git.sesse.net Git - vlc/blob - modules/codec/subsusf.c

   1 /*****************************************************************************
   2  * subsusf.c : USF subtitles decoder
   3  *****************************************************************************
   4  * Copyright (C) 2000-2006 VLC authors and VideoLAN
   5  * $Id$
   6  *
   7  * Authors: Bernie Purcell <bitmap@videolan.org>
   8  *
   9  * This program is free software; you can redistribute it and/or modify it
  10  * under the terms of the GNU Lesser General Public License as published by
  11  * the Free Software Foundation; either version 2.1 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 Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public License
  20  * along with this program; if not, write to the Free Software Foundation,
  21  * Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
  22  *****************************************************************************/
  23 #ifdef HAVE_CONFIG_H
  24 # include "config.h"
  25 #endif
  26 #include <assert.h>
  27
  28 #include <vlc_common.h>
  29 #include <vlc_plugin.h>
  30 #include <vlc_modules.h>
  31 #include <vlc_codec.h>
  32 #include <vlc_input.h>
  33 #include <vlc_charset.h>
  34 #include <vlc_image.h>
  35 #include <vlc_xml.h>
  36 #include <vlc_stream.h>
  37
  38 /*****************************************************************************
  39  * Module descriptor.
  40  *****************************************************************************/
  41 static int  OpenDecoder   ( vlc_object_t * );
  42 static void CloseDecoder  ( vlc_object_t * );
  43
  44 vlc_module_begin ()
  45     set_capability( "decoder", 40 )
  46     set_shortname( N_("USFSubs"))
  47     set_description( N_("USF subtitles decoder") )
  48     set_callbacks( OpenDecoder, CloseDecoder )
  49     set_category( CAT_INPUT )
  50     set_subcategory( SUBCAT_INPUT_SCODEC )
  51     /* We inherit subsdec-align and subsdec-formatted from subsdec.c */
  52 vlc_module_end ()
  53
  54
  55 /*****************************************************************************
  56  * Local prototypes
  57  *****************************************************************************/
  58 enum
  59 {
  60     ATTRIBUTE_ALIGNMENT = (1 << 0),
  61     ATTRIBUTE_X         = (1 << 1),
  62     ATTRIBUTE_X_PERCENT = (1 << 2),
  63     ATTRIBUTE_Y         = (1 << 3),
  64     ATTRIBUTE_Y_PERCENT = (1 << 4),
  65 };
  66
  67 typedef struct
  68 {
  69     char       *psz_filename;
  70     picture_t  *p_pic;
  71 } image_attach_t;
  72
  73 typedef struct
  74 {
  75     char *          psz_stylename; /* The name of the style, no comma's allowed */
  76     text_style_t    font_style;
  77     int             i_align;
  78     int             i_margin_h;
  79     int             i_margin_v;
  80     int             i_margin_percent_h;
  81     int             i_margin_percent_v;
  82 }  ssa_style_t;
  83
  84 struct decoder_sys_t
  85 {
  86     int                 i_original_height;
  87     int                 i_original_width;
  88     int                 i_align;          /* Subtitles alignment on the vout */
  89
  90     ssa_style_t         **pp_ssa_styles;
  91     int                 i_ssa_styles;
  92
  93     image_attach_t      **pp_images;
  94     int                 i_images;
  95 };
  96
  97 static subpicture_t *DecodeBlock   ( decoder_t *, block_t ** );
  98 static char         *CreatePlainText( char * );
  99 static int           ParseImageAttachments( decoder_t *p_dec );
 100
 101 static subpicture_t        *ParseText     ( decoder_t *, block_t * );
 102 static void                 ParseUSFHeader( decoder_t * );
 103 static subpicture_region_t *ParseUSFString( decoder_t *, char * );
 104 static subpicture_region_t *LoadEmbeddedImage( decoder_t *p_dec, const char *psz_filename, int i_transparent_color );
 105
 106 /*****************************************************************************
 107  * OpenDecoder: probe the decoder and return score
 108  *****************************************************************************
 109  * Tries to launch a decoder and return score so that the interface is able
 110  * to chose.
 111  *****************************************************************************/
 112 static int OpenDecoder( vlc_object_t *p_this )
 113 {
 114     decoder_t     *p_dec = (decoder_t*)p_this;
 115     decoder_sys_t *p_sys;
 116
 117     if( p_dec->fmt_in.i_codec != VLC_CODEC_USF )
 118         return VLC_EGENERIC;
 119
 120     /* Allocate the memory needed to store the decoder's structure */
 121     if( ( p_dec->p_sys = p_sys = calloc(1, sizeof(decoder_sys_t)) ) == NULL )
 122         return VLC_ENOMEM;
 123
 124     p_dec->pf_decode_sub = DecodeBlock;
 125     p_dec->fmt_out.i_cat = SPU_ES;
 126     p_dec->fmt_out.i_codec = 0;
 127
 128     /* init of p_sys */
 129     TAB_INIT( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 130     TAB_INIT( p_sys->i_images, p_sys->pp_images );
 131
 132     /* USF subtitles are mandated to be UTF-8, so don't need vlc_iconv */
 133
 134     p_sys->i_align = var_CreateGetInteger( p_dec, "subsdec-align" );
 135
 136     ParseImageAttachments( p_dec );
 137
 138     if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 139     {
 140         if( p_dec->fmt_in.i_extra > 0 )
 141             ParseUSFHeader( p_dec );
 142     }
 143
 144     return VLC_SUCCESS;
 145 }
 146
 147 /****************************************************************************
 148  * DecodeBlock: the whole thing
 149  ****************************************************************************
 150  * This function must be fed with complete subtitles units.
