?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
demux: libavi: remove useless member
[vlc.git] / modules / demux / subtitle.c
blobb6599c8269632a5346786aef09b81791d5cb9579
1 /*****************************************************************************
2 * subtitle.c: Demux for subtitle text files.
3 *****************************************************************************
4 * Copyright (C) 1999-2007 VLC authors and VideoLAN
5 * $Id$
6 *
7 * Authors: Laurent Aimar <fenrir@via.ecp.fr>
8 * Derk-Jan Hartman <hartman at videolan dot org>
9 * Jean-Baptiste Kempf <jb@videolan.org>
10 *
11 * This program is free software; you can redistribute it and/or modify it
12 * under the terms of the GNU Lesser General Public License as published by
13 * the Free Software Foundation; either version 2.1 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 Lesser General Public License for more details.
20 *
21 * You should have received a copy of the GNU Lesser General Public License
22 * along with this program; if not, write to the Free Software Foundation,
23 * Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
24 *****************************************************************************/
25
26 /*****************************************************************************
27 * Preamble
28 *****************************************************************************/
29
30 #ifdef HAVE_CONFIG_H
31 # include "config.h"
32 #endif
33
34 #include <vlc_common.h>
35 #include <vlc_plugin.h>
36 #include <vlc_input.h>
37 #include <vlc_memory.h>
38
39 #include <ctype.h>
40 #include <math.h>
41 #include <assert.h>
42
43 #include <vlc_demux.h>
44 #include <vlc_charset.h>
45
46 /*****************************************************************************
47 * Module descriptor
48 *****************************************************************************/
49 static int Open ( vlc_object_t *p_this );
50 static void Close( vlc_object_t *p_this );
51
52 #define SUB_DELAY_LONGTEXT \
53 N_("Apply a delay to all subtitles (in 1/10s, eg 100 means 10s).")
54 #define SUB_FPS_LONGTEXT \
55 N_("Override the normal frames per second settings. " \
56 "This will only work with MicroDVD and SubRIP (SRT) subtitles.")
57 #define SUB_TYPE_LONGTEXT \
58 N_("Force the subtiles format. Selecting \"auto\" means autodetection and should always work.")
59 #define SUB_DESCRIPTION_LONGTEXT \
60 N_("Override the default track description.")
61
62 static const char *const ppsz_sub_type[] =
63 {
64 "auto", "microdvd", "subrip", "subviewer", "ssa1",
65 "ssa2-4", "ass", "vplayer", "sami", "dvdsubtitle", "mpl2",
66 "aqt", "pjs", "mpsub", "jacosub", "psb", "realtext", "dks",
67 "subviewer1", "sbv"
68 };
69
70 vlc_module_begin ()
71 set_shortname( N_("Subtitles"))
72 set_description( N_("Text subtitle parser") )
73 set_capability( "demux", 0 )
74 set_category( CAT_INPUT )
75 set_subcategory( SUBCAT_INPUT_DEMUX )
76 add_float( "sub-fps", 0.0,
77 N_("Frames per Second"),
78 SUB_FPS_LONGTEXT, true )
79 add_integer( "sub-delay", 0,
80 N_("Subtitle delay"),
81 SUB_DELAY_LONGTEXT, true )
82 add_string( "sub-type", "auto", N_("Subtitle format"),
83 SUB_TYPE_LONGTEXT, true )
84 change_string_list( ppsz_sub_type, ppsz_sub_type )
85 add_string( "sub-description", NULL, N_("Subtitle description"),
86 SUB_DESCRIPTION_LONGTEXT, true )
87 set_callbacks( Open, Close )
88
89 add_shortcut( "subtitle" )
90 vlc_module_end ()
91
92 /*****************************************************************************
93 * Prototypes:
94 *****************************************************************************/
95 enum subtitle_type_e
96 {
97 SUB_TYPE_UNKNOWN = -1,
98 SUB_TYPE_MICRODVD,
99 SUB_TYPE_SUBRIP,
100 SUB_TYPE_SSA1,
101 SUB_TYPE_SSA2_4,
102 SUB_TYPE_ASS,
103 SUB_TYPE_VPLAYER,
104 SUB_TYPE_SAMI,
105 SUB_TYPE_SUBVIEWER, /* SUBVIEWER 2 */
106 SUB_TYPE_DVDSUBTITLE, /* Mplayer calls it subviewer2 */
107 SUB_TYPE_MPL2,
108 SUB_TYPE_AQT,
109 SUB_TYPE_PJS,
110 SUB_TYPE_MPSUB,
111 SUB_TYPE_JACOSUB,
112 SUB_TYPE_PSB,
113 SUB_TYPE_RT,
114 SUB_TYPE_DKS,
115 SUB_TYPE_SUBVIEW1, /* SUBVIEWER 1 - mplayer calls it subrip09,
116 and Gnome subtitles SubViewer 1.0 */
117 SUB_TYPE_SBV,
118 SUB_TYPE_SCC, /* Scenarist Closed Caption */
119 };
120
121 typedef struct
122 {
123 size_t i_line_count;
124 size_t i_line;
125 char **line;
126 } text_t;
127
128 static int TextLoad( text_t *, stream_t *s );
129 static void TextUnload( text_t * );
130
131 typedef struct
132 {
133 int64_t i_start;
134 int64_t i_stop;
135
136 char *psz_text;
137 } subtitle_t;
138
139 typedef struct
140 {
141 enum subtitle_type_e i_type;
142 int64_t i_microsecperframe;
143
144 char *psz_header; /* SSA */
145
146 struct
147 {
148 bool b_inited;
149
150 int i_comment;
151 int i_time_resolution;
152 int i_time_shift;
153 } jss;
154
155 struct
156 {
157 bool b_inited;
158
159 float f_total;
160 float f_factor;
161 } mpsub;
162
163 struct
164 {
165 const char *psz_start;
166 } sami;
167
168 } subs_properties_t;
169
170 struct demux_sys_t
171 {
172 es_out_id_t *es;
173 bool b_slave;
174 bool b_first_time;
175
176 int64_t i_next_demux_date;
177
178 struct
179 {
180 subtitle_t *p_array;
181 size_t i_count;
182 size_t i_current;
183 } subtitles;
184
185 int64_t i_length;
186
187 /* */
188 subs_properties_t props;
189
190 block_t * (*pf_convert)( const subtitle_t * );
191 };
192
193 static int ParseMicroDvd ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
194 static int ParseSubRip ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
195 static int ParseSubViewer ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
196 static int ParseSSA ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
197 static int ParseVplayer ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
198 static int ParseSami ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
199 static int ParseDVDSubtitle( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
200 static int ParseMPL2 ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
201 static int ParseAQT ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
202 static int ParsePJS ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
203 static int ParseMPSub ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
204 static int ParseJSS ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
205 static int ParsePSB ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
206 static int ParseRealText ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
207 static int ParseDKS ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
208 static int ParseSubViewer1 ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
209 static int ParseCommonSBV ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
210 static int ParseSCC ( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t *, size_t );
211
212 static const struct
213 {
214 const char *psz_type_name;
215 int i_type;
216 const char *psz_name;
217 int (*pf_read)( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t*, size_t );
218 } sub_read_subtitle_function [] =
219 {
220 { "microdvd", SUB_TYPE_MICRODVD, "MicroDVD", ParseMicroDvd },
221 { "subrip", SUB_TYPE_SUBRIP, "SubRIP", ParseSubRip },
222 { "subviewer", SUB_TYPE_SUBVIEWER, "SubViewer", ParseSubViewer },
223 { "ssa1", SUB_TYPE_SSA1, "SSA-1", ParseSSA },
224 { "ssa2-4", SUB_TYPE_SSA2_4, "SSA-2/3/4", ParseSSA },
225 { "ass", SUB_TYPE_ASS, "SSA/ASS", ParseSSA },
226 { "vplayer", SUB_TYPE_VPLAYER, "VPlayer", ParseVplayer },
227 { "sami", SUB_TYPE_SAMI, "SAMI", ParseSami },
228 { "dvdsubtitle",SUB_TYPE_DVDSUBTITLE, "DVDSubtitle", ParseDVDSubtitle },
229 { "mpl2", SUB_TYPE_MPL2, "MPL2", ParseMPL2 },
230 { "aqt", SUB_TYPE_AQT, "AQTitle", ParseAQT },
231 { "pjs", SUB_TYPE_PJS, "PhoenixSub", ParsePJS },
232 { "mpsub", SUB_TYPE_MPSUB, "MPSub", ParseMPSub },
233 { "jacosub", SUB_TYPE_JACOSUB, "JacoSub", ParseJSS },
234 { "psb", SUB_TYPE_PSB, "PowerDivx", ParsePSB },
235 { "realtext", SUB_TYPE_RT, "RealText", ParseRealText },
236 { "dks", SUB_TYPE_DKS, "DKS", ParseDKS },
237 { "subviewer1", SUB_TYPE_SUBVIEW1, "Subviewer 1", ParseSubViewer1 },
238 { "sbv", SUB_TYPE_SBV, "SBV", ParseCommonSBV },
239 { "scc", SUB_TYPE_SCC, "SCC", ParseSCC },
240 { NULL, SUB_TYPE_UNKNOWN, "Unknown", NULL }
241 };
242 /* When adding support for more formats, be sure to add their file extension
243 * to src/input/subtitles.c to enable auto-detection.
