Subsdec: Split the decoder for subsdec / SSA / USF in three files, since the code...
[vlc.git] / modules / codec / subtitles / subsass.c
1 /*****************************************************************************
2  * subsass.c : ASS/SSA subtitles decoder
3  *****************************************************************************
4  * Copyright (C) 2000-2006 the VideoLAN team
5  * $Id: subsdec.c 20996 2007-08-05 20:01:21Z jb $
6  *
7  * Authors: Gildas Bazin <gbazin@videolan.org>
8  *          Samuel Hocevar <sam@zoy.org>
9  *          Derk-Jan Hartman <hartman at videolan dot org>
10  *          Bernie Purcell <b dot purcell at adbglobal dot com>
11  *
12  * This program is free software; you can redistribute it and/or modify
13  * it under the terms of the GNU General Public License as published by
14  * the Free Software Foundation; either version 2 of the License, or
15  * (at your option) any later version.
16  *
17  * This program is distributed in the hope that it will be useful,
18  * but WITHOUT ANY WARRANTY; without even the implied warranty of
19  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
20  * GNU General Public License for more details.
21  *
22  * You should have received a copy of the GNU General Public License
23  * along with this program; if not, write to the Free Software
24  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
25  *****************************************************************************/
26
27 #include "subsdec.h"
28
29
30 void ParseSSAString( decoder_t *p_dec,
31                      char *psz_subtitle,
32                      subpicture_t *p_spu_in )
33 {
34     /* We expect MKV formatted SSA:
35      * ReadOrder, Layer, Style, CharacterName, MarginL, MarginR,
36      * MarginV, Effect, Text */
37     decoder_sys_t   *p_sys = p_dec->p_sys;
38     subpicture_t    *p_spu = p_spu_in;
39     ssa_style_t     *p_style = NULL;
40     char            *psz_new_subtitle = NULL;
41     char            *psz_buffer_sub = NULL;
42     char            *psz_style = NULL;
43     char            *psz_style_start = NULL;
44     char            *psz_style_end = NULL;
45     int             i_text = 0, i_comma = 0, i_strlen = 0, i;
46     int             i_margin_l = 0, i_margin_r = 0, i_margin_v = 0;
47
48     psz_buffer_sub = psz_subtitle;
49
50     p_spu->p_region->psz_html = NULL;
51
52     i_comma = 0;
53     while( i_comma < 8 && *psz_buffer_sub != '\0' )
54     {
55         if( *psz_buffer_sub == ',' )
56         {
57             i_comma++;
58             if( i_comma == 2 )
59                 psz_style_start = &psz_buffer_sub[1];
60             else if( i_comma == 3 )
61                 psz_style_end = &psz_buffer_sub[0];
62             else if( i_comma == 4 )
63                 i_margin_l = (int)strtol( &psz_buffer_sub[1], NULL, 10 );
64             else if( i_comma == 5 )
65                 i_margin_r = (int)strtol( &psz_buffer_sub[1], NULL, 10 );
66             else if( i_comma == 6 )
67                 i_margin_v = (int)strtol( &psz_buffer_sub[1], NULL, 10 );
68         }
69         psz_buffer_sub++;
70     }
71
72     if( *psz_buffer_sub == '\0' && i_comma == 8 )
73     {
74         msg_Dbg( p_dec, "couldn't find all fields in this SSA line" );
75         return;
76     }
77
78     psz_new_subtitle = malloc( strlen( psz_buffer_sub ) + 1);
79     i_text = 0;
80     while( psz_buffer_sub[0] != '\0' )
81     {
82         if( psz_buffer_sub[0] == '\\' && psz_buffer_sub[1] == 'n' )
83         {
84             psz_new_subtitle[i_text] = ' ';
85             i_text++;
86             psz_buffer_sub += 2;
87         }
88         else if( psz_buffer_sub[0] == '\\' && psz_buffer_sub[1] == 'N' )
89         {
90             psz_new_subtitle[i_text] = '\n';
91             i_text++;
92             psz_buffer_sub += 2;
93         }
94         else if( psz_buffer_sub[0] == '{' &&
95                  psz_buffer_sub[1] == '\\' )
96         {
97             /* SSA control code */
98             while( psz_buffer_sub[0] != '\0' &&
99                    psz_buffer_sub[0] != '}' )