 151  ****************************************************************************/
 152 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 153 {
 154     subpicture_t *p_spu;
 155     block_t *p_block;
 156
 157     if( !pp_block || *pp_block == NULL )
 158         return NULL;
 159
 160     p_block = *pp_block;
 161
 162     p_spu = ParseText( p_dec, p_block );
 163
 164     block_Release( p_block );
 165     *pp_block = NULL;
 166
 167     return p_spu;
 168 }
 169
 170 /*****************************************************************************
 171  * CloseDecoder: clean up the decoder
 172  *****************************************************************************/
 173 static void CloseDecoder( vlc_object_t *p_this )
 174 {
 175     decoder_t *p_dec = (decoder_t *)p_this;
 176     decoder_sys_t *p_sys = p_dec->p_sys;
 177
 178     if( p_sys->pp_ssa_styles )
 179     {
 180         int i;
 181         for( i = 0; i < p_sys->i_ssa_styles; i++ )
 182         {
 183             if( !p_sys->pp_ssa_styles[i] )
 184                 continue;
 185
 186             free( p_sys->pp_ssa_styles[i]->psz_stylename );
 187             //FIXME: Make font_style a pointer and use text_style_* functions
 188             free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
 189             free( p_sys->pp_ssa_styles[i] );
 190         }
 191         TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 192     }
 193     if( p_sys->pp_images )
 194     {
 195         int i;
 196         for( i = 0; i < p_sys->i_images; i++ )
 197         {
 198             if( !p_sys->pp_images[i] )
 199                 continue;
 200
 201             if( p_sys->pp_images[i]->p_pic )
 202                 picture_Release( p_sys->pp_images[i]->p_pic );
 203             free( p_sys->pp_images[i]->psz_filename );
 204
 205             free( p_sys->pp_images[i] );
 206         }
 207         TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
 208     }
 209
 210     free( p_sys );
 211 }
 212
 213 /*****************************************************************************
 214  * ParseText: parse an text subtitle packet and send it to the video output
 215  *****************************************************************************/
 216 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 217 {
 218     decoder_sys_t *p_sys = p_dec->p_sys;
 219     subpicture_t *p_spu = NULL;
 220     char *psz_subtitle = NULL;
 221
 222     /* We cannot display a subpicture with no date */
 223     if( p_block->i_pts <= VLC_TS_INVALID )
 224     {
 225         msg_Warn( p_dec, "subtitle without a date" );
 226         return NULL;
 227     }
 228
 229     /* Check validity of packet data */
 230     /* An "empty" line containing only \0 can be used to force
 231        and ephemer picture from the screen */
 232     if( p_block->i_buffer < 1 )
 233     {
 234         msg_Warn( p_dec, "no subtitle data" );
 235         return NULL;
 236     }
 237
 238     /* Should be resiliant against bad subtitles */
 239     psz_subtitle = strndup( (const char *)p_block->p_buffer,
 240                             p_block->i_buffer );
 241     if( psz_subtitle == NULL )
 242         return NULL;
 243
 244     /* USF Subtitles are mandated to be UTF-8 -- make sure it is */
 245     if (EnsureUTF8( psz_subtitle ) == NULL)
 246     {
 247         msg_Err( p_dec, "USF subtitles must be in UTF-8 format.\n"
 248                  "This stream contains USF subtitles which aren't." );
 249     }
 250
 251     /* Create the subpicture unit */
 252     p_spu = decoder_NewSubpicture( p_dec, NULL );
 253     if( !p_spu )
 254     {
 255         msg_Warn( p_dec, "can't get spu buffer" );
 256         free( psz_subtitle );
 257         return NULL;
 258     }
 259
 260     /* Decode USF strings */
 261     p_spu->p_region = ParseUSFString( p_dec, psz_subtitle );
 262
 263     p_spu->i_start = p_block->i_pts;
 264     p_spu->i_stop = p_block->i_pts + p_block->i_length;
 265     p_spu->b_ephemer = (p_block->i_length == 0);
 266     p_spu->b_absolute = false;
 267     p_spu->i_original_picture_width = p_sys->i_original_width;
 268     p_spu->i_original_picture_height = p_sys->i_original_height;
 269
 270     free( psz_subtitle );
 271
 272     return p_spu;
 273 }
 274
 275 static char *GrabAttributeValue( const char *psz_attribute,
 276                                  const char *psz_tag_start )
 277 {
 278     if( psz_attribute && psz_tag_start )
 279     {
 280         char *psz_tag_end = strchr( psz_tag_start, '>' );
 281         char *psz_found   = strcasestr( psz_tag_start, psz_attribute );
 282
 283         if( psz_found )
 284         {
 285             psz_found += strlen( psz_attribute );
 286
 287             if(( *(psz_found++) == '=' ) &&
 288                ( *(psz_found++) == '\"' ))
 289             {
 290                 if( psz_found < psz_tag_end )
 291                 {
 292                     int   i_len = strcspn( psz_found, "\"" );
 293                     return strndup( psz_found, i_len );
 294                 }
 295             }
 296         }
 297     }
 298     return NULL;
 299 }
 300
 301 static ssa_style_t *ParseStyle( decoder_sys_t *p_sys, char *psz_subtitle )
 302 {
 303     ssa_style_t *p_ssa_style = NULL;
 304     char        *psz_style = GrabAttributeValue( "style", psz_subtitle );
 305
 306     if( psz_style )
 307     {
 308         int i;
 309
 310         for( i = 0; i < p_sys->i_ssa_styles; i++ )
 311         {
 312             if( !strcmp( p_sys->pp_ssa_styles[i]->psz_stylename, psz_style ) )
 313                 p_ssa_style = p_sys->pp_ssa_styles[i];
 314         }
 315         free( psz_style );
 316     }
 317     return p_ssa_style;
