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

   1 /*****************************************************************************
   2  * subsass.c : ASS/SSA subtitles decoder
   3  *****************************************************************************
   4  * Copyright (C) 2000-2006 the VideoLAN team
   5  * $Id$
   6  *
   7  * Authors: Gildas Bazin <gbazin@videolan.org>
   8  *          Samuel Hocevar <sam@zoy.org>
   9  *          Derk-Jan Hartman <hartman at videolan dot org>
  10  *
  11  * This program is free software; you can redistribute it and/or modify
  12  * it under the terms of the GNU General Public License as published by
  13  * the Free Software Foundation; either version 2 of the License, or
  14  * (at your option) any later version.
  15  *
  16  * This program is distributed in the hope that it will be useful,
  17  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  18  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19  * GNU General Public License for more details.
  20  *
  21  * You should have received a copy of the GNU General Public License
  22  * along with this program; if not, write to the Free Software
  23  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
  24  *****************************************************************************/
  25
  26 #include "subsdec.h"
  27
  28
  29 void ParseSSAString( decoder_t *p_dec,
  30                      char *psz_subtitle,
  31                      subpicture_t *p_spu_in )
  32 {
  33     /* We expect MKV formatted SSA:
  34      * ReadOrder, Layer, Style, CharacterName, MarginL, MarginR,
  35      * MarginV, Effect, Text */
  36     decoder_sys_t   *p_sys = p_dec->p_sys;
  37     subpicture_t    *p_spu = p_spu_in;
  38     ssa_style_t     *p_style = NULL;
  39     char            *psz_new_subtitle = NULL;
  40     char            *psz_buffer_sub = NULL;
  41     char            *psz_style = NULL;
  42     char            *psz_style_start = NULL;
  43     char            *psz_style_end = NULL;
  44     int             i_text = 0, i_comma = 0, i_strlen = 0, i;
  45     int             i_margin_l = 0, i_margin_r = 0, i_margin_v = 0;
  46
  47     psz_buffer_sub = psz_subtitle;
  48
  49     p_spu->p_region->psz_html = NULL;
  50
  51     i_comma = 0;
  52     while( i_comma < 8 && *psz_buffer_sub != '\0' )
  53     {
  54         if( *psz_buffer_sub == ',' )
  55         {
  56             i_comma++;
  57             if( i_comma == 2 )
  58                 psz_style_start = &psz_buffer_sub[1];
  59             else if( i_comma == 3 )
  60                 psz_style_end = &psz_buffer_sub[0];
  61             else if( i_comma == 4 )
  62                 i_margin_l = (int)strtol( &psz_buffer_sub[1], NULL, 10 );
  63             else if( i_comma == 5 )
  64                 i_margin_r = (int)strtol( &psz_buffer_sub[1], NULL, 10 );
  65             else if( i_comma == 6 )
  66                 i_margin_v = (int)strtol( &psz_buffer_sub[1], NULL, 10 );
  67         }
  68         psz_buffer_sub++;
  69     }
  70
  71     if( *psz_buffer_sub == '\0' && i_comma == 8 )
  72     {
  73         msg_Dbg( p_dec, "couldn't find all fields in this SSA line" );
  74         return;
  75     }
  76
  77     psz_new_subtitle = malloc( strlen( psz_buffer_sub ) + 1);
  78     i_text = 0;
  79     while( psz_buffer_sub[0] != '\0' )
  80     {
  81         if( psz_buffer_sub[0] == '\\' && psz_buffer_sub[1] == 'n' )
  82         {
  83             psz_new_subtitle[i_text] = ' ';
  84             i_text++;
  85             psz_buffer_sub += 2;
  86         }
  87         else if( psz_buffer_sub[0] == '\\' && psz_buffer_sub[1] == 'N' )
  88         {
  89             psz_new_subtitle[i_text] = '\n';
  90             i_text++;
  91             psz_buffer_sub += 2;
  92         }
  93         else if( psz_buffer_sub[0] == '{' &&
  94                  psz_buffer_sub[1] == '\\' )
  95         {
  96             /* SSA control code */
  97             while( psz_buffer_sub[0] != '\0' &&
  98                    psz_buffer_sub[0] != '}' )
  99             {
 100                 psz_buffer_sub++;