244 */
245
246 static int Demux( demux_t * );
247 static int Control( demux_t *, int, va_list );
248
249 static void Fix( demux_t * );
250 static char * get_language_from_filename( const char * );
251
252 /*****************************************************************************
253 * Decoder format output function
254 *****************************************************************************/
255
256 static block_t *ToTextBlock( const subtitle_t *p_subtitle )
257 {
258 block_t *p_block;
259 size_t i_len = strlen( p_subtitle->psz_text ) + 1;
260
261 if( i_len <= 1 || !(p_block = block_Alloc( i_len )) )
262 return NULL;
263
264 memcpy( p_block->p_buffer, p_subtitle->psz_text, i_len );
265
266 return p_block;
267 }
268
269 static block_t *ToEIA608Block( const subtitle_t *p_subtitle )
270 {
271 block_t *p_block;
272 const size_t i_len = strlen( p_subtitle->psz_text );
273 const size_t i_block = (1 + i_len / 5) * 3;
274
275 if( i_len < 4 || !(p_block = block_Alloc( i_block )) )
276 return NULL;
277
278 p_block->i_buffer = 0;
279
280 char *saveptr = NULL;
281 char *psz_tok = strtok_r( p_subtitle->psz_text, " ", &saveptr );
282 unsigned a, b;
283 while( psz_tok &&
284 sscanf( psz_tok, "%2x%2x", &a, &b ) == 2 &&
285 i_block - p_block->i_buffer >= 3 )
286 {
287 uint8_t *p_data = &p_block->p_buffer[p_block->i_buffer];
288 p_data[0] = 0xFC;
289 p_data[1] = a;
290 p_data[2] = b;
291 p_block->i_buffer += 3;
292 psz_tok = strtok_r( NULL, " ", &saveptr );
293 }
294
295 return p_block;
296 }
297
298 /*****************************************************************************
299 * Module initializer
300 *****************************************************************************/
301 static int Open ( vlc_object_t *p_this )
302 {
303 demux_t *p_demux = (demux_t*)p_this;
304 demux_sys_t *p_sys;
305 es_format_t fmt;
306 float f_fps;
307 char *psz_type;
308 int (*pf_read)( vlc_object_t *, subs_properties_t *, text_t *, subtitle_t*, size_t );
309
310 if( !p_demux->obj.force )
311 {
312 msg_Dbg( p_demux, "subtitle demux discarded" );
313 return VLC_EGENERIC;
314 }
315
316 p_demux->pf_demux = Demux;
317 p_demux->pf_control = Control;
318 p_demux->p_sys = p_sys = malloc( sizeof( demux_sys_t ) );
319 if( p_sys == NULL )
320 return VLC_ENOMEM;
321
322 p_sys->b_slave = false;
323 p_sys->b_first_time = true;
324 p_sys->i_next_demux_date = 0;
325
326 p_sys->pf_convert = ToTextBlock;
327
328 p_sys->subtitles.i_current= 0;
329 p_sys->subtitles.i_count = 0;
330 p_sys->subtitles.p_array = NULL;
331
332 p_sys->props.psz_header = NULL;
333 p_sys->props.i_microsecperframe = 40000;
334 p_sys->props.jss.b_inited = false;
335 p_sys->props.mpsub.b_inited = false;
336 p_sys->props.sami.psz_start = NULL;
337
338 /* Get the FPS */
339 f_fps = var_CreateGetFloat( p_demux, "sub-original-fps" ); /* FIXME */
340 if( f_fps >= 1.f )
341 p_sys->props.i_microsecperframe = llroundf( 1000000.f / f_fps );
342
343 msg_Dbg( p_demux, "Movie fps: %f", (double) f_fps );
344
345 /* Check for override of the fps */
346 f_fps = var_CreateGetFloat( p_demux, "sub-fps" );
347 if( f_fps >= 1.f )
348 {
349 p_sys->props.i_microsecperframe = llroundf( 1000000.f / f_fps );
350 msg_Dbg( p_demux, "Override subtitle fps %f", (double) f_fps );
351 }
352
353 /* Get or probe the type */
354 p_sys->props.i_type = SUB_TYPE_UNKNOWN;
355 psz_type = var_CreateGetString( p_demux, "sub-type" );
356 if( psz_type && *psz_type )
357 {
358 for( int i = 0; ; i++ )
359 {
360 if( sub_read_subtitle_function[i].psz_type_name == NULL )
361 break;
362
363 if( !strcmp( sub_read_subtitle_function[i].psz_type_name,
364 psz_type ) )
365 {
366 p_sys->props.i_type = sub_read_subtitle_function[i].i_type;
367 break;
368 }
369 }
370 }
371 free( psz_type );
372
373 #ifndef NDEBUG
374 const uint64_t i_start_pos = vlc_stream_Tell( p_demux->s );
375 #endif
376
377 size_t i_peek;
378 const uint8_t *p_peek;
379 if( vlc_stream_Peek( p_demux->s, &p_peek, 16 ) < 16 )
380 {
381 free( p_sys );
382 return VLC_EGENERIC;
383 }
384
385 enum
386 {
387 UTF8BOM,
388 UTF16LE,
389 UTF16BE,
390 NOBOM,
391 } e_bom = NOBOM;
392 const char *psz_bom = NULL;
393
394 i_peek = 4096;
395 /* Detect Unicode while skipping the UTF-8 Byte Order Mark */
396 if( !memcmp( p_peek, "\xEF\xBB\xBF", 3 ) )
397 {
398 e_bom = UTF8BOM;
399 psz_bom = "UTF-8";
400 }
401 else if( !memcmp( p_peek, "\xFF\xFE", 2 ) )
402 {
403 e_bom = UTF16LE;
404 psz_bom = "UTF-16LE";
405 i_peek *= 2;
406 }
407 else if( !memcmp( p_peek, "\xFE\xFF", 2 ) )
408 {
409 e_bom = UTF16BE;
410 psz_bom = "UTF-16BE";
411 i_peek *= 2;
412 }
413
414 if( e_bom != NOBOM )
415 msg_Dbg( p_demux, "detected %s Byte Order Mark", psz_bom );
416
417 i_peek = vlc_stream_Peek( p_demux->s, &p_peek, i_peek );
418 if( unlikely(i_peek < 16) )
419 {
420 free( p_sys );
421 return VLC_EGENERIC;
422 }
423
424 stream_t *p_probestream = NULL;
425 if( e_bom != UTF8BOM && e_bom != NOBOM )
426 {
427 if( i_peek > 16 )
428 {
429 char *p_outbuf = FromCharset( psz_bom, p_peek, i_peek );
430 if( p_outbuf != NULL )
431 p_probestream = vlc_stream_MemoryNew( p_demux, (uint8_t *)p_outbuf,
432 strlen( p_outbuf ),
433 false ); /* free p_outbuf on release */
434 }
435 }
436 else
437 {
438 const size_t i_skip = (e_bom == UTF8BOM) ? 3 : 0;
439 p_probestream = vlc_stream_MemoryNew( p_demux, (uint8_t *) &p_peek[i_skip],
440 i_peek - i_skip, true );
441 }
442
443 if( p_probestream == NULL )
444 {
445 free( p_sys );
446 return VLC_EGENERIC;
447 }
448
449 /* Probe if unknown type */
450 if( p_sys->props.i_type == SUB_TYPE_UNKNOWN )
451 {
452 int i_try;
453 char *s = NULL;
454
455 msg_Dbg( p_demux, "autodetecting subtitle format" );
456 for( i_try = 0; i_try < 256; i_try++ )
457 {
458 int i_dummy;
459 char p_dummy;
460
461 if( (s = vlc_stream_ReadLine( p_probestream ) ) == NULL )
462 break;
463
464 if( strcasestr( s, "<SAMI>" ) )
465 {
466 p_sys->props.i_type = SUB_TYPE_SAMI;
467 break;
468 }
469 else if( sscanf( s, "{%d}{%d}", &i_dummy, &i_dummy ) == 2 ||
470 sscanf( s, "{%d}{}", &i_dummy ) == 1)
471 {
472 p_sys->props.i_type = SUB_TYPE_MICRODVD;
473 break;
474 }
475 else if( sscanf( s, "%d:%d:%d,%d --> %d:%d:%d,%d",
476 &i_dummy,&i_dummy,&i_dummy,&i_dummy,
477 &i_dummy,&i_dummy,&i_dummy,&i_dummy ) == 8 ||
478 sscanf( s, "%d:%d:%d --> %d:%d:%d,%d",
479 &i_dummy,&i_dummy,&i_dummy,&i_dummy,
480 &i_dummy,&i_dummy,&i_dummy ) == 7 ||
481 sscanf( s, "%d:%d:%d,%d --> %d:%d:%d",
482 &i_dummy,&i_dummy,&i_dummy,&i_dummy,
483 &i_dummy,&i_dummy,&i_dummy ) == 7 ||
484 sscanf( s, "%d:%d:%d.%d --> %d:%d:%d.%d",
485 &i_dummy,&i_dummy,&i_dummy,&i_dummy,
486 &i_dummy,&i_dummy,&i_dummy,&i_dummy ) == 8 ||
487 sscanf( s, "%d:%d:%d --> %d:%d:%d.%d",
488 &i_dummy,&i_dummy,&i_dummy,&i_dummy,
489 &i_dummy,&i_dummy,&i_dummy ) == 7 ||
490 sscanf( s, "%d:%d:%d.%d --> %d:%d:%d",
491 &i_dummy,&i_dummy,&i_dummy,&i_dummy,
492 &i_dummy,&i_dummy,&i_dummy ) == 7 ||
493 sscanf( s, "%d:%d:%d --> %d:%d:%d",
494 &i_dummy,&i_dummy,&i_dummy,
495 &i_dummy,&i_dummy,&i_dummy ) == 6 )
496 {
497 p_sys->props.i_type = SUB_TYPE_SUBRIP;
498 break;
499 }
500 else if( !strncasecmp( s, "!: This is a Sub Station Alpha v1", 33 ) )
501 {
502 p_sys->props.i_type = SUB_TYPE_SSA1;
503 break;
504 }
505 else if( !strncasecmp( s, "ScriptType: v4.00+", 18 ) )
506 {
507 p_sys->props.i_type = SUB_TYPE_ASS;
508 break;
509 }
510 else if( !strncasecmp( s, "ScriptType: v4.00", 17 ) )
511 {
512 p_sys->props.i_type = SUB_TYPE_SSA2_4;
513 break;
514 }
515 else if( !strncasecmp( s, "Dialogue: Marked", 16 ) )
516 {
517 p_sys->props.i_type = SUB_TYPE_SSA2_4;
518 break;
519 }
520 else if( !strncasecmp( s, "Dialogue:", 9 ) )
521 {
522 p_sys->props.i_type = SUB_TYPE_ASS;
523 break;
524 }
525 else if( strcasestr( s, "[INFORMATION]" ) )
526 {
527 p_sys->props.i_type = SUB_TYPE_SUBVIEWER; /* I hope this will work */
528 break;
529 }
530 else if( sscanf( s, "%d:%d:%d.%d %d:%d:%d",
531 &i_dummy, &i_dummy, &i_dummy, &i_dummy,
532 &i_dummy, &i_dummy, &i_dummy ) == 7 ||
533 sscanf( s, "@%d @%d", &i_dummy, &i_dummy) == 2)
534 {
535 p_sys->props.i_type = SUB_TYPE_JACOSUB;
536 break;
537 }
538 else if( sscanf( s, "%d:%d:%d.%d,%d:%d:%d.%d",
539 &i_dummy, &i_dummy, &i_dummy, &i_dummy,
540 &i_dummy, &i_dummy, &i_dummy, &i_dummy ) == 8 )
541 {
542 p_sys->props.i_type = SUB_TYPE_SBV;
543 break;
544 }
545 else if( sscanf( s, "%d:%d:%d:", &i_dummy, &i_dummy, &i_dummy ) == 3 ||
546 sscanf( s, "%d:%d:%d ", &i_dummy, &i_dummy, &i_dummy ) == 3 )
547 {
