle", psz_node ) && (i_style_level == 2) )
 657                 {
 658                     while ( xml_ReaderNextAttr( p_xml_reader ) == VLC_SUCCESS )
 659                     {
 660                         char *psz_name = xml_ReaderName ( p_xml_reader );
 661                         char *psz_value = xml_ReaderValue ( p_xml_reader );
 662
 663                         if( psz_name && psz_value )
 664                         {
 665                             if( !strcasecmp( "face", psz_name ) )
 666                             {
 667                                 free( p_ssa_style->font_style.psz_fontname );
 668                                 p_ssa_style->font_style.psz_fontname = strdup( psz_value );
 669                             }
 670                             else if( !strcasecmp( "size", psz_name ) )
 671                             {
 672                                 if( ( *psz_value == '+' ) || ( *psz_value == '-' ) )
 673                                 {
 674                                     int i_value = atoi( psz_value );
 675
 676                                     if( ( i_value >= -5 ) && ( i_value <= 5 ) )
 677                                         p_ssa_style->font_style.i_font_size  +=
 678                                             ( i_value * p_ssa_style->font_style.i_font_size ) / 10;
 679                                     else if( i_value < -5 )
 680                                         p_ssa_style->font_style.i_font_size  = - i_value;
 681                                     else if( i_value > 5 )
 682                                         p_ssa_style->font_style.i_font_size  = i_value;
 683                                 }
 684                                 else
 685                                     p_ssa_style->font_style.i_font_size  = atoi( psz_value );
 686                             }
 687                             else if( !strcasecmp( "italic", psz_name ) )
 688                             {
 689                                 if( !strcasecmp( "yes", psz_value ))
 690                                     p_ssa_style->font_style.i_style_flags |= STYLE_ITALIC;
 691                                 else
 692                                     p_ssa_style->font_style.i_style_flags &= ~STYLE_ITALIC;
 693                             }
 694                             else if( !strcasecmp( "weight", psz_name ) )
 695                             {
 696                                 if( !strcasecmp( "bold", psz_value ))
 697                                     p_ssa_style->font_style.i_style_flags |= STYLE_BOLD;
 698                                 else
 699                                     p_ssa_style->font_style.i_style_flags &= ~STYLE_BOLD;
 700                             }
 701                             else if( !strcasecmp( "underline", psz_name ) )
 702                             {
 703                                 if( !strcasecmp( "yes", psz_value ))
 704                                     p_ssa_style->font_style.i_style_flags |= STYLE_UNDERLINE;
 705                                 else
 706                                     p_ssa_style->font_style.i_style_flags &= ~STYLE_UNDERLINE;
 707                             }
 708                             else if( !strcasecmp( "color", psz_name ) )
 709                             {
 710                                 if( *psz_value == '#' )
 711                                 {
 712                                     unsigned long col = strtol(psz_value+1, NULL, 16);
 713                                     p_ssa_style->font_style.i_font_color = (col & 0x00ffffff);
 714                                     p_ssa_style->font_style.i_font_alpha = (col >> 24) & 0xff;
 715                                 }
 716                             }
 717                             else if( !strcasecmp( "outline-color", psz_name ) )
 718                             {
 719                                 if( *psz_value == '#' )
 720                                 {
 721                                     unsigned long col = strtol(psz_value+1, NULL, 16);
 722                                     p_ssa_style->font_style.i_outline_color = (col & 0x00ffffff);
 723                                     p_ssa_style->font_style.i_outline_alpha = (col >> 24) & 0xff;
 724                                 }
 725                             }
 726                             else if( !strcasecmp( "outline-level", psz_name ) )
 727                             {
 728                                 p_ssa_style->font_style.i_outline_width = atoi( psz_value );
 729                             }
 730                             else if( !strcasecmp( "shadow-color", psz_name ) )
 731                             {
 732                                 if( *psz_value == '#' )
 733                                 {
 734                                     unsigned long col = strtol(psz_value+1, NULL, 16);
 735                                     p_ssa_style->font_style.i_shadow_color = (col & 0x00ffffff);
 736                                     p_ssa_style->font_style.i_shadow_alpha = (col >> 24) & 0xff;
 737                                 }