 318 }
 319
 320 static int ParsePositionAttributeList( char *psz_subtitle, int *i_align,
 321                                        int *i_x, int *i_y )
 322 {
 323     int   i_mask = 0;
 324
 325     char *psz_align    = GrabAttributeValue( "alignment", psz_subtitle );
 326     char *psz_margin_x = GrabAttributeValue( "horizontal-margin", psz_subtitle );
 327     char *psz_margin_y = GrabAttributeValue( "vertical-margin", psz_subtitle );
 328     /* -- UNSUPPORTED
 329     char *psz_relative = GrabAttributeValue( "relative-to", psz_subtitle );
 330     char *psz_rotate_x = GrabAttributeValue( "rotate-x", psz_subtitle );
 331     char *psz_rotate_y = GrabAttributeValue( "rotate-y", psz_subtitle );
 332     char *psz_rotate_z = GrabAttributeValue( "rotate-z", psz_subtitle );
 333     */
 334
 335     *i_align = SUBPICTURE_ALIGN_BOTTOM;
 336     *i_x = 0;
 337     *i_y = 0;
 338
 339     if( psz_align )
 340     {
 341         if( !strcasecmp( "TopLeft", psz_align ) )
 342             *i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_LEFT;
 343         else if( !strcasecmp( "TopCenter", psz_align ) )
 344             *i_align = SUBPICTURE_ALIGN_TOP;
 345         else if( !strcasecmp( "TopRight", psz_align ) )
 346             *i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_RIGHT;
 347         else if( !strcasecmp( "MiddleLeft", psz_align ) )
 348             *i_align = SUBPICTURE_ALIGN_LEFT;
 349         else if( !strcasecmp( "MiddleCenter", psz_align ) )
 350             *i_align = 0;
 351         else if( !strcasecmp( "MiddleRight", psz_align ) )
 352             *i_align = SUBPICTURE_ALIGN_RIGHT;
 353         else if( !strcasecmp( "BottomLeft", psz_align ) )
 354             *i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_LEFT;
 355         else if( !strcasecmp( "BottomCenter", psz_align ) )
 356             *i_align = SUBPICTURE_ALIGN_BOTTOM;
 357         else if( !strcasecmp( "BottomRight", psz_align ) )
 358             *i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_RIGHT;
 359
 360         i_mask |= ATTRIBUTE_ALIGNMENT;
 361         free( psz_align );
 362     }
 363     if( psz_margin_x )
 364     {
 365         *i_x = atoi( psz_margin_x );
 366         if( strchr( psz_margin_x, '%' ) )
 367             i_mask |= ATTRIBUTE_X_PERCENT;
 368         else
 369             i_mask |= ATTRIBUTE_X;
 370
 371         free( psz_margin_x );
 372     }
 373     if( psz_margin_y )
 374     {
 375         *i_y = atoi( psz_margin_y );
 376         if( strchr( psz_margin_y, '%' ) )
 377             i_mask |= ATTRIBUTE_Y_PERCENT;
 378         else
 379             i_mask |= ATTRIBUTE_Y;
 380
 381         free( psz_margin_y );
 382     }
 383     return i_mask;
 384 }
 385
 386 static void SetupPositions( subpicture_region_t *p_region, char *psz_subtitle )
 387 {
 388     int           i_mask = 0;
 389     int           i_align;
 390     int           i_x, i_y;
 391
 392     i_mask = ParsePositionAttributeList( psz_subtitle, &i_align, &i_x, &i_y );
 393
 394     if( i_mask & ATTRIBUTE_ALIGNMENT )
 395         p_region->i_align = i_align;
 396
 397     /* TODO: Setup % based offsets properly, without adversely affecting
 398      *       everything else in vlc. Will address with separate patch, to
 399      *       prevent this one being any more complicated.
 400      */
 401     if( i_mask & ATTRIBUTE_X )
 402         p_region->i_x = i_x;
 403     else if( i_mask & ATTRIBUTE_X_PERCENT )
 404         p_region->i_x = 0;
 405
 406     if( i_mask & ATTRIBUTE_Y )
 407         p_region->i_y = i_y;
 408     else if( i_mask & ATTRIBUTE_Y_PERCENT )
 409         p_region->i_y = 0;
 410 }
 411
 412 static subpicture_region_t *CreateTextRegion( decoder_t *p_dec,
 413                                               char *psz_subtitle,
 414                                               int i_len,
 415                                               int i_sys_align )
 416 {
 417     decoder_sys_t        *p_sys = p_dec->p_sys;
 418     subpicture_region_t  *p_text_region;
 419     video_format_t        fmt;
 420
 421     /* Create a new subpicture region */
 422     memset( &fmt, 0, sizeof(video_format_t) );
 423     fmt.i_chroma = VLC_CODEC_TEXT;
 424     fmt.i_width = fmt.i_height = 0;
 425     fmt.i_x_offset = fmt.i_y_offset = 0;
 426     p_text_region = subpicture_region_New( &fmt );
 427
 428     if( p_text_region != NULL )
 429     {
 430         ssa_style_t  *p_ssa_style = NULL;
 431
 432         p_text_region->psz_text = NULL;
 433         p_text_region->psz_html = strndup( psz_subtitle, i_len );
 434         if( ! p_text_region->psz_html )
 435         {
 436             subpicture_region_Delete( p_text_region );
 437             return NULL;
 438         }
 439
 440         p_ssa_style = ParseStyle( p_sys, p_text_region->psz_html );
 441         if( !p_ssa_style )
 442         {
 443             int i;
 444
 445             for( i = 0; i < p_sys->i_ssa_styles; i++ )
 446             {
 447                 if( !strcasecmp( p_sys->pp_ssa_styles[i]->psz_stylename, "Default" ) )
 448                     p_ssa_style = p_sys->pp_ssa_styles[i];
 449             }
 450         }
 451
 452         if( p_ssa_style )
 453         {
 454             msg_Dbg( p_dec, "style is: %s", p_ssa_style->psz_stylename );
 455
 456             p_text_region->p_style = text_style_Duplicate( &p_ssa_style->font_style );
 457             p_text_region->i_align = p_ssa_style->i_align;
 458
 459             /* TODO: Setup % based offsets properly, without adversely affecting
 460              *       everything else in vlc. Will address with separate patch,
 461              *       to prevent this one being any more complicated.