 101             }
 102             psz_buffer_sub++;
 103         }
 104         else
 105         {
 106             psz_new_subtitle[i_text] = psz_buffer_sub[0];
 107             i_text++;
 108             psz_buffer_sub++;
 109         }
 110     }
 111     psz_new_subtitle[i_text] = '\0';
 112
 113     i_strlen = __MAX( psz_style_end - psz_style_start, 0);
 114     psz_style = strndup( psz_style_start, i_strlen );
 115
 116     for( i = 0; i < p_sys->i_ssa_styles; i++ )
 117     {
 118         if( !strcmp( p_sys->pp_ssa_styles[i]->psz_stylename, psz_style ) )
 119             p_style = p_sys->pp_ssa_styles[i];
 120     }
 121     if( psz_style ) free( psz_style );
 122
 123     p_spu->p_region->psz_text = psz_new_subtitle;
 124     if( p_style == NULL )
 125     {
 126         p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
 127         p_spu->i_x = p_sys->i_align ? 20 : 0;
 128         p_spu->i_y = 10;
 129     }
 130     else
 131     {
 132         msg_Dbg( p_dec, "style is: %s", p_style->psz_stylename);
 133         p_spu->p_region->p_style = &p_style->font_style;
 134         p_spu->p_region->i_align = p_style->i_align;
 135         if( p_style->i_align & SUBPICTURE_ALIGN_LEFT )
 136         {
 137             p_spu->i_x = (i_margin_l) ? i_margin_l : p_style->i_margin_h;
 138         }
 139         else if( p_style->i_align & SUBPICTURE_ALIGN_RIGHT )
 140         {
 141             p_spu->i_x = (i_margin_r) ? i_margin_r : p_style->i_margin_h;
 142         }
 143         p_spu->i_y = (i_margin_v) ? i_margin_v : p_style->i_margin_v;
 144     }
 145 }
 146
 147 /*****************************************************************************
 148  * ParseColor: SSA stores color in BBGGRR, in ASS it uses AABBGGRR
 149  * The string value in the string can be a pure integer, or hexadecimal &HBBGGRR
 150  *****************************************************************************/
 151 static void ParseColor( decoder_t *p_dec,
 152                         char *psz_color,
 153                         int *pi_color,
 154                         int *pi_alpha )
 155 {
 156     int i_color = 0;
 157     if( !strncasecmp( psz_color, "&H", 2 ) )
 158     {
 159         /* textual HEX representation */
 160         i_color = (int) strtol( psz_color+2, NULL, 16 );
 161     }
 162     else i_color = (int) strtol( psz_color, NULL, 0 );
 163
 164     *pi_color = 0;
 165     *pi_color |= ( ( i_color & 0x000000FF ) << 16 ); /* Red */
 166     *pi_color |= ( ( i_color & 0x0000FF00 ) );       /* Green */
 167     *pi_color |= ( ( i_color & 0x00FF0000 ) >> 16 ); /* Blue */
 168
 169     if( pi_alpha != NULL )
 170         *pi_alpha = ( i_color & 0xFF000000 ) >> 24;
 171 }
 172
 173 /*****************************************************************************
 174  * ParseSSAHeader: Retrieve global formatting information etc
 175  *****************************************************************************/
 176 void ParseSSAHeader( decoder_t *p_dec )
 177 {
 178     decoder_sys_t *p_sys = p_dec->p_sys;
 179     char *psz_parser = NULL;
 180     char *psz_header = malloc( p_dec->fmt_in.i_extra+1 );
 181     int i_section_type = 1;
 182
 183     memcpy( psz_header, p_dec->fmt_in.p_extra, p_dec->fmt_in.i_extra );
 184     psz_header[ p_dec->fmt_in.i_extra] = '\0';
 185
 186     /* Handle [Script Info] section */
 187     psz_parser = strcasestr( psz_header, "[Script Info]" );
 188     if( psz_parser == NULL ) goto eof;
 189
 190     psz_parser = GotoNextLine( psz_parser );
 191
 192     while( psz_parser[0] != '\0' )
 193     {
 194         int temp;
 195         char buffer_text[MAX_LINE + 1];
 196
 197         if( psz_parser[0] == '!' || psz_parser[0] == ';' ) /* comment */;
 198         else if( sscanf( psz_parser, "PlayResX: %d", &temp ) == 1 )
 199             p_sys->i_original_width = ( temp > 0 ) ? temp : -1;
 200         else if( sscanf( psz_parser, "PlayResY: %d", &temp ) == 1 )
 201             p_sys->i_original_height = ( temp > 0 ) ? temp : -1;
 202         else if( sscanf( psz_parser, "Script Type: %8192s", buffer_text ) == 1 )
 203         {