 738                             }
 739                             else if( !strcasecmp( "shadow-level", psz_name ) )
 740                             {
 741                                 p_ssa_style->font_style.i_shadow_width = atoi( psz_value );
 742                             }
 743                             else if( !strcasecmp( "back-color", psz_name ) )
 744                             {
 745                                 if( *psz_value == '#' )
 746                                 {
 747                                     unsigned long col = strtol(psz_value+1, NULL, 16);
 748                                     p_ssa_style->font_style.i_karaoke_background_color = (col & 0x00ffffff);
 749                                     p_ssa_style->font_style.i_karaoke_background_alpha = (col >> 24) & 0xff;
 750                                 }
 751                             }
 752                             else if( !strcasecmp( "spacing", psz_name ) )
 753                             {
 754                                 p_ssa_style->font_style.i_spacing = atoi( psz_value );
 755                             }
 756                         }
 757                         free( psz_name );
 758                         free( psz_value );
 759                     }
 760                 }
 761                 else if( !strcasecmp( "position", psz_node ) && (i_style_level == 2) )
 762                 {
 763                     while ( xml_ReaderNextAttr( p_xml_reader ) == VLC_SUCCESS )
 764                     {
 765                         char *psz_name = xml_ReaderName ( p_xml_reader );
 766                         char *psz_value = xml_ReaderValue ( p_xml_reader );
 767
 768                         if( psz_name && psz_value )
 769                         {
 770                             if( !strcasecmp( "alignment", psz_name ) )
 771                             {
 772                                 if( !strcasecmp( "TopLeft", psz_value ) )
 773                                     p_ssa_style->i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_LEFT;
 774                                 else if( !strcasecmp( "TopCenter", psz_value ) )
 775                                     p_ssa_style->i_align = SUBPICTURE_ALIGN_TOP;
 776                                 else if( !strcasecmp( "TopRight", psz_value ) )
 777                                     p_ssa_style->i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_RIGHT;
 778                                 else if( !strcasecmp( "MiddleLeft", psz_value ) )
 779                                     p_ssa_style->i_align = SUBPICTURE_ALIGN_LEFT;
 780                                 else if( !strcasecmp( "MiddleCenter", psz_value ) )
 781                                     p_ssa_style->i_align = 0;
 782                                 else if( !strcasecmp( "MiddleRight", psz_value ) )
 783                                     p_ssa_style->i_align = SUBPICTURE_ALIGN_RIGHT;
 784                                 else if( !strcasecmp( "BottomLeft", psz_value ) )
 785                                     p_ssa_style->i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_LEFT;
 786                                 else if( !strcasecmp( "BottomCenter", psz_value ) )
 787                                     p_ssa_style->i_align = SUBPICTURE_ALIGN_BOTTOM;
 788                                 else if( !strcasecmp( "BottomRight", psz_value ) )
 789                                     p_ssa_style->i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_RIGHT;
 790                             }
 791                             else if( !strcasecmp( "horizontal-margin", psz_name ) )
 792                             {
 793                                 if( strchr( psz_value, '%' ) )
 794                                 {
 795                                     p_ssa_style->i_margin_h = 0;
 796                                     p_ssa_style->i_margin_percent_h = atoi( psz_value );
 797                                 }
 798                                 else
 799                                 {
 800                                     p_ssa_style->i_margin_h = atoi( psz_value );
 801                                     p_ssa_style->i_margin_percent_h = 0;
 802                                 }
 803                             }
 804                             else if( !strcasecmp( "vertical-margin", psz_name ) )
 805                             {
 806                                 if( strchr( psz_value, '%' ) )
 807                                 {
 808                                     p_ssa_style->i_margin_v = 0;
 809                                     p_ssa_style->i_margin_percent_v = atoi( psz_value );
 810                                 }
 811                                 else
 812                                 {
 813                                     p_ssa_style->i_margin_v = atoi( psz_value );
 814                                     p_ssa_style->i_margin_percent_v = 0;
 815                                 }
 816                             }
 817                         }
 818                         free( psz_name );
 819                         free( psz_value );