 462
 463                      * p_ssa_style->i_margin_percent_h;
 464                      * p_ssa_style->i_margin_percent_v;
 465              */
 466             p_text_region->i_x         = p_ssa_style->i_margin_h;
 467             p_text_region->i_y         = p_ssa_style->i_margin_v;
 468
 469         }
 470         else
 471         {
 472             p_text_region->i_align = SUBPICTURE_ALIGN_BOTTOM | i_sys_align;
 473             p_text_region->i_x = i_sys_align ? 20 : 0;
 474             p_text_region->i_y = 10;
 475         }
 476         /* Look for position arguments which may override the style-based
 477          * defaults.
 478          */
 479         SetupPositions( p_text_region, psz_subtitle );
 480
 481         p_text_region->p_next = NULL;
 482     }
 483     return p_text_region;
 484 }
 485
 486 static int ParseImageAttachments( decoder_t *p_dec )
 487 {
 488     decoder_sys_t        *p_sys = p_dec->p_sys;
 489     input_attachment_t  **pp_attachments;
 490     int                   i_attachments_cnt;
 491     int                   k = 0;
 492
 493     if( VLC_SUCCESS != decoder_GetInputAttachments( p_dec, &pp_attachments, &i_attachments_cnt ))
 494         return VLC_EGENERIC;
 495
 496     for( k = 0; k < i_attachments_cnt; k++ )
 497     {
 498         input_attachment_t *p_attach = pp_attachments[k];
 499
 500         vlc_fourcc_t type = image_Mime2Fourcc( p_attach->psz_mime );
 501
 502         if( ( type != 0 ) &&
 503             ( p_attach->i_data > 0 ) &&
 504             ( p_attach->p_data != NULL ) )
 505         {
 506             picture_t         *p_pic = NULL;
 507             image_handler_t   *p_image;
 508
 509             p_image = image_HandlerCreate( p_dec );
 510             if( p_image != NULL )
 511             {
 512                 block_t   *p_block;
 513
 514                 p_block = block_Alloc( p_attach->i_data );
 515
 516                 if( p_block != NULL )
 517                 {
 518                     video_format_t     fmt_in;
 519                     video_format_t     fmt_out;
 520
 521                     memcpy( p_block->p_buffer, p_attach->p_data, p_attach->i_data );
 522
 523                     memset( &fmt_in,  0, sizeof( video_format_t));
 524                     memset( &fmt_out, 0, sizeof( video_format_t));
 525
 526                     fmt_in.i_chroma  = type;
 527                     fmt_out.i_chroma = VLC_CODEC_YUVA;
 528
 529                     /* Find a suitable decoder module */
 530                     if( module_exists( "sdl_image" ) )
 531                     {
 532                         /* ffmpeg thinks it can handle bmp properly but it can't (at least
 533                          * not all of them), so use sdl_image if it is available */
 534
 535                         var_Create( p_dec, "codec", VLC_VAR_STRING | VLC_VAR_DOINHERIT );
 536                         var_SetString( p_dec, "codec", "sdl_image" );
 537                     }
 538
 539                     p_pic = image_Read( p_image, p_block, &fmt_in, &fmt_out );
 540                     var_Destroy( p_dec, "codec" );
 541                 }
 542
 543                 image_HandlerDelete( p_image );
 544             }
 545             if( p_pic )
 546             {
 547                 image_attach_t *p_picture = malloc( sizeof(image_attach_t) );
 548
 549                 if( p_picture )
 550                 {
 551                     p_picture->psz_filename = strdup( p_attach->psz_name );
 552                     p_picture->p_pic = p_pic;
 553
 554                     TAB_APPEND( p_sys->i_images, p_sys->pp_images, p_picture );
 555                 }
 556             }
 557         }
 558         vlc_input_attachment_Delete( pp_attachments[ k ] );
 559     }
 560     free( pp_attachments );
 561
 562     return VLC_SUCCESS;
 563 }
 564
 565 static void ParseUSFHeaderTags( decoder_t *p_dec, xml_reader_t *p_xml_reader )
 566 {
 567     decoder_sys_t *p_sys = p_dec->p_sys;
 568     const char *node;
 569     ssa_style_t *p_ssa_style = NULL;
 570     int i_style_level = 0;
 571     int i_metadata_level = 0;
 572     int type;
 573
 574     while( (type = xml_ReaderNextNode( p_xml_reader, &node )) > 0 )
 575     {
 576         switch( type )
 577         {
 578             case XML_READER_ENDELEM:
 579                 switch (i_style_level)
 580                 {
 581                     case 0:
 582                         if( !strcasecmp( "metadata", node ) && (i_metadata_level == 1) )
 583                             i_metadata_level--;
 584                         break;
 585                     case 1:
 586                         if( !strcasecmp( "styles", node ) )
 587                             i_style_level--;
 588                         break;
 589                     case 2:
 590                         if( !strcasecmp( "style", node ) )
 591                         {
 592                             TAB_APPEND( p_sys->i_ssa_styles, p_sys->pp_ssa_styles, p_ssa_style );
 593
 594                             p_ssa_style = NULL;
 595                             i_style_level--;
 596                         }
 597                         break;
 598                 }
 599                 break;
 600
 601             case XML_READER_STARTELEM:
 602                 if( !strcasecmp( "metadata", node ) && (i_style_level == 0) )
 603                     i_metadata_level++;
 604                 else if( !strcasecmp( "resolution", node ) &&
 605                          ( i_metadata_level == 1) )
 606                 {
 607                     const char *attr, *val;
 608                     while( (attr = xml_ReaderNextAttr( p_xml_reader, &val )) )
 609                     {
 610                         if( !strcasecmp( "x", attr ) )
 611                             p_sys->i_original_width = atoi( val );
 612                         else if( !strcasecmp( "y", attr ) )
 613                             p_sys->i_original_height = atoi( val );
 614                     }
 615                 }
 616                 else if( !strcasecmp( "styles", node ) && (i_style_level == 0) )
 617                 {
 618                     i_style_level++;
 619                 }
 620                 else if( !strcasecmp( "style", node ) && (i_style_level == 1) )
 621                 {
 622                     i_style_level++;
 623
 624                     p_ssa_style = calloc( 1, sizeof(ssa_style_t) );
 625                     if( unlikely(!p_ssa_style) )