 204             if( !strcasecmp( buffer_text, "V4.00+" ) ) p_sys->b_ass = VLC_TRUE;
 205         }
 206         else if( !strncasecmp( psz_parser, "[V4 Styles]", 11 ) )
 207             i_section_type = 1;
 208         else if( !strncasecmp( psz_parser, "[V4+ Styles]", 12) )
 209         {
 210             i_section_type = 2;
 211             p_sys->b_ass = VLC_TRUE;
 212         }
 213         else if( !strncasecmp( psz_parser, "[Events]", 8 ) )
 214             i_section_type = 4;
 215         else if( !strncasecmp( psz_parser, "Style:", 6 ) )
 216         {
 217             int i_font_size, i_bold, i_italic, i_border, i_outline, i_shadow,
 218                 i_underline, i_strikeout, i_scale_x, i_scale_y, i_spacing,
 219                 i_align, i_margin_l, i_margin_r, i_margin_v;
 220
 221             char psz_temp_stylename[MAX_LINE+1];
 222             char psz_temp_fontname[MAX_LINE+1];
 223             char psz_temp_color1[MAX_LINE+1];
 224             char psz_temp_color2[MAX_LINE+1];
 225             char psz_temp_color3[MAX_LINE+1];
 226             char psz_temp_color4[MAX_LINE+1];
 227
 228             if( i_section_type == 1 ) /* V4 */
 229             {
 230                 if( sscanf( psz_parser, "Style: %8192[^,],%8192[^,],%d,%8192[^,],%8192[^,],%8192[^,],%8192[^,],%d,%d,%d,%d,%d,%d,%d,%d,%d%*[^\r\n]",
 231                     psz_temp_stylename, psz_temp_fontname, &i_font_size,
 232                     psz_temp_color1, psz_temp_color2, psz_temp_color3,
 233                     psz_temp_color4, &i_bold, &i_italic,
 234                     &i_border, &i_outline, &i_shadow, &i_align, &i_margin_l,
 235                     &i_margin_r, &i_margin_v ) == 16 )
 236                 {
 237                     ssa_style_t *p_style = malloc( sizeof(ssa_style_t) );
 238
 239                     p_style->psz_stylename = strdup( psz_temp_stylename );
 240                     p_style->font_style.psz_fontname = strdup( psz_temp_fontname );
 241                     p_style->font_style.i_font_size = i_font_size;
 242
 243                     ParseColor( p_dec, psz_temp_color1, &p_style->font_style.i_font_color, NULL );
 244                     ParseColor( p_dec, psz_temp_color4, &p_style->font_style.i_shadow_color, NULL );
 245                     p_style->font_style.i_outline_color = p_style->font_style.i_shadow_color;
 246                     p_style->font_style.i_font_alpha = p_style->font_style.i_outline_alpha
 247                                                      = p_style->font_style.i_shadow_alpha = 0x00;
 248                     p_style->font_style.i_style_flags = 0;
 249                     if( i_bold ) p_style->font_style.i_style_flags |= STYLE_BOLD;
 250                     if( i_italic ) p_style->font_style.i_style_flags |= STYLE_ITALIC;
 251
 252                     if( i_border == 1 )
 253                         p_style->font_style.i_style_flags |= (STYLE_ITALIC | STYLE_OUTLINE);
 254                     else if( i_border == 3 )
 255                     {
 256                         p_style->font_style.i_style_flags |= STYLE_BACKGROUND;
 257                         p_style->font_style.i_background_color = p_style->font_style.i_shadow_color;
 258                         p_style->font_style.i_background_alpha = p_style->font_style.i_shadow_alpha;
 259                     }
 260                     p_style->font_style.i_shadow_width = i_shadow;
 261                     p_style->font_style.i_outline_width = i_outline;
 262
 263                     p_style->i_align = 0;
 264                     if( i_align == 1 || i_align == 5 || i_align == 9 )
 265                         p_style->i_align |= SUBPICTURE_ALIGN_LEFT;
 266                     if( i_align == 3 || i_align == 7 || i_align == 11 )
 267                         p_style->i_align |= SUBPICTURE_ALIGN_RIGHT;
 268                     if( i_align < 4 )
 269                         p_style->i_align |= SUBPICTURE_ALIGN_BOTTOM;
 270                     else if( i_align < 8 )
 271                         p_style->i_align |= SUBPICTURE_ALIGN_TOP;
 272
 273                     p_style->i_margin_h = ( p_style->i_align & SUBPICTURE_ALIGN_RIGHT ) ?