 820                     }
 821                 }
 822
 823                 free( psz_node );
 824                 break;
 825         }
 826     }
 827     free( p_ssa_style );
 828 }
 829
 830
 831
 832 static subpicture_region_t *ParseUSFString( decoder_t *p_dec,
 833                                             char *psz_subtitle,
 834                                             subpicture_t *p_spu_in )
 835 {
 836     decoder_sys_t        *p_sys = p_dec->p_sys;
 837     subpicture_t         *p_spu = p_spu_in;
 838     subpicture_region_t  *p_region_first = NULL;
 839     subpicture_region_t  *p_region_upto  = p_region_first;
 840
 841     while( *psz_subtitle )
 842     {
 843         if( *psz_subtitle == '<' )
 844         {
 845             char *psz_end = NULL;
 846
 847             if(( !strncasecmp( psz_subtitle, "<text ", 6 )) ||
 848                ( !strncasecmp( psz_subtitle, "<text>", 6 )))
 849             {
 850                 psz_end = strcasestr( psz_subtitle, "</text>" );
 851
 852                 if( psz_end )
 853                 {
 854                     subpicture_region_t  *p_text_region;
 855
 856                     psz_end += strcspn( psz_end, ">" ) + 1;
 857
 858                     p_text_region = CreateTextRegion( p_dec,
 859                                                       p_spu,
 860                                                       psz_subtitle,
 861                                                       psz_end - psz_subtitle,
 862                                                       p_sys->i_align );
 863
 864                     if( p_text_region )
 865                     {
 866                         p_text_region->psz_text = CreatePlainText( p_text_region->psz_html );
 867
 868                         if( ! var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 869                         {
 870                             free( p_text_region->psz_html );
 871                             p_text_region->psz_html = NULL;
 872                         }
 873                     }
 874
 875                     if( !p_region_first )
 876                     {
 877                         p_region_first = p_region_upto = p_text_region;
 878                     }
 879                     else if( p_text_region )
 880                     {
 881                         p_region_upto->p_next = p_text_region;
 882                         p_region_upto = p_region_upto->p_next;
 883                     }
 884                 }
 885             }
 886             else if(( !strncasecmp( psz_subtitle, "<karaoke ", 9 )) ||
 887                     ( !strncasecmp( psz_subtitle, "<karaoke>", 9 )))
 888             {
 889                 psz_end = strcasestr( psz_subtitle, "</karaoke>" );
 890
 891                 if( psz_end )
 892                 {
 893                     subpicture_region_t  *p_text_region;
 894
 895                     psz_end += strcspn( psz_end, ">" ) + 1;
 896
 897                     p_text_region = CreateTextRegion( p_dec,
 898                                                       p_spu,
 899                                                       psz_subtitle,
 900                                                       psz_end - psz_subtitle,
 901                                                       p_sys->i_align );
 902
 903                     if( p_text_region )
 904                     {
 905                         if( ! var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 906                         {
 907                             free( p_text_region->psz_html );
 908                             p_text_region->psz_html = NULL;
 909                         }
 910                     }
 911                     if( !p_region_first )
 912                     {
 913                         p_region_first = p_region_upto = p_text_region;
 914                     }
 915                     else if( p_text_region )
 916                     {
 917                         p_region_upto->p_next = p_text_region;
 918                         p_region_upto = p_region_upto->p_next;
 919                     }
 920                 }
 921             }
 922             else if(( !strncasecmp( psz_subtitle, "<image ", 7 )) ||
 923                     ( !strncasecmp( psz_subtitle, "<image>", 7 )))
 924             {
 925                 subpicture_region_t *p_image_region = NULL;
 926
 927                 char *psz_end = strcasestr( psz_subtitle, "</image>" );
 928                 char *psz_content = strchr( psz_subtitle, '>' );
 929                 int   i_transparent = -1;
 930
 931                 /* If a colorkey parameter is specified, then we have to map
 932                  * that index in the picture through as transparent (it is
 933                  * required by the USF spec but is also recommended that if the
 934                  * creator really wants a transparent colour that they use a
 935                  * type like PNG that properly supports it; this goes doubly
 936                  * for VLC because the pictures are stored internally in YUV
 937                  * and the resulting colour-matching may not produce the
 938                  * desired results.)