 626                         return;
 627                     /* All styles are supposed to default to Default, and then
 628                      * one or more settings are over-ridden.
 629                      * At the moment this only effects styles defined AFTER
 630                      * Default in the XML
 631                      */
 632                     for( int i = 0; i < p_sys->i_ssa_styles; i++ )
 633                     {
 634                         if( !strcasecmp( p_sys->pp_ssa_styles[i]->psz_stylename, "Default" ) )
 635                         {
 636                             ssa_style_t *p_default_style = p_sys->pp_ssa_styles[i];
 637
 638                             memcpy( p_ssa_style, p_default_style, sizeof( ssa_style_t ) );
 639                             //FIXME: Make font_style a pointer. Actually we double copy some data here,
 640                             //   we use text_style_Copy to avoid copying psz_fontname, though .
 641                             text_style_Copy( &p_ssa_style->font_style, &p_default_style->font_style );
 642                             p_ssa_style->psz_stylename = NULL;
 643                         }
 644                     }
 645
 646                     const char *attr, *val;
 647                     while( (attr = xml_ReaderNextAttr( p_xml_reader, &val )) )
 648                     {
 649                         if( !strcasecmp( "name", attr ) )
 650                         {
 651                             free( p_ssa_style->psz_stylename );
 652                             p_ssa_style->psz_stylename = strdup( val );
 653                         }
 654                     }
 655                 }
 656                 else if( !strcasecmp( "fontstyle", node ) && (i_style_level == 2) )
 657                 {
 658                     const char *attr, *val;
 659                     while( (attr = xml_ReaderNextAttr( p_xml_reader, &val )) )
 660                     {
 661                         if( !strcasecmp( "face", attr ) )
 662                         {
 663                             free( p_ssa_style->font_style.psz_fontname );
 664                             p_ssa_style->font_style.psz_fontname = strdup( val );
 665                         }
 666                         else if( !strcasecmp( "size", attr ) )
 667                         {
 668                             if( ( *val == '+' ) || ( *val == '-' ) )
 669                             {
 670                                 int i_value = atoi( val );
 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( val );
 682                         }
 683                         else if( !strcasecmp( "italic", attr ) )
 684                         {
 685                             if( !strcasecmp( "yes", val ))
 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", attr ) )
 691                         {
 692                             if( !strcasecmp( "bold", val ))
 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", attr ) )
 698                         {
 699                             if( !strcasecmp( "yes", val ))
 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", attr ) )
 705                         {
 706                             if( *val == '#' )
 707                             {
 708                                 unsigned long col = strtol(val+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", attr ) )
 714                         {
 715                             if( *val == '#' )
 716                             {
 717                                 unsigned long col = strtol(val+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", attr ) )
 723                         {
 724                             p_ssa_style->font_style.i_outline_width = atoi( val );
 725                         }
 726                         else if( !strcasecmp( "shadow-color", attr ) )
 727                         {
 728                             if( *val == '#' )
 729                             {
 730                                 unsigned long col = strtol(val+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", attr ) )
 736                         {
 737                             p_ssa_style->font_style.i_shadow_width = atoi( val );
 738                         }
 739                         else if( !strcasecmp( "back-color", attr ) )
 740                         {
 741                             if( *val == '#' )
 742                             {
 743                                 unsigned long col = strtol(val+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", attr ) )
 749                         {
 750                             p_ssa_style->font_style.i_spacing = atoi( val );
 751                         }
 752                     }
 753                 }
 754                 else if( !strcasecmp( "position", node ) && (i_style_level == 2) )
 755                 {
 756                     const char *attr, *val;
 757                     while( (attr = xml_ReaderNextAttr( p_xml_reader, &val )) )
 758                     {
 759                         if( !strcasecmp( "alignment", attr ) )
 760                         {
 761                             if( !strcasecmp( "TopLeft", val ) )
 762                                 p_ssa_style->i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_LEFT;
 763                             else if( !strcasecmp( "TopCenter", val ) )
 764                                 p_ssa_style->i_align = SUBPICTURE_ALIGN_TOP;
 765                             else if( !strcasecmp( "TopRight", val ) )
 766                                 p_ssa_style->i_align = SUBPICTURE_ALIGN_TOP | SUBPICTURE_ALIGN_RIGHT;
 767                             else if( !strcasecmp( "MiddleLeft", val ) )
 768                                 p_ssa_style->i_align = SUBPICTURE_ALIGN_LEFT;
 769                             else if( !strcasecmp( "MiddleCenter", val ) )
 770                                 p_ssa_style->i_align = 0;
 771                             else if( !strcasecmp( "MiddleRight", val ) )
 772                                 p_ssa_style->i_align = SUBPICTURE_ALIGN_RIGHT;
 773                             else if( !strcasecmp( "BottomLeft", val ) )
 774                                 p_ssa_style->i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_LEFT;
 775                             else if( !strcasecmp( "BottomCenter", val ) )
 776                                 p_ssa_style->i_align = SUBPICTURE_ALIGN_BOTTOM;