548 p_sys->props.i_type = SUB_TYPE_VPLAYER;
549 break;
550 }
551 else if( sscanf( s, "{T %d:%d:%d:%d", &i_dummy, &i_dummy,
552 &i_dummy, &i_dummy ) == 4 )
553 {
554 p_sys->props.i_type = SUB_TYPE_DVDSUBTITLE;
555 break;
556 }
557 else if( sscanf( s, "[%d:%d:%d]%c",
558 &i_dummy, &i_dummy, &i_dummy, &p_dummy ) == 4 )
559 {
560 p_sys->props.i_type = SUB_TYPE_DKS;
561 break;
562 }
563 else if( strstr( s, "*** START SCRIPT" ) )
564 {
565 p_sys->props.i_type = SUB_TYPE_SUBVIEW1;
566 break;
567 }
568 else if( sscanf( s, "[%d][%d]", &i_dummy, &i_dummy ) == 2 ||
569 sscanf( s, "[%d][]", &i_dummy ) == 1)
570 {
571 p_sys->props.i_type = SUB_TYPE_MPL2;
572 break;
573 }
574 else if( sscanf (s, "FORMAT=%d", &i_dummy) == 1 ||
575 ( sscanf (s, "FORMAT=TIM%c", &p_dummy) == 1
576 && p_dummy =='E' ) )
577 {
578 p_sys->props.i_type = SUB_TYPE_MPSUB;
579 break;
580 }
581 else if( sscanf( s, "-->> %d", &i_dummy) == 1 )
582 {
583 p_sys->props.i_type = SUB_TYPE_AQT;
584 break;
585 }
586 else if( sscanf( s, "%d,%d,", &i_dummy, &i_dummy ) == 2 )
587 {
588 p_sys->props.i_type = SUB_TYPE_PJS;
589 break;
590 }
591 else if( sscanf( s, "{%d:%d:%d}",
592 &i_dummy, &i_dummy, &i_dummy ) == 3 )
593 {
594 p_sys->props.i_type = SUB_TYPE_PSB;
595 break;
596 }
597 else if( strcasestr( s, "<time" ) )
598 {
599 p_sys->props.i_type = SUB_TYPE_RT;
600 break;
601 }
602 else if( !strncasecmp( s, "WEBVTT",6 ) )
603 {
604 /* FAIL */
605 break;
606 }
607 else if( !strncasecmp( s, "Scenarist_SCC V1.0", 18 ) )
608 {
609 p_sys->props.i_type = SUB_TYPE_SCC;
610 p_sys->pf_convert = ToEIA608Block;
611 break;
612 }
613
614 free( s );
615 s = NULL;
616 }
617
618 free( s );
619 }
620
621 vlc_stream_Delete( p_probestream );
622
623 /* Quit on unknown subtitles */
624 if( p_sys->props.i_type == SUB_TYPE_UNKNOWN )
625 {
626 #ifndef NDEBUG
627 /* Ensure it will work with non seekable streams */
628 assert( i_start_pos == vlc_stream_Tell( p_demux->s ) );
629 #endif
630 msg_Warn( p_demux, "failed to recognize subtitle type" );
631 free( p_sys );
632 return VLC_EGENERIC;
633 }
634
635 for( int i = 0; ; i++ )
636 {
637 if( sub_read_subtitle_function[i].i_type == p_sys->props.i_type )
638 {
639 msg_Dbg( p_demux, "detected %s format",
640 sub_read_subtitle_function[i].psz_name );
641 pf_read = sub_read_subtitle_function[i].pf_read;
642 break;
643 }
644 }
645
646 msg_Dbg( p_demux, "loading all subtitles..." );
647
648 if( e_bom == UTF8BOM && /* skip BOM */
649 vlc_stream_Read( p_demux->s, NULL, 3 ) != 3 )
650 {
651 Close( p_this );
652 return VLC_EGENERIC;
653 }
654
655 /* Load the whole file */
656 text_t txtlines;
657 TextLoad( &txtlines, p_demux->s );
658
659 /* Parse it */
660 for( size_t i_max = 0; i_max < SIZE_MAX - 500 * sizeof(subtitle_t); )
661 {
662 if( p_sys->subtitles.i_count >= i_max )
663 {
664 i_max += 500;
665 subtitle_t *p_realloc = realloc( p_sys->subtitles.p_array, sizeof(subtitle_t) * i_max );
666 if( p_realloc == NULL )
667 {
668 TextUnload( &txtlines );
669 Close( p_this );
670 return VLC_ENOMEM;
671 }
672 p_sys->subtitles.p_array = p_realloc;
673 }
674
675 if( pf_read( VLC_OBJECT(p_demux), &p_sys->props, &txtlines,
676 &p_sys->subtitles.p_array[p_sys->subtitles.i_count],
677 p_sys->subtitles.i_count ) )
678 break;
679
680 p_sys->subtitles.i_count++;
681 }
682 /* Unload */
683 TextUnload( &txtlines );
684
685 msg_Dbg(p_demux, "loaded %zu subtitles", p_sys->subtitles.i_count );
686
687 /* Fix subtitle (order and time) *** */
688 p_sys->subtitles.i_current = 0;
689 p_sys->i_length = 0;
690 if( p_sys->subtitles.i_count > 0 )
691 p_sys->i_length = p_sys->subtitles.p_array[p_sys->subtitles.i_count-1].i_stop;
692
693 /* *** add subtitle ES *** */
694 if( p_sys->props.i_type == SUB_TYPE_SSA1 ||
695 p_sys->props.i_type == SUB_TYPE_SSA2_4 ||
696 p_sys->props.i_type == SUB_TYPE_ASS )
697 {
698 Fix( p_demux );
699 es_format_Init( &fmt, SPU_ES, VLC_CODEC_SSA );
700 }
701 else if( p_sys->props.i_type == SUB_TYPE_SCC )
702 {
703 es_format_Init( &fmt, SPU_ES, VLC_CODEC_CEA608 );
704 fmt.subs.cc.i_reorder_depth = -1;
705 }
706 else
707 es_format_Init( &fmt, SPU_ES, VLC_CODEC_SUBT );
708
709 /* Stupid language detection in the filename */
710 char * psz_language = get_language_from_filename( p_demux->psz_file );
711
712 if( psz_language )
713 {
714 fmt.psz_language = psz_language;
715 msg_Dbg( p_demux, "detected language %s of subtitle: %s", psz_language,
716 p_demux->psz_location );
717 }
718
719 if( psz_bom )
720 fmt.subs.psz_encoding = strdup( psz_bom );
721 char *psz_description = var_InheritString( p_demux, "sub-description" );
722 if( psz_description && *psz_description )
723 fmt.psz_description = psz_description;
724 else
725 free( psz_description );
726 if( p_sys->props.psz_header != NULL &&
727 (fmt.p_extra = strdup( p_sys->props.psz_header )) )
728 {
729 fmt.i_extra = strlen( p_sys->props.psz_header ) + 1;
730 }
731
732 p_sys->es = es_out_Add( p_demux->out, &fmt );
733 es_format_Clean( &fmt );
734 if( p_sys->es == NULL )
735 {
736 Close( p_this );
737 return VLC_EGENERIC;
738 }
739
740 return VLC_SUCCESS;
741 }
742
743 /*****************************************************************************
744 * Close: Close subtitle demux
745 *****************************************************************************/
746 static void Close( vlc_object_t *p_this )
747 {
748 demux_t *p_demux = (demux_t*)p_this;
749 demux_sys_t *p_sys = p_demux->p_sys;
750
751 for( size_t i = 0; i < p_sys->subtitles.i_count; i++ )
752 free( p_sys->subtitles.p_array[i].psz_text );
753 free( p_sys->subtitles.p_array );
754 free( p_sys->props.psz_header );
755
756 free( p_sys );
757 }
758
759 /*****************************************************************************
760 * Control:
761 *****************************************************************************/
762 static int Control( demux_t *p_demux, int i_query, va_list args )
763 {
764 demux_sys_t *p_sys = p_demux->p_sys;
765 int64_t *pi64, i64;
766 double *pf, f;
767
768 switch( i_query )
769 {
770 case DEMUX_CAN_SEEK:
771 *va_arg( args, bool * ) = true;
772 return VLC_SUCCESS;
773
774 case DEMUX_GET_LENGTH:
775 pi64 = va_arg( args, int64_t * );
776 *pi64 = p_sys->i_length;
777 return VLC_SUCCESS;
778
779 case DEMUX_GET_TIME:
780 pi64 = va_arg( args, int64_t * );
781 *pi64 = p_sys->i_next_demux_date - var_GetInteger( p_demux->obj.parent, "spu-delay" );
782 if( *pi64 < 0 )
783 *pi64 = p_sys->i_next_demux_date;
784 return VLC_SUCCESS;
785
786 case DEMUX_SET_TIME:
787 i64 = va_arg( args, int64_t );
788 for( size_t i = 0; i + 1< p_sys->subtitles.i_count; i++ )
789 {
790 if( p_sys->subtitles.p_array[i + 1].i_start >= i64 )
791 {
792 p_sys->subtitles.i_current = i;
793 p_sys->i_next_demux_date = i64;
794 p_sys->b_first_time = true;
795 return VLC_SUCCESS;
796 }
797 }
798 break;
799
800 case DEMUX_GET_POSITION:
801 pf = va_arg( args, double * );
802 if( p_sys->subtitles.i_current >= p_sys->subtitles.i_count )
803 {
804 *pf = 1.0;
805 }
806 else if( p_sys->subtitles.i_count > 0 && p_sys->i_length )
807 {
808 *pf = p_sys->i_next_demux_date - var_GetInteger( p_demux->obj.parent, "spu-delay" );
809 if( *pf < 0 )
810 *pf = p_sys->i_next_demux_date;
811 *pf /= p_sys->i_length;
812 }
813 else
814 {
815 *pf = 0.0;
816 }
817 return VLC_SUCCESS;
818
819 case DEMUX_SET_POSITION:
820 f = va_arg( args, double );
821 if( p_sys->subtitles.i_count && p_sys->i_length )
822 {
823 i64 = VLC_TS_0 + f * p_sys->i_length;
824 return demux_Control( p_demux, DEMUX_SET_TIME, i64 );
825 }
826 break;
827
828 case DEMUX_SET_NEXT_DEMUX_TIME:
829 p_sys->b_slave = true;
830 p_sys->i_next_demux_date = va_arg( args, int64_t ) - VLC_TS_0;
831 return VLC_SUCCESS;
832
833 case DEMUX_GET_PTS_DELAY:
834 case DEMUX_GET_FPS:
835 case DEMUX_GET_META:
836 case DEMUX_GET_ATTACHMENTS:
837 case DEMUX_GET_TITLE_INFO:
838 case DEMUX_HAS_UNSUPPORTED_META:
839 case DEMUX_CAN_RECORD:
840 default:
841 break;
842
843 }
844 return VLC_EGENERIC;
845 }
846
847 /*****************************************************************************
848 * Demux: Send subtitle to decoder
849 *****************************************************************************/
850 static int Demux( demux_t *p_demux )
851 {
852 demux_sys_t *p_sys = p_demux->p_sys;
853
854 int64_t i_barrier = p_sys->i_next_demux_date - var_GetInteger( p_demux->obj.parent, "spu-delay" );
855 if( i_barrier < 0 )
856 i_barrier = p_sys->i_next_demux_date;
857
858 while( p_sys->subtitles.i_current < p_sys->subtitles.i_count &&