 939                  */
 940                 char *psz_tmp = GrabAttributeValue( "colorkey", psz_subtitle );
 941                 if( psz_tmp )
 942                 {
 943                     if( *psz_tmp == '#' )
 944                         i_transparent = strtol( psz_tmp + 1, NULL, 16 ) & 0x00ffffff;
 945                     free( psz_tmp );
 946                 }
 947                 if( psz_content && ( psz_content < psz_end ) )
 948                 {
 949                     char *psz_filename = strndup( &psz_content[1], psz_end - &psz_content[1] );
 950                     if( psz_filename )
 951                     {
 952                         p_image_region = LoadEmbeddedImage( p_dec, p_spu,
 953                                             psz_filename, i_transparent );
 954                         free( psz_filename );
 955                     }
 956                 }
 957
 958                 if( psz_end ) psz_end += strcspn( psz_end, ">" ) + 1;
 959
 960                 if( p_image_region )
 961                 {
 962                     SetupPositions( p_image_region, psz_subtitle );
 963
 964                     p_image_region->p_next   = NULL;
 965                     p_image_region->psz_text = NULL;
 966                     p_image_region->psz_html = NULL;
 967
 968                 }
 969                 if( !p_region_first )
 970                 {
 971                     p_region_first = p_region_upto = p_image_region;
 972                 }
 973                 else if( p_image_region )
 974                 {
 975                     p_region_upto->p_next = p_image_region;
 976                     p_region_upto = p_region_upto->p_next;
 977                 }
 978             }
 979             if( psz_end )
 980                 psz_subtitle = psz_end - 1;
 981
 982             psz_subtitle += strcspn( psz_subtitle, ">" );
 983         }
 984
 985         psz_subtitle++;
 986     }
 987
 988     return p_region_first;
 989 }
 990
 991 /*****************************************************************************
 992  * ParseUSFHeader: Retrieve global formatting information etc
 993  *****************************************************************************/
 994 static void ParseUSFHeader( decoder_t *p_dec )
 995 {
 996     stream_t      *p_sub = NULL;
 997     xml_t         *p_xml = NULL;
 998     xml_reader_t  *p_xml_reader = NULL;
 999
1000     p_sub = stream_MemoryNew( VLC_OBJECT(p_dec),
1001                               p_dec->fmt_in.p_extra,
1002                               p_dec->fmt_in.i_extra,
1003                               true );
1004     if( !p_sub )
1005         return;
1006
1007     p_xml = xml_Create( p_dec );
1008     if( p_xml )
1009     {
1010         p_xml_reader = xml_ReaderCreate( p_xml, p_sub );
1011         if( p_xml_reader )
1012         {
1013             /* Look for Root Node */
1014             if( xml_ReaderRead( p_xml_reader ) == 1 )
1015             {
1016                 char *psz_node = xml_ReaderName( p_xml_reader );
1017
1018                 if( !strcasecmp( "usfsubtitles", psz_node ) )
1019                     ParseUSFHeaderTags( p_dec, p_xml_reader );
1020
1021                 free( psz_node );
1022             }
1023
1024             xml_ReaderDelete( p_xml, p_xml_reader );
1025         }
1026         xml_Delete( p_xml );
1027     }
1028     stream_Delete( p_sub );
1029 }
1030
1031 /* Function now handles tags which has attribute values, and tries
1032  * to deal with &' commands too. It no longer modifies the string
1033  * in place, so that the original text can be reused
1034  */
1035 static char *StripTags( char *psz_subtitle )
1036 {
1037     char *psz_text_start;
1038     char *psz_text;
1039
1040     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
1041     if( !psz_text_start )
1042         return NULL;
1043
1044     while( *psz_subtitle )
1045     {
1046         /* Mask out any pre-existing LFs in the subtitle */
1047         if( *psz_subtitle == '\n' )
1048             *psz_subtitle = ' ';
1049
1050         if( *psz_subtitle == '<' )
1051         {
1052             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
1053                 *psz_text++ = '\n';
1054
1055             psz_subtitle += strcspn( psz_subtitle, ">" );
1056         }
1057         else if( *psz_subtitle == '&' )
1058         {
1059             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
1060             {
1061                 *psz_text++ = '<';
1062                 psz_subtitle += strcspn( psz_subtitle, ";" );
1063             }
1064             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
1065             {
1066                 *psz_text++ = '>';
1067                 psz_subtitle += strcspn( psz_subtitle, ";" );
1068             }
1069             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
1070             {
1071                 *psz_text++ = '&';
1072                 psz_subtitle += strcspn( psz_subtitle, ";" );
1073             }
1074             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
1075             {
1076                 *psz_text++ = '\"';
1077                 psz_subtitle += strcspn( psz_subtitle, ";" );
1078             }
1079             else