100             {
101                 psz_buffer_sub++;
102             }
103             psz_buffer_sub++;
104         }
105         else
106         {
107             psz_new_subtitle[i_text] = psz_buffer_sub[0];
108             i_text++;
109             psz_buffer_sub++;
110         }
111     }
112     psz_new_subtitle[i_text] = '\0';
113
114     i_strlen = __MAX( psz_style_end - psz_style_start, 0);
115     psz_style = strndup( psz_style_start, i_strlen );
116
117     for( i = 0; i < p_sys->i_ssa_styles; i++ )
118     {
119         if( !strcmp( p_sys->pp_ssa_styles[i]->psz_stylename, psz_style ) )
120             p_style = p_sys->pp_ssa_styles[i];
121     }
122     if( psz_style ) free( psz_style );
123
124     p_spu->p_region->psz_text = psz_new_subtitle;
125     if( p_style == NULL )
126     {
127         p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
128         p_spu->i_x = p_sys->i_align ? 20 : 0;
129         p_spu->i_y = 10;
130     }
131     else
132     {
133         msg_Dbg( p_dec, "style is: %s", p_style->psz_stylename);
134         p_spu->p_region->p_style = &p_style->font_style;
135         p_spu->p_region->i_align = p_style->i_align;
136         if( p_style->i_align & SUBPICTURE_ALIGN_LEFT )
137         {
138             p_spu->i_x = (i_margin_l) ? i_margin_l : p_style->i_margin_h;
139         }
140         else if( p_style->i_align & SUBPICTURE_ALIGN_RIGHT )
141         {
142             p_spu->i_x = (i_margin_r) ? i_margin_r : p_style->i_margin_h;
143         }
144         p_spu->i_y = (i_margin_v) ? i_margin_v : p_style->i_margin_v;
145     }
146 }
147
148 /*****************************************************************************
149  * ParseColor: SSA stores color in BBGGRR, in ASS it uses AABBGGRR
150  * The string value in the string can be a pure integer, or hexadecimal &HBBGGRR
151  *****************************************************************************/
152 static void ParseColor( decoder_t *p_dec,
153                         char *psz_color,
154                         int *pi_color,
155                         int *pi_alpha )
156 {
157     int i_color = 0;
158     if( !strncasecmp( psz_color, "&H", 2 ) )
159     {
160         /* textual HEX representation */
161         i_color = (int) strtol( psz_color+2, NULL, 16 );
162     }
163     else i_color = (int) strtol( psz_color, NULL, 0 );
164
165     *pi_color = 0;
166     *pi_color |= ( ( i_color & 0x000000FF ) << 16 ); /* Red */
167     *pi_color |= ( ( i_color & 0x0000FF00 ) );       /* Green */
168     *pi_color |= ( ( i_color & 0x00FF0000 ) >> 16 ); /* Blue */
169
170     if( pi_alpha != NULL )
171         *pi_alpha = ( i_color & 0xFF000000 ) >> 24;
172 }
173
174 /*****************************************************************************
175  * ParseSSAHeader: Retrieve global formatting information etc
176  *****************************************************************************/
177 void ParseSSAHeader( decoder_t *p_dec )
178 {
179     decoder_sys_t *p_sys = p_dec->p_sys;
180     char *psz_parser = NULL;
181     char *psz_header = malloc( p_dec->fmt_in.i_extra+1 );
182     int i_section_type = 1;
183
184     memcpy( psz_header, p_dec->fmt_in.p_extra, p_dec->fmt_in.i_extra );
185     psz_header[ p_dec->fmt_in.i_extra] = '\0';
186
187     /* Handle [Script Info] section */
188     psz_parser = strcasestr( psz_header, "[Script Info]" );
189     if( psz_parser == NULL ) goto eof;
190
191     psz_parser = GotoNextLine( psz_parser );
192
193     while( psz_parser[0] != '\0' )
194     {
195         int temp;
196         char buffer_text[MAX_LINE + 1];
197
198         if( psz_parser[0] == '!' || psz_parser[0] == ';' ) /* comment */;
199         else if( sscanf( psz_parser, "PlayResX: %d", &temp ) == 1 )
200             p_sys->i_original_width = ( temp > 0 ) ? temp : -1;
201         else if( sscanf( psz_parser, "PlayResY: %d", &temp ) == 1 )
202             p_sys->i_original_height = ( temp > 0 ) ? temp : -1;
203         else if( sscanf( psz_parser, "Script Type: %8192s", buffer_text ) == 1 )
204         {