 777                             else if( !strcasecmp( "BottomRight", val ) )
 778                                 p_ssa_style->i_align = SUBPICTURE_ALIGN_BOTTOM | SUBPICTURE_ALIGN_RIGHT;
 779                         }
 780                         else if( !strcasecmp( "horizontal-margin", attr ) )
 781                         {
 782                             if( strchr( val, '%' ) )
 783                             {
 784                                 p_ssa_style->i_margin_h = 0;
 785                                 p_ssa_style->i_margin_percent_h = atoi( val );
 786                             }
 787                             else
 788                             {
 789                                 p_ssa_style->i_margin_h = atoi( val );
 790                                 p_ssa_style->i_margin_percent_h = 0;
 791                             }
 792                         }
 793                         else if( !strcasecmp( "vertical-margin", attr ) )
 794                         {
 795                             if( strchr( val, '%' ) )
 796                             {
 797                                 p_ssa_style->i_margin_v = 0;
 798                                 p_ssa_style->i_margin_percent_v = atoi( val );
 799                             }
 800                             else
 801                             {
 802                                 p_ssa_style->i_margin_v = atoi( val );
 803                                 p_ssa_style->i_margin_percent_v = 0;
 804                             }
 805                         }
 806                     }
 807                 }
 808                 break;
 809         }
 810     }
 811     free( p_ssa_style );
 812 }
 813
 814
 815
 816 static subpicture_region_t *ParseUSFString( decoder_t *p_dec,
 817                                             char *psz_subtitle )
 818 {
 819     decoder_sys_t        *p_sys = p_dec->p_sys;
 820     subpicture_region_t  *p_region_first = NULL;
 821     subpicture_region_t  *p_region_upto  = p_region_first;
 822
 823     while( *psz_subtitle )
 824     {
 825         if( *psz_subtitle == '<' )
 826         {
 827             char *psz_end = NULL;
 828
 829             if(( !strncasecmp( psz_subtitle, "<text ", 6 )) ||
 830                ( !strncasecmp( psz_subtitle, "<text>", 6 )))
 831             {
 832                 psz_end = strcasestr( psz_subtitle, "</text>" );
 833
 834                 if( psz_end )
 835                 {
 836                     subpicture_region_t  *p_text_region;
 837
 838                     psz_end += strcspn( psz_end, ">" ) + 1;
 839
 840                     p_text_region = CreateTextRegion( p_dec,
 841                                                       psz_subtitle,
 842                                                       psz_end - psz_subtitle,
 843                                                       p_sys->i_align );
 844
 845                     if( p_text_region )
 846                     {
 847                         p_text_region->psz_text = CreatePlainText( p_text_region->psz_html );
 848
 849                         if( ! var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 850                         {
 851                             free( p_text_region->psz_html );
 852                             p_text_region->psz_html = NULL;
 853                         }
 854                     }
 855
 856                     if( !p_region_first )
 857                     {
 858                         p_region_first = p_region_upto = p_text_region;
 859                     }
 860                     else if( p_text_region )
 861                     {
 862                         p_region_upto->p_next = p_text_region;
 863                         p_region_upto = p_region_upto->p_next;
 864                     }
 865                 }
 866             }
 867             else if(( !strncasecmp( psz_subtitle, "<karaoke ", 9 )) ||
 868                     ( !strncasecmp( psz_subtitle, "<karaoke>", 9 )))
 869             {
 870                 psz_end = strcasestr( psz_subtitle, "</karaoke>" );
 871
 872                 if( psz_end )
 873                 {
 874                     subpicture_region_t  *p_text_region;
 875
 876                     psz_end += strcspn( psz_end, ">" ) + 1;
 877
 878                     p_text_region = CreateTextRegion( p_dec,
 879                                                       psz_subtitle,
 880                                                       psz_end - psz_subtitle,
 881                                                       p_sys->i_align );
 882
 883                     if( p_text_region )
 884                     {
 885                         if( ! var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 886                         {
 887                             free( p_text_region->psz_html );
 888                             p_text_region->psz_html = NULL;
 889                         }
 890                     }
 891                     if( !p_region_first )
 892                     {
 893                         p_region_first = p_region_upto = p_text_region;
 894                     }
 895                     else if( p_text_region )
 896                     {
 897                         p_region_upto->p_next = p_text_region;
 898                         p_region_upto = p_region_upto->p_next;
 899                     }
 900                 }
 901             }
 902             else if(( !strncasecmp( psz_subtitle, "<image ", 7 )) ||
 903                     ( !strncasecmp( psz_subtitle, "<image>", 7 )))
 904             {
 905                 subpicture_region_t *p_image_region = NULL;
 906
 907                 char *psz_end = strcasestr( psz_subtitle, "</image>" );
 908                 char *psz_content = strchr( psz_subtitle, '>' );
 909                 int   i_transparent = -1;
 910
 911                 /* If a colorkey parameter is specified, then we have to map
 912                  * that index in the picture through as transparent (it is
 913                  * required by the USF spec but is also recommended that if the
 914                  * creator really wants a transparent colour that they use a
 915                  * type like PNG that properly supports it; this goes doubly
 916                  * for VLC because the pictures are stored internally in YUV
 917                  * and the resulting colour-matching may not produce the
 918                  * desired results.)