1080             {
1081                 /* Assume it is just a normal ampersand */
1082                 *psz_text++ = '&';
1083             }
1084         }
1085         else
1086         {
1087             *psz_text++ = *psz_subtitle;
1088         }
1089
1090         psz_subtitle++;
1091     }
1092     *psz_text = '\0';
1093     psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
1094
1095     return psz_text_start;
1096 }
1097
1098 /* Turn a HTML subtitle, turn into a plain-text version,
1099  *  complete with sensible whitespace compaction
1100  */
1101
1102 static char *CreatePlainText( char *psz_subtitle )
1103 {
1104     char *psz_text = StripTags( psz_subtitle );
1105     char *s;
1106
1107     if( !psz_text )
1108         return NULL;
1109
1110     s = strpbrk( psz_text, "\t\r\n " );
1111     while( s )
1112     {
1113         int   k;
1114         char  spc = ' ';
1115         int   i_whitespace = strspn( s, "\t\r\n " );
1116
1117         /* Favour '\n' over other whitespaces - if one of these
1118          * occurs in the whitespace use a '\n' as our value,
1119          * otherwise just use a ' '
1120          */
1121         for( k = 0; k < i_whitespace; k++ )
1122             if( s[k] == '\n' ) spc = '\n';
1123
1124         if( i_whitespace > 1 )
1125         {
1126             memmove( &s[1],
1127                      &s[i_whitespace],
1128                      strlen( s ) - i_whitespace + 1 );
1129         }
1130         *s++ = spc;
1131
1132         s = strpbrk( s, "\t\r\n " );
1133     }
1134     return psz_text;
1135 }
1136
1137 /****************************************************************************
1138  * download and resize image located at psz_url
1139  ***************************************************************************/
1140 static subpicture_region_t *LoadEmbeddedImage( decoder_t *p_dec,
1141                                                subpicture_t *p_spu,
1142                                                const char *psz_filename,
1143                                                int i_transparent_color )
1144 {
1145     decoder_sys_t         *p_sys = p_dec->p_sys;
1146     subpicture_region_t   *p_region;
1147     video_format_t         fmt_out;
1148     int                    k;
1149     picture_t             *p_pic = NULL;
1150
1151     for( k = 0; k < p_sys->i_images; k++ )
1152     {
1153         if( p_sys->pp_images &&
1154             !strcmp( p_sys->pp_images[k]->psz_filename, psz_filename ) )
1155         {
1156             p_pic = p_sys->pp_images[k]->p_pic;
1157             break;
1158         }
1159     }
1160
1161     if( !p_pic )
1162     {
1163         msg_Err( p_dec, "Unable to read image %s", psz_filename );
1164         return NULL;
1165     }
1166
1167     /* Display the feed's image */
1168     memset( &fmt_out, 0, sizeof( video_format_t));
1169
1170     fmt_out.i_chroma = VLC_FOURCC('Y','U','V','A');
1171     fmt_out.i_aspect = VOUT_ASPECT_FACTOR;
1172     fmt_out.i_sar_num = fmt_out.i_sar_den = 1;
1173     fmt_out.i_width =
1174         fmt_out.i_visible_width = p_pic->format.i_visible_width;
1175     fmt_out.i_height =
1176         fmt_out.i_visible_height = p_pic->format.i_visible_height;
1177
1178     p_region = subpicture_region_New( &fmt_out );
1179     if( !p_region )
1180     {
1181         msg_Err( p_dec, "cannot allocate SPU region" );
1182         return NULL;
1183     }
1184     assert( p_pic->format.i_chroma == VLC_FOURCC('Y','U','V','A') );
1185     /* FIXME the copy is probably not needed anymore */
1186     picture_CopyPixels( p_region->p_picture, p_pic );
1187
1188     /* This isn't the best way to do this - if you really want transparency, then
1189      * you're much better off using an image type that supports it like PNG. The
1190      * spec requires this support though.
1191      */
1192     if( i_transparent_color > 0 )
1193     {
1194         int i_r = ( i_transparent_color >> 16 ) & 0xff;
1195         int i_g = ( i_transparent_color >>  8 ) & 0xff;
1196         int i_b = ( i_transparent_color       ) & 0xff;
1197
1198         /* FIXME it cannot work as the yuv conversion code will probably NOT match
1199          * this one  */
1200         int i_y = ( ( (  66 * i_r + 129 * i_g +  25 * i_b + 128 ) >> 8 ) + 16 );
1201         int i_u =   ( ( -38 * i_r -  74 * i_g + 112 * i_b + 128 ) >> 8 ) + 128 ;
1202         int i_v =   ( ( 112 * i_r -  94 * i_g -  18 * i_b + 128 ) >> 8 ) + 128 ;
1203
1204         assert( p_region->fmt.i_chroma == VLC_FOURCC('Y','U','V','A') );
1205         for( unsigned int y = 0; y < p_region->fmt.i_height; y++ )
1206         {
1207             for( unsigned int x = 0; x < p_region->fmt.i_width; x++ )
1208             {
1209                 if( p_region->p_picture->Y_PIXELS[y*p_region->p_picture->Y_PITCH + x] != i_y ||
1210                     p_region->p_picture->U_PIXELS[y*p_region->p_picture->U_PITCH + x] != i_u ||
1211                     p_region->p_picture->V_PIXELS[y*p_region->p_picture->V_PITCH + x] != i_v )
1212                     continue;
1213                 p_region->p_picture->A_PIXELS[y*p_region->p_picture->A_PITCH + x] = 0;
1214
1215             }
1216         }
1217     }
1218     return p_region;
1219 }