205             if( !strcasecmp( buffer_text, "V4.00+" ) ) p_sys->b_ass = VLC_TRUE;
206         }
207         else if( !strncasecmp( psz_parser, "[V4 Styles]", 11 ) )
208             i_section_type = 1;
209         else if( !strncasecmp( psz_parser, "[V4+ Styles]", 12) )
210         {
211             i_section_type = 2;
212             p_sys->b_ass = VLC_TRUE;
213         }
214         else if( !strncasecmp( psz_parser, "[Events]", 8 ) )
215             i_section_type = 4;
216         else if( !strncasecmp( psz_parser, "Style:", 6 ) )
217         {
218             int i_font_size, i_bold, i_italic, i_border, i_outline, i_shadow,
219                 i_underline, i_strikeout, i_scale_x, i_scale_y, i_spacing,
220                 i_align, i_margin_l, i_margin_r, i_margin_v;
221
222             char psz_temp_stylename[MAX_LINE+1];
223             char psz_temp_fontname[MAX_LINE+1];
224             char psz_temp_color1[MAX_LINE+1];
225             char psz_temp_color2[MAX_LINE+1];
226             char psz_temp_color3[MAX_LINE+1];
227             char psz_temp_color4[MAX_LINE+1];
228
229             if( i_section_type == 1 ) /* V4 */
230             {
231                 if( sscanf( psz_parser, "Style: %8192[^,],%8192[^,],%d,%8192[^,],%8192[^,],%8192[^,],%8192[^,],%d,%d,%d,%d,%d,%d,%d,%d,%d%*[^\r\n]",
232                     psz_temp_stylename, psz_temp_fontname, &i_font_size,
233                     psz_temp_color1, psz_temp_color2, psz_temp_color3,
234                     psz_temp_color4, &i_bold, &i_italic,
235                     &i_border, &i_outline, &i_shadow, &i_align, &i_margin_l, 
236                     &i_margin_r, &i_margin_v ) == 16 )
237                 {
238                     ssa_style_t *p_style = malloc( sizeof(ssa_style_t) );
239
240                     p_style->psz_stylename = strdup( psz_temp_stylename );
241                     p_style->font_style.psz_fontname = strdup( psz_temp_fontname );
242                     p_style->font_style.i_font_size = i_font_size;
243
244                     ParseColor( p_dec, psz_temp_color1, &p_style->font_style.i_font_color, NULL );
245                     ParseColor( p_dec, psz_temp_color4, &p_style->font_style.i_shadow_color, NULL );
246                     p_style->font_style.i_outline_color = p_style->font_style.i_shadow_color;
247                     p_style->font_style.i_font_alpha = p_style->font_style.i_outline_alpha 
248                                                      = p_style->font_style.i_shadow_alpha = 0x00;
249                     p_style->font_style.i_style_flags = 0;
250                     if( i_bold ) p_style->font_style.i_style_flags |= STYLE_BOLD;
251                     if( i_italic ) p_style->font_style.i_style_flags |= STYLE_ITALIC;
252
253                     if( i_border == 1 ) 
254                         p_style->font_style.i_style_flags |= (STYLE_ITALIC | STYLE_OUTLINE);
255                     else if( i_border == 3 )
256                     {
257                         p_style->font_style.i_style_flags |= STYLE_BACKGROUND;
258                         p_style->font_style.i_background_color = p_style->font_style.i_shadow_color;
259                         p_style->font_style.i_background_alpha = p_style->font_style.i_shadow_alpha;
260                     }
261                     p_style->font_style.i_shadow_width = i_shadow;
262                     p_style->font_style.i_outline_width = i_outline;
263
264                     p_style->i_align = 0;
265                     if( i_align == 1 || i_align == 5 || i_align == 9 )
266                         p_style->i_align |= SUBPICTURE_ALIGN_LEFT;
267                     if( i_align == 3 || i_align == 7 || i_align == 11 )
268                         p_style->i_align |= SUBPICTURE_ALIGN_RIGHT;
269                     if( i_align < 4 )
270                         p_style->i_align |= SUBPICTURE_ALIGN_BOTTOM;
271                     else if( i_align < 8 ) 
272                         p_style->i_align |= SUBPICTURE_ALIGN_TOP;
273
274                     p_style->i_margin_h = ( p_style->i_align & SUBPICTURE_ALIGN_RIGHT ) ?