859 p_sys->subtitles.p_array[p_sys->subtitles.i_current].i_start <= i_barrier )
860 {
861 const subtitle_t *p_subtitle = &p_sys->subtitles.p_array[p_sys->subtitles.i_current];
862
863 if ( !p_sys->b_slave && p_sys->b_first_time )
864 {
865 es_out_SetPCR( p_demux->out, VLC_TS_0 + i_barrier );
866 p_sys->b_first_time = false;
867 }
868
869 if( p_subtitle->i_start >= 0 )
870 {
871 block_t *p_block = p_sys->pf_convert( p_subtitle );
872 if( p_block )
873 {
874 p_block->i_dts =
875 p_block->i_pts = VLC_TS_0 + p_subtitle->i_start;
876 if( p_subtitle->i_stop >= 0 && p_subtitle->i_stop >= p_subtitle->i_start )
877 p_block->i_length = p_subtitle->i_stop - p_subtitle->i_start;
878
879 es_out_Send( p_demux->out, p_sys->es, p_block );
880 }
881 }
882
883 p_sys->subtitles.i_current++;
884 }
885
886 if ( !p_sys->b_slave )
887 {
888 es_out_SetPCR( p_demux->out, VLC_TS_0 + i_barrier );
889 p_sys->i_next_demux_date += CLOCK_FREQ / 8;
890 }
891
892 if( p_sys->subtitles.i_current >= p_sys->subtitles.i_count )
893 return VLC_DEMUXER_EOF;
894
895 return VLC_DEMUXER_SUCCESS;
896 }
897
898
899 static int subtitle_cmp( const void *first, const void *second )
900 {
901 int64_t result = ((subtitle_t *)(first))->i_start - ((subtitle_t *)(second))->i_start;
902 /* Return -1, 0 ,1, and not directly subtraction
903 * as result can be > INT_MAX */
904 return result == 0 ? 0 : result > 0 ? 1 : -1;
905 }
906 /*****************************************************************************
907 * Fix: fix time stamp and order of subtitle
908 *****************************************************************************/
909 static void Fix( demux_t *p_demux )
910 {
911 demux_sys_t *p_sys = p_demux->p_sys;
912
913 /* *** fix order (to be sure...) *** */
914 qsort( p_sys->subtitles.p_array, p_sys->subtitles.i_count, sizeof( p_sys->subtitles.p_array[0] ), subtitle_cmp);
915 }
916
917 static int TextLoad( text_t *txt, stream_t *s )
918 {
919 size_t i_line_max;
920
921 /* init txt */
922 i_line_max = 500;
923 txt->i_line_count = 0;
924 txt->i_line = 0;
925 txt->line = calloc( i_line_max, sizeof( char * ) );
926 if( !txt->line )
927 return VLC_ENOMEM;
928
929 /* load the complete file */
930 for( ;; )
931 {
932 char *psz = vlc_stream_ReadLine( s );
933
934 if( psz == NULL )
935 break;
936
937 txt->line[txt->i_line_count] = psz;
938 if( txt->i_line_count + 1 >= i_line_max )
939 {
940 i_line_max += 100;
941 char **p_realloc = realloc( txt->line, i_line_max * sizeof( char * ) );
942 if( p_realloc == NULL )
943 return VLC_ENOMEM;
944 txt->line = p_realloc;
945 }
946 txt->i_line_count++;
947 }
948
949 if( txt->i_line_count == 0 )
950 {
951 free( txt->line );
952 return VLC_EGENERIC;
953 }
954
955 return VLC_SUCCESS;
956 }
957 static void TextUnload( text_t *txt )
958 {
959 if( txt->i_line_count )
960 {
961 for( size_t i = 0; i < txt->i_line_count; i++ )
962 free( txt->line[i] );
963 free( txt->line );
964 }
965 txt->i_line = 0;
966 txt->i_line_count = 0;
967 }
968
969 static char *TextGetLine( text_t *txt )
970 {
971 if( txt->i_line >= txt->i_line_count )
972 return( NULL );
973
974 return txt->line[txt->i_line++];
975 }
976 static void TextPreviousLine( text_t *txt )
977 {
978 if( txt->i_line > 0 )
979 txt->i_line--;
980 }
981
982 /*****************************************************************************
983 * Specific Subtitle function
984 *****************************************************************************/
985 /* ParseMicroDvd:
986 * Format:
987 * {n1}{n2}Line1|Line2|Line3....
988 * where n1 and n2 are the video frame number (n2 can be empty)
989 */
990 static int ParseMicroDvd( vlc_object_t *p_obj, subs_properties_t *p_props,
991 text_t *txt, subtitle_t *p_subtitle,
992 size_t i_idx )
993 {
994 VLC_UNUSED( i_idx );
995 char *psz_text;
996 int i_start;
997 int i_stop;
998 int i;
999
1000 for( ;; )
1001 {
1002 const char *s = TextGetLine( txt );
1003 if( !s )
1004 return VLC_EGENERIC;
1005
1006 psz_text = malloc( strlen(s) + 1 );
1007 if( !psz_text )
1008 return VLC_ENOMEM;
1009
1010 i_start = 0;
1011 i_stop = -1;
1012 if( sscanf( s, "{%d}{}%[^\r\n]", &i_start, psz_text ) == 2 ||
1013 sscanf( s, "{%d}{%d}%[^\r\n]", &i_start, &i_stop, psz_text ) == 3)
1014 {
1015 if( i_start != 1 || i_stop != 1 )
1016 break;
1017
1018 /* We found a possible setting of the framerate "{1}{1}23.976" */
1019 /* Check if it's usable, and if the sub-fps is not set */
1020 float f_fps = us_strtof( psz_text, NULL );
1021 if( f_fps > 0.f && var_GetFloat( p_obj, "sub-fps" ) <= 0.f )
1022 p_props->i_microsecperframe = llroundf(1000000.f / f_fps);
1023 }
1024 free( psz_text );
1025 }
1026
1027 /* replace | by \n */
1028 for( i = 0; psz_text[i] != '\0'; i++ )
1029 {
1030 if( psz_text[i] == '|' )
1031 psz_text[i] = '\n';
1032 }
1033
1034 /* */
1035 p_subtitle->i_start = i_start * p_props->i_microsecperframe;
1036 p_subtitle->i_stop = i_stop >= 0 ? (i_stop * p_props->i_microsecperframe) : -1;
1037 p_subtitle->psz_text = psz_text;
1038 return VLC_SUCCESS;
1039 }
1040
1041 /* ParseSubRipSubViewer
1042 * Format SubRip
1043 * n
1044 * h1:m1:s1,d1 --> h2:m2:s2,d2
1045 * Line1
1046 * Line2
1047 * ....
1048 * [Empty line]
1049 * Format SubViewer v1/v2
1050 * h1:m1:s1.d1,h2:m2:s2.d2
1051 * Line1[br]Line2
1052 * Line3
1053 * ...
1054 * [empty line]
1055 * We ignore line number for SubRip
1056 */
1057 static int ParseSubRipSubViewer( vlc_object_t *p_obj, subs_properties_t *p_props,
1058 text_t *txt, subtitle_t *p_subtitle,
1059 int (* pf_parse_timing)(subtitle_t *, const char *),
1060 bool b_replace_br )
1061 {
1062 VLC_UNUSED(p_obj);
1063 VLC_UNUSED(p_props);
1064 char *psz_text;
1065
1066 for( ;; )
1067 {
1068 const char *s = TextGetLine( txt );
1069
1070 if( !s )
1071 return VLC_EGENERIC;
1072
1073 if( pf_parse_timing( p_subtitle, s) == VLC_SUCCESS &&
1074 p_subtitle->i_start < p_subtitle->i_stop )
1075 {
1076 break;
1077 }
1078 }
1079
1080 /* Now read text until an empty line */
1081 psz_text = strdup("");
1082 if( !psz_text )
1083 return VLC_ENOMEM;
1084
1085 for( ;; )
1086 {
1087 const char *s = TextGetLine( txt );
1088 size_t i_len;
1089 size_t i_old;
1090
1091 i_len = s ? strlen( s ) : 0;
1092 if( i_len <= 0 )
1093 {
1094 p_subtitle->psz_text = psz_text;
1095 return VLC_SUCCESS;
1096 }
1097
1098 i_old = strlen( psz_text );
1099 psz_text = realloc_or_free( psz_text, i_old + i_len + 1 + 1 );
1100 if( !psz_text )
1101 {
1102 return VLC_ENOMEM;
1103 }
1104 strcat( psz_text, s );
1105 strcat( psz_text, "\n" );
1106
1107 /* replace [br] by \n */
1108 if( b_replace_br )
1109 {
1110 char *p;
1111
1112 while( ( p = strstr( psz_text, "[br]" ) ) )
1113 {
1114 *p++ = '\n';
1115 memmove( p, &p[3], strlen(&p[3])+1 );
1116 }
1117 }
1118 }
1119 }
1120
1121 /* subtitle_ParseSubRipTimingValue
1122 * Parses SubRip timing value.
1123 */
1124 static int subtitle_ParseSubRipTimingValue(int64_t *timing_value,
1125 const char *s)
1126 {
1127 int h1, m1, s1, d1 = 0;
1128
1129 if ( sscanf( s, "%d:%d:%d,%d",
1130 &h1, &m1, &s1, &d1 ) == 4 ||
1131 sscanf( s, "%d:%d:%d.%d",
1132 &h1, &m1, &s1, &d1 ) == 4 ||
1133 sscanf( s, "%d:%d:%d",
1134 &h1, &m1, &s1) == 3 )
1135 {
1136 (*timing_value) = ( (int64_t)h1 * 3600 * 1000 +
1137 (int64_t)m1 * 60 * 1000 +
1138 (int64_t)s1 * 1000 +
1139 (int64_t)d1 ) * 1000;
1140
1141 return VLC_SUCCESS;
1142 }
1143
1144 return VLC_EGENERIC;
1145 }
1146
1147 /* subtitle_ParseSubRipTiming
1148 * Parses SubRip timing.
1149 */
1150 static int subtitle_ParseSubRipTiming( subtitle_t *p_subtitle,
1151 const char *s )
1152 {
1153 int i_result = VLC_EGENERIC;
1154 char *psz_start, *psz_stop;
1155 psz_start = malloc( strlen(s) + 1 );
1156 psz_stop = malloc( strlen(s) + 1 );
1157
1158 if( sscanf( s, "%s --> %s", psz_start, psz_stop) == 2 &&
1159 subtitle_ParseSubRipTimingValue( &p_subtitle->i_start, psz_start ) == VLC_SUCCESS &&
1160 subtitle_ParseSubRipTimingValue( &p_subtitle->i_stop, psz_stop ) == VLC_SUCCESS )
1161 {
1162 i_result = VLC_SUCCESS;
1163 }
1164
1165 free(psz_start);
1166 free(psz_stop);
1167
1168 return i_result;
1169 }
1170 /* ParseSubRip
1171 */
1172 static int ParseSubRip( vlc_object_t *p_obj, subs_properties_t *p_props,
1173 text_t *txt, subtitle_t *p_subtitle,
1174 size_t i_idx )
1175 {
1176 VLC_UNUSED( i_idx );
1177 return ParseSubRipSubViewer( p_obj, p_props, txt, p_subtitle,
1178 &subtitle_ParseSubRipTiming,
1179 false );
1180 }
1181
1182 /* subtitle_ParseSubViewerTiming
1183 * Parses SubViewer timing.