 919                  */
 920                 char *psz_tmp = GrabAttributeValue( "colorkey", psz_subtitle );
 921                 if( psz_tmp )
 922                 {
 923                     if( *psz_tmp == '#' )
 924                         i_transparent = strtol( psz_tmp + 1, NULL, 16 ) & 0x00ffffff;
 925                     free( psz_tmp );
 926                 }
 927                 if( psz_content && ( psz_content < psz_end ) )
 928                 {
 929                     char *psz_filename = strndup( &psz_content[1], psz_end - &psz_content[1] );
 930                     if( psz_filename )
 931                     {
 932                         p_image_region = LoadEmbeddedImage( p_dec,
 933                                             psz_filename, i_transparent );
 934                         free( psz_filename );
 935                     }
 936                 }
 937
 938                 if( psz_end ) psz_end += strcspn( psz_end, ">" ) + 1;
 939
 940                 if( p_image_region )
 941                 {
 942                     SetupPositions( p_image_region, psz_subtitle );
 943
 944                     p_image_region->p_next   = NULL;
 945                     p_image_region->psz_text = NULL;
 946                     p_image_region->psz_html = NULL;
 947
 948                 }
 949                 if( !p_region_first )
 950                 {
 951                     p_region_first = p_region_upto = p_image_region;
 952                 }
 953                 else if( p_image_region )
 954                 {
 955                     p_region_upto->p_next = p_image_region;
 956                     p_region_upto = p_region_upto->p_next;
 957                 }
 958             }
 959             if( psz_end )
 960                 psz_subtitle = psz_end - 1;
 961
 962             psz_subtitle += strcspn( psz_subtitle, ">" );
 963         }
 964
 965         psz_subtitle++;
 966     }
 967
 968     return p_region_first;
 969 }
 970
 971 /*****************************************************************************
 972  * ParseUSFHeader: Retrieve global formatting information etc
 973  *****************************************************************************/
 974 static void ParseUSFHeader( decoder_t *p_dec )
 975 {
 976     stream_t      *p_sub = NULL;
 977     xml_reader_t  *p_xml_reader = NULL;
 978
 979     p_sub = stream_MemoryNew( VLC_OBJECT(p_dec),
 980                               p_dec->fmt_in.p_extra,
 981                               p_dec->fmt_in.i_extra,
 982                               true );
 983     if( !p_sub )
 984         return;
 985
 986     p_xml_reader = xml_ReaderCreate( p_dec, p_sub );
 987     if( likely(p_xml_reader) )
 988     {
 989         const char *node;
 990
 991         /* Look for Root Node */
 992         if( xml_ReaderNextNode( p_xml_reader, &node ) == XML_READER_STARTELEM
 993          && !strcasecmp( "usfsubtitles", node ) )
 994             ParseUSFHeaderTags( p_dec, p_xml_reader );
 995
 996         xml_ReaderDelete( p_xml_reader );
 997     }
 998     stream_Delete( p_sub );
 999 }
1000
1001 /* Function now handles tags which has attribute values, and tries
1002  * to deal with &' commands too. It no longer modifies the string
1003  * in place, so that the original text can be reused
1004  */
1005 static char *StripTags( char *psz_subtitle )
1006 {
1007     char *psz_text_start;
1008     char *psz_text;
1009
1010     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
1011     if( !psz_text_start )
1012         return NULL;
1013
1014     while( *psz_subtitle )
1015     {
1016         /* Mask out any pre-existing LFs in the subtitle */
1017         if( *psz_subtitle == '\n' )
1018             *psz_subtitle = ' ';
1019
1020         if( *psz_subtitle == '<' )
1021         {
1022             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
1023                 *psz_text++ = '\n';
1024
1025             psz_subtitle += strcspn( psz_subtitle, ">" );
1026         }
1027         else if( *psz_subtitle == '&' )
1028         {
1029             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
1030             {
1031                 *psz_text++ = '<';
1032                 psz_subtitle += strcspn( psz_subtitle, ";" );
1033             }
1034             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
1035             {
1036                 *psz_text++ = '>';
1037                 psz_subtitle += strcspn( psz_subtitle, ";" );
1038             }
1039             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
1040             {
1041                 *psz_text++ = '&';
1042                 psz_subtitle += strcspn( psz_subtitle, ";" );
1043             }
1044             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
1045             {
1046                 *psz_text++ = '\"';
1047                 psz_subtitle += strcspn( psz_subtitle, ";" );
1048             }
1049             else
1050             {
1051                 /* Assume it is just a normal ampersand */
1052                 *psz_text++ = '&';