275                                                         i_margin_r : i_margin_l;
276                     p_style->i_margin_v = i_margin_v;
277                     p_style->i_margin_percent_h = 0;
278                     p_style->i_margin_percent_v = 0;
279
280                     p_style->font_style.i_karaoke_background_color = 0xffffff;
281                     p_style->font_style.i_karaoke_background_alpha = 0xff;
282
283                     TAB_APPEND( p_sys->i_ssa_styles, p_sys->pp_ssa_styles, p_style );
284                 }
285                 else msg_Warn( p_dec, "SSA v4 styleline parsing failed" );
286             }
287             else if( i_section_type == 2 ) /* V4+ */
288             {
289                 /* Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour,
290                    Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline,
291                    Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
292                 */
293                 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]",
294                     psz_temp_stylename, psz_temp_fontname, &i_font_size,
295                     psz_temp_color1, psz_temp_color2, psz_temp_color3, psz_temp_color4, &i_bold, &i_italic,
296                     &i_underline, &i_strikeout, &i_scale_x, &i_scale_y, &i_spacing, &i_border, &i_outline,
297                     &i_shadow, &i_align, &i_margin_l, &i_margin_r, &i_margin_v ) == 21 )
298                 {
299                     ssa_style_t *p_style = malloc( sizeof(ssa_style_t) );
300
301                     p_style->psz_stylename = strdup( psz_temp_stylename );
302                     p_style->font_style.psz_fontname = strdup( psz_temp_fontname );
303                     p_style->font_style.i_font_size = i_font_size;
304                     msg_Dbg( p_dec, psz_temp_color1 );
305                     ParseColor( p_dec, psz_temp_color1, &p_style->font_style.i_font_color,
306                                 &p_style->font_style.i_font_alpha );
307                     ParseColor( p_dec, psz_temp_color3, &p_style->font_style.i_outline_color,
308                                 &p_style->font_style.i_outline_alpha );
309                     ParseColor( p_dec, psz_temp_color4, &p_style->font_style.i_shadow_color,
310                                 &p_style->font_style.i_shadow_alpha );
311
312                     p_style->font_style.i_style_flags = 0;
313                     if( i_bold ) p_style->font_style.i_style_flags |= STYLE_BOLD;
314                     if( i_italic ) p_style->font_style.i_style_flags |= STYLE_ITALIC;
315                     if( i_underline ) p_style->font_style.i_style_flags |= STYLE_UNDERLINE;
316                     if( i_strikeout ) p_style->font_style.i_style_flags |= STYLE_STRIKEOUT;
317                     if( i_border == 1 ) p_style->font_style.i_style_flags |= (STYLE_ITALIC | STYLE_OUTLINE);
318                     else if( i_border == 3 )
319                     {
320                         p_style->font_style.i_style_flags |= STYLE_BACKGROUND;
321                         p_style->font_style.i_background_color = p_style->font_style.i_shadow_color;
322                         p_style->font_style.i_background_alpha = p_style->font_style.i_shadow_alpha;
323                     }
324                     p_style->font_style.i_shadow_width  = ( i_border == 1 ) ? i_shadow : 0;
325                     p_style->font_style.i_outline_width = ( i_border == 1 ) ? i_outline : 0;
326                     p_style->font_style.i_spacing = i_spacing;
327                     //p_style->font_style.f_angle = f_angle;
328
329                     p_style->i_align = 0;
330                     if( i_align == 0x1 || i_align == 0x4 || i_align == 0x7 )
331                         p_style->i_align |= SUBPICTURE_ALIGN_LEFT;
332                     if( i_align == 0x3 || i_align == 0x6 || i_align == 0x9 )
333                         p_style->i_align |= SUBPICTURE_ALIGN_RIGHT;
334                     if( i_align == 0x7 || i_align == 0x8 || i_align == 0x9 )
335                         p_style->i_align |= SUBPICTURE_ALIGN_TOP;
336                     if( i_align == 0x1 || i_align == 0x2 || i_align == 0x3 )
337                         p_style->i_align |= SUBPICTURE_ALIGN_BOTTOM;
338                     p_style->i_margin_h = ( p_style->i_align & SUBPICTURE_ALIGN_RIGHT ) ?
339                                             i_margin_r : i_margin_l;
340                     p_style->i_margin_v = i_margin_v;
341                     p_style->i_margin_percent_h = 0;
342                     p_style->i_margin_percent_v = 0;
343
344                     p_style->font_style.i_karaoke_background_color = 0xffffff;
345                     p_style->font_style.i_karaoke_background_alpha = 0xff;
346
347                     /*TODO: Ignored: angle i_scale_x|y (fontscaling), i_encoding */
348                     TAB_APPEND( p_sys->i_ssa_styles, p_sys->pp_ssa_styles, p_style );
349                 }
350                 else msg_Dbg( p_dec, "SSA V4+ styleline parsing failed" );
351             }
352         }
353         psz_parser = GotoNextLine( psz_parser );
354     }
355
356 eof:
357     if( psz_header ) free( psz_header );
358     return;
359 }
360
361