1184 */
1185 static int subtitle_ParseSubViewerTiming( subtitle_t *p_subtitle,
1186 const char *s )
1187 {
1188 int h1, m1, s1, d1, h2, m2, s2, d2;
1189
1190 if( sscanf( s, "%d:%d:%d.%d,%d:%d:%d.%d",
1191 &h1, &m1, &s1, &d1, &h2, &m2, &s2, &d2) == 8 )
1192 {
1193 p_subtitle->i_start = ( (int64_t)h1 * 3600*1000 +
1194 (int64_t)m1 * 60*1000 +
1195 (int64_t)s1 * 1000 +
1196 (int64_t)d1 ) * 1000;
1197
1198 p_subtitle->i_stop = ( (int64_t)h2 * 3600*1000 +
1199 (int64_t)m2 * 60*1000 +
1200 (int64_t)s2 * 1000 +
1201 (int64_t)d2 ) * 1000;
1202 return VLC_SUCCESS;
1203 }
1204 return VLC_EGENERIC;
1205 }
1206
1207 /* ParseSubViewer
1208 */
1209 static int ParseSubViewer( vlc_object_t *p_obj, subs_properties_t *p_props,
1210 text_t *txt, subtitle_t *p_subtitle,
1211 size_t i_idx )
1212 {
1213 VLC_UNUSED( i_idx );
1214
1215 return ParseSubRipSubViewer( p_obj, p_props, txt, p_subtitle,
1216 &subtitle_ParseSubViewerTiming,
1217 true );
1218 }
1219
1220 /* ParseSSA
1221 */
1222 static int ParseSSA( vlc_object_t *p_obj, subs_properties_t *p_props,
1223 text_t *txt, subtitle_t *p_subtitle,
1224 size_t i_idx )
1225 {
1226 VLC_UNUSED(p_obj);
1227 size_t header_len = 0;
1228
1229 for( ;; )
1230 {
1231 const char *s = TextGetLine( txt );
1232 int h1, m1, s1, c1, h2, m2, s2, c2;
1233 char *psz_text, *psz_temp;
1234 char temp[16];
1235
1236 if( !s )
1237 return VLC_EGENERIC;
1238
1239 /* We expect (SSA2-4):
1240 * Format: Marked, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
1241 * Dialogue: Marked=0,0:02:40.65,0:02:41.79,Wolf main,Cher,0000,0000,0000,,Et les enregistrements de ses ondes delta ?
1242 *
1243 * SSA-1 is similar but only has 8 commas up untill the subtitle text. Probably the Effect field is no present, but not 100 % sure.
1244 */
1245
1246 /* For ASS:
1247 * Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
1248 * Dialogue: Layer#,0:02:40.65,0:02:41.79,Wolf main,Cher,0000,0000,0000,,Et les enregistrements de ses ondes delta ?
1249 */
1250
1251 /* The output text is - at least, not removing numbers - 18 chars shorter than the input text. */
1252 psz_text = malloc( strlen(s) );
1253 if( !psz_text )
1254 return VLC_ENOMEM;
1255
1256 if( sscanf( s,
1257 "Dialogue: %15[^,],%d:%d:%d.%d,%d:%d:%d.%d,%[^\r\n]",
1258 temp,
1259 &h1, &m1, &s1, &c1,
1260 &h2, &m2, &s2, &c2,
1261 psz_text ) == 10 )
1262 {
1263 /* The dec expects: ReadOrder, Layer, Style, Name, MarginL, MarginR, MarginV, Effect, Text */
1264 /* (Layer comes from ASS specs ... it's empty for SSA.) */
1265 if( p_props->i_type == SUB_TYPE_SSA1 )
1266 {
1267 /* SSA1 has only 8 commas before the text starts, not 9 */
1268 memmove( &psz_text[1], psz_text, strlen(psz_text)+1 );
1269 psz_text[0] = ',';
1270 }
1271 else
1272 {
1273 int i_layer = ( p_props->i_type == SUB_TYPE_ASS ) ? atoi( temp ) : 0;
1274
1275 /* ReadOrder, Layer, %s(rest of fields) */
1276 if( asprintf( &psz_temp, "%zu,%d,%s", i_idx, i_layer, psz_text ) == -1 )
1277 {
1278 free( psz_text );
1279 return VLC_ENOMEM;
1280 }
1281
1282 free( psz_text );
1283 psz_text = psz_temp;
1284 }
1285
1286 p_subtitle->i_start = ( (int64_t)h1 * 3600*1000 +
1287 (int64_t)m1 * 60*1000 +
1288 (int64_t)s1 * 1000 +
1289 (int64_t)c1 * 10 ) * 1000;
1290 p_subtitle->i_stop = ( (int64_t)h2 * 3600*1000 +
1291 (int64_t)m2 * 60*1000 +
1292 (int64_t)s2 * 1000 +
1293 (int64_t)c2 * 10 ) * 1000;
1294 p_subtitle->psz_text = psz_text;
1295 return VLC_SUCCESS;
1296 }
1297 free( psz_text );
1298
1299 /* All the other stuff we add to the header field */
1300 if( header_len == 0 && p_props->psz_header )
1301 header_len = strlen( p_props->psz_header );
1302
1303 size_t s_len = strlen( s );
1304 p_props->psz_header = realloc_or_free( p_props->psz_header, header_len + s_len + 2 );
1305 if( !p_props->psz_header )
1306 return VLC_ENOMEM;
1307 snprintf( p_props->psz_header + header_len, s_len + 2, "%s\n", s );
1308 header_len += s_len + 1;
1309 }
1310 }
1311
1312 /* ParseVplayer
1313 * Format
1314 * h:m:s:Line1|Line2|Line3....
1315 * or
1316 * h:m:s Line1|Line2|Line3....
1317 */
1318 static int ParseVplayer( vlc_object_t *p_obj, subs_properties_t *p_props,
1319 text_t *txt, subtitle_t *p_subtitle,
1320 size_t i_idx )
1321 {
1322 VLC_UNUSED(p_obj);
1323 VLC_UNUSED(p_props);
1324 VLC_UNUSED( i_idx );
1325 char *psz_text;
1326
1327 for( ;; )
1328 {
1329 const char *s = TextGetLine( txt );
1330 int h1, m1, s1;
1331
1332 if( !s )
1333 return VLC_EGENERIC;
1334
1335 psz_text = malloc( strlen( s ) + 1 );
1336 if( !psz_text )
1337 return VLC_ENOMEM;
1338
1339 if( sscanf( s, "%d:%d:%d%*c%[^\r\n]",
1340 &h1, &m1, &s1, psz_text ) == 4 )
1341 {
1342 p_subtitle->i_start = ( (int64_t)h1 * 3600*1000 +
1343 (int64_t)m1 * 60*1000 +
1344 (int64_t)s1 * 1000 ) * 1000;
1345 p_subtitle->i_stop = -1;
1346 break;
1347 }
1348 free( psz_text );
1349 }
1350
1351 /* replace | by \n */
1352 for( size_t i = 0; psz_text[i] != '\0'; i++ )
1353 {
1354 if( psz_text[i] == '|' )
1355 psz_text[i] = '\n';
1356 }
1357 p_subtitle->psz_text = psz_text;
1358 return VLC_SUCCESS;
1359 }
1360
1361 /* ParseSami
1362 */
1363 static const char *ParseSamiSearch( text_t *txt,
1364 const char *psz_start, const char *psz_str )
1365 {
1366 if( psz_start && strcasestr( psz_start, psz_str ) )
1367 {
1368 const char *s = strcasestr( psz_start, psz_str );
1369 return &s[strlen( psz_str )];
1370 }
1371
1372 for( ;; )
1373 {
1374 const char *p = TextGetLine( txt );
1375 if( !p )
1376 return NULL;
1377
1378 const char *s = strcasestr( p, psz_str );
1379 if( s != NULL )
1380 return &s[strlen( psz_str )];
1381 }
1382 }
1383 static int ParseSami( vlc_object_t *p_obj, subs_properties_t *p_props,
1384 text_t *txt, subtitle_t *p_subtitle, size_t i_idx )
1385 {
1386 VLC_UNUSED(p_obj);
1387 VLC_UNUSED(p_props);
1388 VLC_UNUSED( i_idx );
1389 const char *s;
1390 int64_t i_start;
1391
1392 unsigned int i_text;
1393 char text[8192]; /* Arbitrary but should be long enough */
1394
1395 /* search "Start=" */
1396 s = ParseSamiSearch( txt, p_props->sami.psz_start, "Start=" );
1397 p_props->sami.psz_start = NULL;
1398 if( !s )
1399 return VLC_EGENERIC;
1400
1401 /* get start value */
1402 char *psz_end;
1403 i_start = strtol( s, &psz_end, 0 );
1404 s = psz_end;
1405
1406 /* search <P */
1407 if( !( s = ParseSamiSearch( txt, s, "<P" ) ) )
1408 return VLC_EGENERIC;
1409
1410 /* search > */
1411 if( !( s = ParseSamiSearch( txt, s, ">" ) ) )
1412 return VLC_EGENERIC;
1413
1414 i_text = 0;
1415 text[0] = '\0';
1416 /* now get all txt until a "Start=" line */
1417 for( ;; )
1418 {
1419 char c = '\0';
1420 /* Search non empty line */
1421 while( s && *s == '\0' )
1422 s = TextGetLine( txt );
1423 if( !s )
1424 break;
1425
1426 if( *s == '<' )
1427 {
1428 if( !strncasecmp( s, "<br", 3 ) )
1429 {
1430 c = '\n';
1431 }
1432 else if( strcasestr( s, "Start=" ) )
1433 {
1434 p_props->sami.psz_start = s;
1435 break;
1436 }
1437 s = ParseSamiSearch( txt, s, ">" );
1438 }
1439 else if( !strncmp( s, "&nbsp;", 6 ) )
1440 {
1441 c = ' ';
1442 s += 6;
1443 }
1444 else if( *s == '\t' )
1445 {
1446 c = ' ';
1447 s++;
1448 }
1449 else
1450 {
1451 c = *s;
1452 s++;
1453 }
1454 if( c != '\0' && i_text+1 < sizeof(text) )
1455 {
1456 text[i_text++] = c;
1457 text[i_text] = '\0';
1458 }
1459 }
1460
1461 p_subtitle->i_start = i_start * 1000;
1462 p_subtitle->i_stop = -1;
1463 p_subtitle->psz_text = strdup( text );
1464
1465 return VLC_SUCCESS;
1466 }
1467
1468 /* ParseDVDSubtitle
1469 * Format
1470 * {T h1:m1:s1:c1
1471 * Line1
1472 * Line2
1473 * ...
1474 * }
1475 * TODO it can have a header
1476 * { HEAD
1477 * ...
1478 * CODEPAGE=...
1479 * FORMAT=...