 274                                                         i_margin_r : i_margin_l;
 275                     p_style->i_margin_v = i_margin_v;
 276                     p_style->i_margin_percent_h = 0;
 277                     p_style->i_margin_percent_v = 0;
 278
 279                     p_style->font_style.i_karaoke_background_color = 0xffffff;
 280                     p_style->font_style.i_karaoke_background_alpha = 0xff;
 281
 282                     TAB_APPEND( p_sys->i_ssa_styles, p_sys->pp_ssa_styles, p_style );
 283                 }
 284                 else msg_Warn( p_dec, "SSA v4 styleline parsing failed" );
 285             }
 286             else if( i_section_type == 2 ) /* V4+ */
 287             {
 288                 /* Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour,
 289                    Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline,
 290                    Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
 291                 */
 292                 if( sscanf( psz_parser, "Style: %8192[^,],%8192[^,],%d,%8192[^,],%8192[^,],%8192[^,],%8192[^,],%d,%d,%d,%d,%d,%d,%d,%*f,%d,%d,%d,%d,%d,%d,%d%*[^\r\n]",
 293                     psz_temp_stylename, psz_temp_fontname, &i_font_size,
 294                     psz_temp_color1, psz_temp_color2, psz_temp_color3, psz_temp_color4, &i_bold, &i_italic,
 295                     &i_underline, &i_strikeout, &i_scale_x, &i_scale_y, &i_spacing, &i_border, &i_outline,
 296                     &i_shadow, &i_align, &i_margin_l, &i_margin_r, &i_margin_v ) == 21 )
 297                 {
 298                     ssa_style_t *p_style = malloc( sizeof(ssa_style_t) );
 299
 300                     p_style->psz_stylename = strdup( psz_temp_stylename );
 301                     p_style->font_style.psz_fontname = strdup( psz_temp_fontname );
 302                     p_style->font_style.i_font_size = i_font_size;
 303                     msg_Dbg( p_dec, psz_temp_color1 );
 304                     ParseColor( p_dec, psz_temp_color1, &p_style->font_style.i_font_color,
 305                                 &p_style->font_style.i_font_alpha );
 306                     ParseColor( p_dec, psz_temp_color3, &p_style->font_style.i_outline_color,
 307                                 &p_style->font_style.i_outline_alpha );
 308                     ParseColor( p_dec, psz_temp_color4, &p_style->font_style.i_shadow_color,
 309                                 &p_style->font_style.i_shadow_alpha );
 310
 311                     p_style->font_style.i_style_flags = 0;
 312                     if( i_bold ) p_style->font_style.i_style_flags |= STYLE_BOLD;
 313                     if( i_italic ) p_style->font_style.i_style_flags |= STYLE_ITALIC;
 314                     if( i_underline ) p_style->font_style.i_style_flags |= STYLE_UNDERLINE;
 315                     if( i_strikeout ) p_style->font_style.i_style_flags |= STYLE_STRIKEOUT;
 316                     if( i_border == 1 ) p_style->font_style.i_style_flags |= (STYLE_ITALIC | STYLE_OUTLINE);
 317                     else if( i_border == 3 )
 318                     {
 319                         p_style->font_style.i_style_flags |= STYLE_BACKGROUND;
 320                         p_style->font_style.i_background_color = p_style->font_style.i_shadow_color;
 321                         p_style->font_style.i_background_alpha = p_style->font_style.i_shadow_alpha;
 322                     }
 323                     p_style->font_style.i_shadow_width  = ( i_border == 1 ) ? i_shadow : 0;
 324                     p_style->font_style.i_outline_width = ( i_border == 1 ) ? i_outline : 0;
 325                     p_style->font_style.i_spacing = i_spacing;
 326                     //p_style->font_style.f_angle = f_angle;
 327
 328                     p_style->i_align = 0;
 329                     if( i_align == 0x1 || i_align == 0x4 || i_align == 0x7 )
 330                         p_style->i_align |= SUBPICTURE_ALIGN_LEFT;
 331                     if( i_align == 0x3 || i_align == 0x6 || i_align == 0x9 )
 332                         p_style->i_align |= SUBPICTURE_ALIGN_RIGHT;
 333                     if( i_align == 0x7 || i_align == 0x8 || i_align == 0x9 )
 334                         p_style->i_align |= SUBPICTURE_ALIGN_TOP;
 335                     if( i_align == 0x1 || i_align == 0x2 || i_align == 0x3 )
 336                         p_style->i_align |= SUBPICTURE_ALIGN_BOTTOM;
 337                     p_style->i_margin_h = ( p_style->i_align & SUBPICTURE_ALIGN_RIGHT ) ?
 338                                             i_margin_r : i_margin_l;
 339                     p_style->i_margin_v = i_margin_v;
 340                     p_style->i_margin_percent_h = 0;
 341                     p_style->i_margin_percent_v = 0;
 342
 343                     p_style->font_style.i_karaoke_background_color = 0xffffff;
 344                     p_style->font_style.i_karaoke_background_alpha = 0xff;
 345
 346                     /*TODO: Ignored: angle i_scale_x|y (fontscaling), i_encoding */
 347                     TAB_APPEND( p_sys->i_ssa_styles, p_sys->pp_ssa_styles, p_style );
 348                 }
 349                 else msg_Dbg( p_dec, "SSA V4+ styleline parsing failed" );
 350             }
 351         }
 352         psz_parser = GotoNextLine( psz_parser );
 353     }
 354
 355 eof:
 356     if( psz_header ) free( psz_header );
 357     return;
 358 }
 359
 360