1053             }
1054         }
1055         else
1056         {
1057             *psz_text++ = *psz_subtitle;
1058         }
1059
1060         /* Security fix: Account for the case where input ends early */
1061         if( *psz_subtitle == '\0' ) break;
1062
1063         psz_subtitle++;
1064     }
1065     *psz_text++ = '\0';
1066
1067     char *psz = realloc( psz_text_start, psz_text - psz_text_start );
1068     return likely(psz != NULL) ? psz : psz_text_start;
1069 }
1070
1071 /* Turn a HTML subtitle, turn into a plain-text version,
1072  *  complete with sensible whitespace compaction
1073  */
1074
1075 static char *CreatePlainText( char *psz_subtitle )
1076 {
1077     char *psz_text = StripTags( psz_subtitle );
1078     char *s;
1079
1080     if( !psz_text )
1081         return NULL;
1082
1083     s = strpbrk( psz_text, "\t\r\n " );
1084     while( s )
1085     {
1086         int   k;
1087         char  spc = ' ';
1088         int   i_whitespace = strspn( s, "\t\r\n " );
1089
1090         /* Favour '\n' over other whitespaces - if one of these
1091          * occurs in the whitespace use a '\n' as our value,
1092          * otherwise just use a ' '
1093          */
1094         for( k = 0; k < i_whitespace; k++ )
1095             if( s[k] == '\n' ) spc = '\n';
1096
1097         if( i_whitespace > 1 )
1098         {
1099             memmove( &s[1],
1100                      &s[i_whitespace],
1101                      strlen( s ) - i_whitespace + 1 );
1102         }
1103         *s++ = spc;
1104
1105         s = strpbrk( s, "\t\r\n " );
1106     }
1107     return psz_text;
1108 }
1109
1110 /****************************************************************************
1111  * download and resize image located at psz_url
1112  ***************************************************************************/
1113 static subpicture_region_t *LoadEmbeddedImage( decoder_t *p_dec,
1114                                                const char *psz_filename,
1115                                                int i_transparent_color )
1116 {
1117     decoder_sys_t         *p_sys = p_dec->p_sys;
1118     subpicture_region_t   *p_region;
1119     video_format_t         fmt_out;
1120     picture_t             *p_pic = NULL;
1121
1122     for( int k = 0; k < p_sys->i_images; k++ )
1123     {
1124         if( p_sys->pp_images &&
1125             !strcmp( p_sys->pp_images[k]->psz_filename, psz_filename ) )
1126         {
1127             p_pic = p_sys->pp_images[k]->p_pic;
1128             break;
1129         }
1130     }
1131
1132     if( !p_pic )
1133     {
1134         msg_Err( p_dec, "Unable to read image %s", psz_filename );
1135         return NULL;
1136     }
1137
1138     /* Display the feed's image */
1139     memset( &fmt_out, 0, sizeof( video_format_t));
1140
1141     fmt_out.i_chroma = VLC_CODEC_YUVA;
1142     fmt_out.i_sar_num = fmt_out.i_sar_den = 1;
1143     fmt_out.i_width =
1144         fmt_out.i_visible_width = p_pic->format.i_visible_width;
1145     fmt_out.i_height =
1146         fmt_out.i_visible_height = p_pic->format.i_visible_height;
1147
1148     p_region = subpicture_region_New( &fmt_out );
1149     if( !p_region )
1150     {
1151         msg_Err( p_dec, "cannot allocate SPU region" );
1152         return NULL;
1153     }
1154     assert( p_pic->format.i_chroma == VLC_CODEC_YUVA );
1155     /* FIXME the copy is probably not needed anymore */
1156     picture_CopyPixels( p_region->p_picture, p_pic );
1157
1158     /* This isn't the best way to do this - if you really want transparency, then
1159      * you're much better off using an image type that supports it like PNG. The
1160      * spec requires this support though.
1161      */
1162     if( i_transparent_color > 0 )
1163     {
1164         int i_r = ( i_transparent_color >> 16 ) & 0xff;
1165         int i_g = ( i_transparent_color >>  8 ) & 0xff;
1166         int i_b = ( i_transparent_color       ) & 0xff;
1167
1168         /* FIXME it cannot work as the yuv conversion code will probably NOT match
1169          * this one  */
1170         int i_y = ( ( (  66 * i_r + 129 * i_g +  25 * i_b + 128 ) >> 8 ) + 16 );
1171         int i_u =   ( ( -38 * i_r -  74 * i_g + 112 * i_b + 128 ) >> 8 ) + 128 ;
1172         int i_v =   ( ( 112 * i_r -  94 * i_g -  18 * i_b + 128 ) >> 8 ) + 128 ;
1173
1174         assert( p_region->fmt.i_chroma == VLC_CODEC_YUVA );
1175         for( unsigned int y = 0; y < p_region->fmt.i_height; y++ )
1176         {
1177             for( unsigned int x = 0; x < p_region->fmt.i_width; x++ )
1178             {
1179                 if( p_region->p_picture->Y_PIXELS[y*p_region->p_picture->Y_PITCH + x] != i_y ||
1180                     p_region->p_picture->U_PIXELS[y*p_region->p_picture->U_PITCH + x] != i_u ||
1181                     p_region->p_picture->V_PIXELS[y*p_region->p_picture->V_PITCH + x] != i_v )
1182                     continue;
1183                 p_region->p_picture->A_PIXELS[y*p_region->p_picture->A_PITCH + x] = 0;
1184
1185             }
1186         }
1187     }
1188     return p_region;
1189 }