1480 * LANG=English
1481 * }
1482 * LANG support would be cool
1483 * CODEPAGE is probably mandatory FIXME
1484 */
1485 static int ParseDVDSubtitle(vlc_object_t *p_obj, subs_properties_t *p_props,
1486 text_t *txt, subtitle_t *p_subtitle, size_t i_idx )
1487 {
1488 VLC_UNUSED(p_obj);
1489 VLC_UNUSED(p_props);
1490 VLC_UNUSED( i_idx );
1491 char *psz_text;
1492
1493 for( ;; )
1494 {
1495 const char *s = TextGetLine( txt );
1496 int h1, m1, s1, c1;
1497
1498 if( !s )
1499 return VLC_EGENERIC;
1500
1501 if( sscanf( s,
1502 "{T %d:%d:%d:%d",
1503 &h1, &m1, &s1, &c1 ) == 4 )
1504 {
1505 p_subtitle->i_start = ( (int64_t)h1 * 3600*1000 +
1506 (int64_t)m1 * 60*1000 +
1507 (int64_t)s1 * 1000 +
1508 (int64_t)c1 * 10) * 1000;
1509 p_subtitle->i_stop = -1;
1510 break;
1511 }
1512 }
1513
1514 /* Now read text until a line containing "}" */
1515 psz_text = strdup("");
1516 if( !psz_text )
1517 return VLC_ENOMEM;
1518 for( ;; )
1519 {
1520 const char *s = TextGetLine( txt );
1521 int i_len;
1522 int i_old;
1523
1524 if( !s )
1525 {
1526 free( psz_text );
1527 return VLC_EGENERIC;
1528 }
1529
1530 i_len = strlen( s );
1531 if( i_len == 1 && s[0] == '}')
1532 {
1533 p_subtitle->psz_text = psz_text;
1534 return VLC_SUCCESS;
1535 }
1536
1537 i_old = strlen( psz_text );
1538 psz_text = realloc_or_free( psz_text, i_old + i_len + 1 + 1 );
1539 if( !psz_text )
1540 return VLC_ENOMEM;
1541 strcat( psz_text, s );
1542 strcat( psz_text, "\n" );
1543 }
1544 }
1545
1546 /* ParseMPL2
1547 * Format
1548 * [n1][n2]Line1|Line2|Line3...
1549 * where n1 and n2 are the video frame number (n2 can be empty)
1550 */
1551 static int ParseMPL2(vlc_object_t *p_obj, subs_properties_t *p_props,
1552 text_t *txt, subtitle_t *p_subtitle, size_t i_idx )
1553 {
1554 VLC_UNUSED(p_obj);
1555 VLC_UNUSED(p_props);
1556 VLC_UNUSED( i_idx );
1557 char *psz_text;
1558 int i;
1559
1560 for( ;; )
1561 {
1562 const char *s = TextGetLine( txt );
1563 int i_start;
1564 int i_stop;
1565
1566 if( !s )
1567 return VLC_EGENERIC;
1568
1569 psz_text = malloc( strlen(s) + 1 );
1570 if( !psz_text )
1571 return VLC_ENOMEM;
1572
1573 i_start = 0;
1574 i_stop = -1;
1575 if( sscanf( s, "[%d][] %[^\r\n]", &i_start, psz_text ) == 2 ||
1576 sscanf( s, "[%d][%d] %[^\r\n]", &i_start, &i_stop, psz_text ) == 3)
1577 {
1578 p_subtitle->i_start = (int64_t)i_start * 100000;
1579 p_subtitle->i_stop = i_stop >= 0 ? ((int64_t)i_stop * 100000) : -1;
1580 break;
1581 }
1582 free( psz_text );
1583 }
1584
1585 for( i = 0; psz_text[i] != '\0'; )
1586 {
1587 /* replace | by \n */
1588 if( psz_text[i] == '|' )
1589 psz_text[i] = '\n';
1590
1591 /* Remove italic */
1592 if( psz_text[i] == '/' && ( i == 0 || psz_text[i-1] == '\n' ) )
1593 memmove( &psz_text[i], &psz_text[i+1], strlen(&psz_text[i+1])+1 );
1594 else
1595 i++;
1596 }
1597 p_subtitle->psz_text = psz_text;
1598 return VLC_SUCCESS;
1599 }
1600
1601 static int ParseAQT(vlc_object_t *p_obj, subs_properties_t *p_props, text_t *txt, subtitle_t *p_subtitle, size_t i_idx )
1602 {
1603 VLC_UNUSED(p_obj);
1604 VLC_UNUSED(p_props);
1605 VLC_UNUSED( i_idx );
1606
1607 char *psz_text = strdup( "" );
1608 int i_old = 0;
1609 int i_firstline = 1;
1610
1611 for( ;; )
1612 {
1613 int t; /* Time */
1614
1615 const char *s = TextGetLine( txt );
1616
1617 if( !s )
1618 {
1619 free( psz_text );
1620 return VLC_EGENERIC;
1621 }
1622
1623 /* Data Lines */
1624 if( sscanf (s, "-->> %d", &t) == 1)
1625 {
1626 p_subtitle->i_start = (int64_t)t; /* * FPS*/
1627 p_subtitle->i_stop = -1;
1628
1629 /* Starting of a subtitle */
1630 if( i_firstline )
1631 {
1632 i_firstline = 0;
1633 }
1634 /* We have been too far: end of the subtitle, begin of next */
1635 else
1636 {
1637 TextPreviousLine( txt );
1638 break;
1639 }
1640 }
1641 /* Text Lines */
1642 else
1643 {
1644 i_old = strlen( psz_text ) + 1;
1645 psz_text = realloc_or_free( psz_text, i_old + strlen( s ) + 1 );
1646 if( !psz_text )
1647 return VLC_ENOMEM;
1648 strcat( psz_text, s );
1649 strcat( psz_text, "\n" );
1650 if( txt->i_line == txt->i_line_count )
1651 break;
1652 }
1653 }
1654 p_subtitle->psz_text = psz_text;
1655 return VLC_SUCCESS;
1656 }
1657
1658 static int ParsePJS(vlc_object_t *p_obj, subs_properties_t *p_props,
1659 text_t *txt, subtitle_t *p_subtitle, size_t i_idx )
1660 {
1661 VLC_UNUSED(p_obj);
1662 VLC_UNUSED(p_props);
1663 VLC_UNUSED( i_idx );
1664
1665 char *psz_text;
1666 int i;
1667
1668 for( ;; )
1669 {
1670 const char *s = TextGetLine( txt );
1671 int t1, t2;
1672
1673 if( !s )
1674 return VLC_EGENERIC;
1675
1676 psz_text = malloc( strlen(s) + 1 );
1677 if( !psz_text )
1678 return VLC_ENOMEM;
1679
1680 /* Data Lines */
1681 if( sscanf (s, "%d,%d,\"%[^\n\r]", &t1, &t2, psz_text ) == 3 )
1682 {
1683 /* 1/10th of second ? Frame based ? FIXME */
1684 p_subtitle->i_start = 10 * t1;
1685 p_subtitle->i_stop = 10 * t2;
1686 /* Remove latest " */
1687 psz_text[ strlen(psz_text) - 1 ] = '\0';
1688
1689 break;
1690 }
1691 free( psz_text );
1692 }
1693
1694 /* replace | by \n */
1695 for( i = 0; psz_text[i] != '\0'; i++ )
1696 {
1697 if( psz_text[i] == '|' )
1698 psz_text[i] = '\n';
1699 }
1700
1701 p_subtitle->psz_text = psz_text;
1702 msg_Dbg( p_obj, "%s", psz_text );
1703 return VLC_SUCCESS;
1704 }
1705
1706 static int ParseMPSub( vlc_object_t *p_obj, subs_properties_t *p_props,
1707 text_t *txt, subtitle_t *p_subtitle, size_t i_idx )
1708 {
1709 VLC_UNUSED( i_idx );
1710
1711 char *psz_text = strdup( "" );
1712
1713 if( !p_props->mpsub.b_inited )
1714 {
1715 p_props->mpsub.f_total = 0.0;
1716 p_props->mpsub.f_factor = 0.0;
1717
1718 p_props->mpsub.b_inited = true;
1719 }
1720
1721 for( ;; )
1722 {
1723 char p_dummy;
1724 char *psz_temp;
1725
1726 const char *s = TextGetLine( txt );
1727 if( !s )
1728 {
1729 free( psz_text );
1730 return VLC_EGENERIC;
1731 }
1732
1733 if( strstr( s, "FORMAT" ) )
1734 {
1735 if( sscanf (s, "FORMAT=TIM%c", &p_dummy ) == 1 && p_dummy == 'E')
1736 {
1737 p_props->mpsub.f_factor = 100.0;
1738 break;
1739 }
1740
1741 psz_temp = malloc( strlen(s) );
1742 if( !psz_temp )
1743 {
1744 free( psz_text );
1745 return VLC_ENOMEM;
1746 }
1747
1748 if( sscanf( s, "FORMAT=%[^\r\n]", psz_temp ) )
1749 {
1750 float f_fps = us_strtof( psz_temp, NULL );
1751
1752 if( f_fps > 0.f && var_GetFloat( p_obj, "sub-fps" ) <= 0.f )
1753 var_SetFloat( p_obj, "sub-fps", f_fps );
1754
1755 p_props->mpsub.f_factor = 1.f;
1756 free( psz_temp );
1757 break;
1758 }
1759 free( psz_temp );
1760 }
1761
1762 /* Data Lines */
1763 float f1 = us_strtof( s, &psz_temp );
1764 if( *psz_temp )
1765 {
1766 float f2 = us_strtof( psz_temp, NULL );
1767 p_props->mpsub.f_total += f1 * p_props->mpsub.f_factor;
1768 p_subtitle->i_start = llroundf(10000.f * p_props->mpsub.f_total);
1769 p_props->mpsub.f_total += f2 * p_props->mpsub.f_factor;
1770 p_subtitle->i_stop = llroundf(10000.f * p_props->mpsub.f_total);
1771 break;
1772 }
1773 }
1774
1775 for( ;; )
1776 {
1777 const char *s = TextGetLine( txt );
1778
1779 if( !s )
1780 {
1781 free( psz_text );
1782 return VLC_EGENERIC;
1783 }
1784
1785 size_t i_len = strlen( s );
1786 if( i_len == 0 )
1787 break;
1788
1789 size_t i_old = strlen( psz_text );
1790
1791 psz_text = realloc_or_free( psz_text, i_old + i_len + 1 + 1 );
1792 if( !psz_text )
1793 return VLC_ENOMEM;
1794
1795 strcat( psz_text, s );
1796 strcat( psz_text, "\n" );
1797 }
1798
1799 p_subtitle->psz_text = psz_text;
1800 return VLC_SUCCESS;
1801 }
1802
1803 static int ParseJSS( vlc_object_t *p_obj, subs_properties_t *p_props,
1804 text_t *txt, subtitle_t *p_subtitle, size_t i_idx )
1805 {
1806 VLC_UNUSED( i_idx );
1807 char *psz_text, *psz_orig;
1808 char *psz_text2, *psz_orig2;
1809
1810 if( !p_props->jss.b_inited )
1811 {
1812 p_props->jss.i_comment = 0;
1813 p_props->jss.i_time_resolution = 30;
1814 p_props->jss.i_time_shift = 0;
1815
1816 p_props->jss.b_inited = true;
1817 }
1818
1819 /* Parse the main lines */
1820 for( ;; )
1821 {
1822 const char *s = TextGetLine( txt );
1823 if( !s )
1824 return VLC_EGENERIC;
1825
1826 size_t line_length = strlen( s );
1827 psz_orig = malloc( line_length + 1 );
1828 if( !psz_orig )
1829 return VLC_ENOMEM;
1830 psz_text = psz_orig;
1831
1832 /* Complete time lines */
1833 int h1, h2, m1, m2, s1, s2, f1, f2;
1834 if( sscanf( s, "%d:%d:%d.%d %d:%d:%d.%d %[^\n\r]",
1835 &h1, &m1, &s1, &f1, &h2, &m2, &s2, &f2, psz_text ) == 9 )
1836 {
1837 p_subtitle->i_start = ( ( (int64_t) h1 *3600 + m1 * 60 + s1 ) +
1838 (int64_t)( ( f1 + p_props->jss.i_time_shift ) / p_props->jss.i_time_resolution ) )
1839 * 1000000;
1840 p_subtitle->i_stop = ( ( (int64_t) h2 *3600 + m2 * 60 + s2 ) +
1841 (int64_t)( ( f2 + p_props->jss.i_time_shift ) / p_props->jss.i_time_resolution ) )
1842 * 1000000;
1843 break;
1844 }
1845 /* Short time lines */
1846 else if( sscanf( s, "@%d @%d %[^\n\r]", &f1, &f2, psz_text ) == 3 )
1847 {
1848 p_subtitle->i_start = ((int64_t)
1849 ( f1 + p_props->jss.i_time_shift ) / p_props->jss.i_time_resolution * 1000000.0 );
1850 p_subtitle->i_stop = ((int64_t)
1851 ( f2 + p_props->jss.i_time_shift ) / p_props->jss.i_time_resolution * 1000000.0 );
1852 break;
1853 }
1854 /* General Directive lines */
1855 /* Only TIME and SHIFT are supported so far */
1856 else if( s[0] == '#' )
1857 {
1858 int h = 0, m =0, sec = 1, f = 1;
1859 unsigned shift = 1;
1860 int inv = 1;
1861
1862 strcpy( psz_text, s );
1863
1864 switch( toupper( (unsigned char)psz_text[1] ) )
1865 {
1866 case 'S':
1867 shift = isalpha( (unsigned char)psz_text[2] ) ? 6 : 2 ;
1868 if ( shift > line_length )
1869 break;
1870
1871 if( sscanf( &psz_text[shift], "%d", &h ) )
1872 {
1873 /* Negative shifting */
1874 if( h < 0 )
1875 {
1876 h *= -1;
1877 inv = -1;
1878 }
1879
1880 if( sscanf( &psz_text[shift], "%*d:%d", &m ) )
1881 {
1882 if( sscanf( &psz_text[shift], "%*d:%*d:%d", &sec ) )
1883 {
1884 sscanf( &psz_text[shift], "%*d:%*d:%*d.%d", &f );
1885 }
1886 else
1887 {
1888 h = 0;
1889 sscanf( &psz_text[shift], "%d:%d.%d",
1890 &m, &sec, &f );
1891 m *= inv;
1892 }
1893 }
1894 else
1895 {
1896 h = m = 0;
1897 sscanf( &psz_text[shift], "%d.%d", &sec, &f);
1898 sec *= inv;
1899 }
1900 p_props->jss.i_time_shift = ( ( h * 3600 + m * 60 + sec )
1901 * p_props->jss.i_time_resolution + f ) * inv;
1902 }
1903 break;
1904
1905 case 'T':
1906 shift = isalpha( (unsigned char)psz_text[2] ) ? 8 : 2 ;
1907 if ( shift > line_length )
1908 break;
1909
1910 sscanf( &psz_text[shift], "%d", &p_props->jss.i_time_resolution );
1911 break;
1912 }
1913 free( psz_orig );
1914 continue;
1915 }
1916 else
1917 /* Unkown type line, probably a comment */
1918 {
1919 free( psz_orig );
1920 continue;
1921 }
1922 }
1923
1924 while( psz_text[ strlen( psz_text ) - 1 ] == '\\' )
1925 {
1926 const char *s2 = TextGetLine( txt );
1927
1928 if( !s2 )
1929 {
1930 free( psz_orig );
1931 return VLC_EGENERIC;
1932 }
1933
1934 size_t i_len = strlen( s2 );
1935 if( i_len == 0 )
1936 break;
1937
1938 size_t i_old = strlen( psz_text );
1939
1940 psz_text = realloc_or_free( psz_text, i_old + i_len + 1 );
1941 if( !psz_text )
1942 return VLC_ENOMEM;
1943
1944 psz_orig = psz_text;
1945 strcat( psz_text, s2 );
1946 }
1947
1948 /* Skip the blanks */
1949 while( *psz_text == ' ' || *psz_text == '\t' ) psz_text++;
1950
1951 /* Parse the directives */
1952 if( isalpha( (unsigned char)*psz_text ) || *psz_text == '[' )
1953 {
1954 while( *psz_text && *psz_text != ' ' )
1955 ++psz_text;
1956
1957 /* Directives are NOT parsed yet */
1958 /* This has probably a better place in a decoder ? */
1959 /* directive = malloc( strlen( psz_text ) + 1 );
1960 if( sscanf( psz_text, "%s %[^\n\r]", directive, psz_text2 ) == 2 )*/
1961 }
1962
1963 /* Skip the blanks after directives */
1964 while( *psz_text == ' ' || *psz_text == '\t' ) psz_text++;
1965
1966 /* Clean all the lines from inline comments and other stuffs */
1967 psz_orig2 = calloc( strlen( psz_text) + 1, 1 );
1968 psz_text2 = psz_orig2;
1969
1970 for( ; *psz_text != '\0' && *psz_text != '\n' && *psz_text != '\r'; )
1971 {
1972 switch( *psz_text )
1973 {
1974 case '{':
1975 p_props->jss.i_comment++;
1976 break;
1977 case '}':
1978 if( p_props->jss.i_comment )
1979 {
1980 p_props->jss.i_comment = 0;
1981 if( (*(psz_text + 1 ) ) == ' ' ) psz_text++;
1982 }
1983 break;
1984 case '~':
1985 if( !p_props->jss.i_comment )
1986 {
1987 *psz_text2 = ' ';
1988 psz_text2++;
1989 }
1990 break;
1991 case ' ':
1992 case '\t':
1993 if( (*(psz_text + 1 ) ) == ' ' || (*(psz_text + 1 ) ) == '\t' )
1994 break;
1995 if( !p_props->jss.i_comment )
1996 {
1997 *psz_text2 = ' ';
1998 psz_text2++;
1999 }
2000 break;
2001 case '\\':
2002 if( (*(psz_text + 1 ) ) == 'n' )
2003 {
2004 *psz_text2 = '\n';
2005 psz_text++;
2006 psz_text2++;
2007 break;
2008 }
2009 if( ( toupper((unsigned char)*(psz_text + 1 ) ) == 'C' ) ||
2010 ( toupper((unsigned char)*(psz_text + 1 ) ) == 'F' ) )
2011 {
2012 psz_text++;
2013 break;
2014 }
2015 if( (*(psz_text + 1 ) ) == 'B' || (*(psz_text + 1 ) ) == 'b' ||
2016 (*(psz_text + 1 ) ) == 'I' || (*(psz_text + 1 ) ) == 'i' ||
2017 (*(psz_text + 1 ) ) == 'U' || (*(psz_text + 1 ) ) == 'u' ||
2018 (*(psz_text + 1 ) ) == 'D' || (*(psz_text + 1 ) ) == 'N' )
2019 {
2020 psz_text++;
2021 break;
2022 }
2023 if( (*(psz_text + 1 ) ) == '~' || (*(psz_text + 1 ) ) == '{' ||
2024 (*(psz_text + 1 ) ) == '\\' )
2025 psz_text++;
2026 else if( ( *(psz_text + 1 ) == '\r' || *(psz_text + 1 ) == '\n' ) &&
2027 *(psz_text + 1 ) != '\0' )
2028 {
2029 psz_text++;
2030 }
2031 break;
2032 default:
2033 if( !p_props->jss.i_comment )
2034 {
2035 *psz_text2 = *psz_text;
2036 psz_text2++;
2037 }
2038 }
2039 psz_text++;
2040 }
2041
2042 p_subtitle->psz_text = psz_orig2;
2043 msg_Dbg( p_obj, "%s", p_subtitle->psz_text );
2044 free( psz_orig );
2045 return VLC_SUCCESS;
2046 }
2047
2048 static int ParsePSB( vlc_object_t *p_obj, subs_properties_t *p_props,
2049 text_t *txt, subtitle_t *p_subtitle, size_t i_idx )
2050 {
2051 VLC_UNUSED(p_obj);
2052 VLC_UNUSED(p_props);
2053 VLC_UNUSED( i_idx );
2054
2055 char *psz_text;
2056 int i;
2057
2058 for( ;; )
2059 {
2060 int h1, m1, s1;
2061 int h2, m2, s2;
2062 const char *s = TextGetLine( txt );
2063
2064 if( !s )
2065 return VLC_EGENERIC;
2066
2067 psz_text = malloc( strlen( s ) + 1 );
2068 if( !psz_text )
2069 return VLC_ENOMEM;
2070
2071 if( sscanf( s, "{%d:%d:%d}{%d:%d:%d}%[^\r\n]",
2072 &h1, &m1, &s1, &h2, &m2, &s2, psz_text ) == 7 )
2073 {
2074 p_subtitle->i_start = ( (int64_t)h1 * 3600*1000 +
2075 (int64_t)m1 * 60*1000 +
2076 (int64_t)s1 * 1000 ) * 1000;
2077 p_subtitle->i_stop = ( (int64_t)h2 * 3600*1000 +
2078 (int64_t)m2 * 60*1000 +
2079 (int64_t)s2 * 1000 ) * 1000;
2080 break;
2081 }
2082 free( psz_text );
2083 }
2084
2085 /* replace | by \n */
2086 for( i = 0; psz_text[i] != '\0'; i++ )
2087 {
2088 if( psz_text[i] == '|' )
2089 psz_text[i] = '\n';
2090 }
2091 p_subtitle->psz_text = psz_text;
2092 return VLC_SUCCESS;
2093 }
2094
2095 static int64_t ParseRealTime( char *psz, int *h, int *m, int *s, int *f )
2096 {
2097 if( *psz == '\0' ) return 0;
2098 if( sscanf( psz, "%d:%d:%d.%d", h, m, s, f ) == 4 ||
2099 sscanf( psz, "%d:%d.%d", m, s, f ) == 3 ||
2100 sscanf( psz, "%d.%d", s, f ) == 2 ||
2101 sscanf( psz, "%d:%d", m, s ) == 2 ||
2102 sscanf( psz, "%d", s ) == 1 )
2103 {
2104 return (int64_t)((( *h * 60 + *m ) * 60 ) + *s ) * 1000 * 1000
2105 + (int64_t)*f * 10 * 1000;
2106 }
2107 else return VLC_EGENERIC;
2108 }
2109
2110 static int ParseRealText( vlc_object_t *p_obj, subs_properties_t *p_props,
2111 text_t *txt, subtitle_t *p_subtitle, size_t i_idx )
2112 {
2113 VLC_UNUSED(p_obj);
2114 VLC_UNUSED(p_props);
2115 VLC_UNUSED( i_idx );
2116 char *psz_text = NULL;
2117
2118 for( ;; )
2119 {
2120 int h1 = 0, m1 = 0, s1 = 0, f1 = 0;
2121 int h2 = 0, m2 = 0, s2 = 0, f2 = 0;
2122 const char *s = TextGetLine( txt );
2123 free( psz_text );
2124
2125 if( !s )
2126 return VLC_EGENERIC;
2127
2128 psz_text = malloc( strlen( s ) + 1 );
2129 if( !psz_text )
2130 return VLC_ENOMEM;
2131
2132 /* Find the good begining. This removes extra spaces at the beginning
2133 of the line.*/
2134 char *psz_temp = strcasestr( s, "<time");
2135 if( psz_temp != NULL )
2136 {
2137 char psz_end[12], psz_begin[12];
2138 /* Line has begin and end */
2139 if( ( sscanf( psz_temp,
2140 "<%*[t|T]ime %*[b|B]egin=\"%11[^\"]\" %*[e|E]nd=\"%11[^\"]%*[^>]%[^\n\r]",
2141 psz_begin, psz_end, psz_text) != 3 ) &&
2142 /* Line has begin and no end */
2143 ( sscanf( psz_temp,
2144 "<%*[t|T]ime %*[b|B]egin=\"%11[^\"]\"%*[^>]%[^\n\r]",
2145 psz_begin, psz_text ) != 2) )
2146 /* Line is not recognized */
2147 {
2148 continue;
2149 }
2150
2151 /* Get the times */
2152 int64_t i_time = ParseRealTime( psz_begin, &h1, &m1, &s1, &f1 );
2153 p_subtitle->i_start = i_time >= 0 ? i_time : 0;
2154
2155 i_time = ParseRealTime( psz_end, &h2, &m2, &s2, &f2 );
2156 p_subtitle->i_stop = i_time >= 0 ? i_time : -1;
2157 break;
2158 }
2159 }
2160
2161 /* Get the following Lines */
2162 for( ;; )
2163 {
2164 const char *s = TextGetLine( txt );
2165
2166 if( !s )
2167 {
2168 free( psz_text );
2169 return VLC_EGENERIC;
2170 }
2171
2172 size_t i_len = strlen( s );
2173 if( i_len == 0 ) break;
2174
2175 if( strcasestr( s, "<time" ) ||
2176 strcasestr( s, "<clear/") )
2177 {
2178 TextPreviousLine( txt );
2179 break;
2180 }
2181
2182 size_t i_old = strlen( psz_text );
2183
2184 psz_text = realloc_or_free( psz_text, i_old + i_len + 1 + 1 );
2185 if( !psz_text )
2186 return VLC_ENOMEM;
2187
2188 strcat( psz_text, s );
2189 strcat( psz_text, "\n" );
2190 }
2191
2192 /* Remove the starting ">" that remained after the sscanf */
2193 memmove( &psz_text[0], &psz_text[1], strlen( psz_text ) );
2194
2195 p_subtitle->psz_text = psz_text;
2196
2197 return VLC_SUCCESS;
2198 }
2199
2200 static int ParseDKS( vlc_object_t *p_obj, subs_properties_t *p_props,
2201 text_t *txt, subtitle_t *p_subtitle, size_t i_idx )
2202 {
2203 VLC_UNUSED(p_obj);
2204 VLC_UNUSED(p_props);
2205 VLC_UNUSED( i_idx );
2206
2207 char *psz_text;
2208
2209 for( ;; )
2210 {
2211 int h1, m1, s1;
2212 int h2, m2, s2;
2213 char *s = TextGetLine( txt );
2214
2215 if( !s )
2216 return VLC_EGENERIC;
2217
2218 psz_text = malloc( strlen( s ) + 1 );
2219 if( !psz_text )
2220 return VLC_ENOMEM;
2221
2222 if( sscanf( s, "[%d:%d:%d]%[^\r\n]",
2223 &h1, &m1, &s1, psz_text ) == 4 )
2224 {
2225 p_subtitle->i_start = ( (int64_t)h1 * 3600*1000 +
2226 (int64_t)m1 * 60*1000 +
2227 (int64_t)s1 * 1000 ) * 1000;
2228
2229 s = TextGetLine( txt );
2230 if( !s )
2231 {
2232 free( psz_text );
2233 return VLC_EGENERIC;
2234 }
2235
2236 if( sscanf( s, "[%d:%d:%d]", &h2, &m2, &s2 ) == 3 )
2237 p_subtitle->i_stop = ( (int64_t)h2 * 3600*1000 +
2238 (int64_t)m2 * 60*1000 +
2239 (int64_t)s2 * 1000 ) * 1000;
2240 else
2241 p_subtitle->i_stop = -1;
2242 break;
2243 }
2244 free( psz_text );
2245 }
2246
2247 /* replace [br] by \n */
2248 char *p;
2249 while( ( p = strstr( psz_text, "[br]" ) ) )
2250 {
2251 *p++ = '\n';
2252 memmove( p, &p[3], strlen(&p[3])+1 );
2253 }
2254
2255 p_subtitle->psz_text = psz_text;
2256 return VLC_SUCCESS;
2257 }
2258
2259 static int ParseSubViewer1( vlc_object_t *p_obj, subs_properties_t *p_props,
2260 text_t *txt, subtitle_t *p_subtitle, size_t i_idx )
2261 {
2262 VLC_UNUSED(p_obj);
2263 VLC_UNUSED(p_props);
2264 VLC_UNUSED( i_idx );
2265 char *psz_text;
2266
2267 for( ;; )
2268 {
2269 int h1, m1, s1;
2270 int h2, m2, s2;
2271 char *s = TextGetLine( txt );
2272
2273 if( !s )
2274 return VLC_EGENERIC;
2275
2276 if( sscanf( s, "[%d:%d:%d]", &h1, &m1, &s1 ) == 3 )
2277 {
2278 p_subtitle->i_start = ( (int64_t)h1 * 3600*1000 +
2279 (int64_t)m1 * 60*1000 +
2280 (int64_t)s1 * 1000 ) * 1000;
2281
2282 s = TextGetLine( txt );
2283 if( !s )
2284 return VLC_EGENERIC;
2285
2286 psz_text = strdup( s );
2287 if( !psz_text )
2288 return VLC_ENOMEM;
2289
2290 s = TextGetLine( txt );
2291 if( !s )
2292 {
2293 free( psz_text );
2294 return VLC_EGENERIC;
2295 }
2296
2297 if( sscanf( s, "[%d:%d:%d]", &h2, &m2, &s2 ) == 3 )
2298 p_subtitle->i_stop = ( (int64_t)h2 * 3600*1000 +
2299 (int64_t)m2 * 60*1000 +
2300 (int64_t)s2 * 1000 ) * 1000;
2301 else
2302 p_subtitle->i_stop = -1;
2303
2304 break;
2305 }
2306 }
2307
2308 p_subtitle->psz_text = psz_text;
2309
2310 return VLC_SUCCESS;
2311 }
2312
2313 static int ParseCommonSBV( vlc_object_t *p_obj, subs_properties_t *p_props,
2314 text_t *txt, subtitle_t *p_subtitle, size_t i_idx )
2315 {
2316 VLC_UNUSED(p_obj);
2317 VLC_UNUSED( i_idx );
2318 VLC_UNUSED( p_props );
2319 char *psz_text;
2320
2321 for( ;; )
2322 {
2323 const char *s = TextGetLine( txt );
2324 int h1 = 0, m1 = 0, s1 = 0, d1 = 0;
2325 int h2 = 0, m2 = 0, s2 = 0, d2 = 0;
2326
2327 if( !s )
2328 return VLC_EGENERIC;
2329
2330 if( sscanf( s,"%d:%d:%d.%d,%d:%d:%d.%d",
2331 &h1, &m1, &s1, &d1,
2332 &h2, &m2, &s2, &d2 ) == 8 )
2333 {
2334 p_subtitle->i_start = ( (int64_t)h1 * 3600 * 1000 +
2335 (int64_t)m1 * 60 * 1000 +
2336 (int64_t)s1 * 1000 +
2337 (int64_t)d1 ) * 1000;
2338
2339 p_subtitle->i_stop = ( (int64_t)h2 * 3600 * 1000 +
2340 (int64_t)m2 * 60 * 1000 +
2341 (int64_t)s2 * 1000 +
2342 (int64_t)d2 ) * 1000;
2343 if( p_subtitle->i_start < p_subtitle->i_stop )
2344 break;
2345 }
2346 }
2347
2348 /* Now read text until an empty line */
2349 psz_text = strdup("");
2350 if( !psz_text )
2351 return VLC_ENOMEM;
2352
2353 for( ;; )
2354 {
2355 const char *s = TextGetLine( txt );
2356 size_t i_len;
2357 size_t i_old;
2358
2359 i_len = s ? strlen( s ) : 0;
2360 if( i_len <= 0 )
2361 {
2362 p_subtitle->psz_text = psz_text;
2363 return VLC_SUCCESS;
2364 }
2365
2366 i_old = strlen( psz_text );
2367 psz_text = realloc_or_free( psz_text, i_old + i_len + 1 + 1 );
2368 if( !psz_text )
2369 return VLC_ENOMEM;
2370
2371 strcat( psz_text, s );
2372 strcat( psz_text, "\n" );
2373 }
2374 }
2375
2376 static int ParseSCC( vlc_object_t *p_obj, subs_properties_t *p_props,
2377 text_t *txt, subtitle_t *p_subtitle, size_t i_idx )
2378 {
2379 VLC_UNUSED(p_obj);
2380 VLC_UNUSED( i_idx );
2381 VLC_UNUSED( p_props );
2382
2383 for( ;; )
2384 {
2385 const char *psz_line = TextGetLine( txt );
2386 if( !psz_line )
2387 return VLC_EGENERIC;
2388
2389 unsigned h, m, s, f;
2390 if( sscanf( psz_line, "%u:%u:%u:%u ", &h, &m, &s, &f ) != 4 )
2391 continue;
2392
2393 p_subtitle->i_start = CLOCK_FREQ * ( h * 3600 + m * 60 + s ) +
2394 f * p_props->i_microsecperframe;
2395 p_subtitle->i_stop = -1;
2396
2397 const char *psz_text = strchr( psz_line, '\t' );
2398 if( !psz_text && !(psz_text = strchr( psz_line, ' ' )) )
2399 continue;
2400
2401 if ( psz_text[1] == '\0' )
2402 continue;
2403
2404 p_subtitle->psz_text = strdup( psz_text + 1 );
2405 if( !p_subtitle->psz_text )
2406 return VLC_ENOMEM;
2407
2408 break;
2409 }
2410
2411 return VLC_SUCCESS;
2412 }
2413
2414 /* Matches filename.xx.srt */
2415 static char * get_language_from_filename( const char * psz_sub_file )
2416 {
2417 char *psz_ret = NULL;
2418 char *psz_tmp, *psz_language_begin;
2419
2420 if( !psz_sub_file ) return NULL;
2421 char *psz_work = strdup( psz_sub_file );
2422
2423 /* Removing extension, but leaving the dot */
2424 psz_tmp = strrchr( psz_work, '.' );
2425 if( psz_tmp )
2426 {
2427 psz_tmp[0] = '\0';
2428 psz_language_begin = strrchr( psz_work, '.' );
2429 if( psz_language_begin )
2430 psz_ret = strdup(++psz_language_begin);
2431 psz_tmp[0] = '.';
2432 }
2433
2434 free( psz_work );
2435 return psz_ret;
2436 }
VLC Development tree mirror