modules/demux/ttml.c

   1 /*****************************************************************************
   2  * ttml.c : TTML subtitles demux
   3  *****************************************************************************
   4  * Copyright (C) 2015-2017 VLC authors and VideoLAN
   5  *
   6  * Authors: Hugo Beauzée-Luyssen <hugo@beauzee.fr>
   7  *          Sushma Reddy <sushma.reddy@research.iiit.ac.in>
   8  *
   9  * This program is free software; you can redistribute it and/or modify it
  10  * under the terms of the GNU Lesser General Public License as published by
  11  * the Free Software Foundation; either version 2.1 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * This program is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  17  * GNU Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public License
  20  * along with this program; if not, write to the Free Software Foundation,
  21  * Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
  22  *****************************************************************************/
  23
  24 #ifdef HAVE_CONFIG_H
  25 # include "config.h"
  26 #endif
  27
  28 #include <vlc_common.h>
  29 #include <vlc_demux.h>
  30 #include <vlc_xml.h>
  31 #include <vlc_strings.h>
  32 #include <vlc_memstream.h>
  33 #include <vlc_es_out.h>
  34 #include <vlc_charset.h>          /* FromCharset */
  35
  36 #include <assert.h>
  37 #include <stdlib.h>
  38 #include <ctype.h>
  39
  40 #include "../codec/ttml/ttml.h"
  41
  42 //#define TTML_DEMUX_DEBUG
  43
  44 typedef struct
  45 {
  46     xml_t*          p_xml;
  47     xml_reader_t*   p_reader;
  48     es_out_id_t*    p_es;
  49     vlc_tick_t      i_next_demux_time;
  50     bool            b_slave;
  51     bool            b_first_time;
  52
  53     tt_node_t         *p_rootnode;
  54
  55     tt_timings_t    temporal_extent;
  56
  57     /*
  58      * All timings are stored unique and ordered.
  59      * Being begin or end times of sub sequence,
  60      * we use them as 'point of change' for output filtering.
  61     */
  62     struct
  63     {
  64         tt_time_t *p_array;
  65         size_t   i_count;
  66         size_t   i_current;
  67     } times;
  68 } demux_sys_t;
  69
  70 static char *tt_genTiming( tt_time_t t )
  71 {
  72     if( !tt_time_Valid( &t ) )
  73         t.base = 0;
  74     unsigned f = t.base % CLOCK_FREQ;
  75     t.base /= CLOCK_FREQ;
  76     unsigned h = t.base / 3600;
  77     unsigned m = t.base % 3600 / 60;
  78     unsigned s = t.base % 60;
  79
  80     int i_ret;
  81     char *psz;
  82     if( f )
  83     {
  84         const char *lz = "000000";
  85         const char *psz_lz = &lz[6];
  86         /* add leading zeroes */
  87         for( unsigned i=10*f; i<CLOCK_FREQ; i *= 10 )
  88             psz_lz--;
  89         /* strip trailing zeroes */
  90         for( ; f > 0 && (f % 10) == 0; f /= 10 );
  91         i_ret = asprintf( &psz, "%02u:%02u:%02u.%s%u",
  92                                  h, m, s, psz_lz, f );
  93     }
  94     else if( t.frames )
  95     {
  96         i_ret = asprintf( &psz, "%02u:%02u:%02u:%s%u",
  97                                  h, m, s, t.frames < 10 ? "0" : "", t.frames );
  98     }
  99     else
 100     {
 101         i_ret = asprintf( &psz, "%02u:%02u:%02u",
 102                                  h, m, s );
 103     }
 104
 105     return i_ret < 0 ? NULL : psz;
 106 }
 107
 108 static void tt_node_AttributesToText( struct vlc_memstream *p_stream, const tt_node_t* p_node )
 109 {
 110     bool b_timed_node = false;
 111     const vlc_dictionary_t* p_attr_dict = &p_node->attr_dict;
 112     for( int i = 0; i < p_attr_dict->i_size; ++i )
 113     {
 114         for ( vlc_dictionary_entry_t* p_entry = p_attr_dict->p_entries[i];
 115                                       p_entry != NULL; p_entry = p_entry->p_next )
 116         {
 117             const char *psz_value = NULL;
 118
 119             if( !strcmp(p_entry->psz_key, "begin") ||
 120                 !strcmp(p_entry->psz_key, "end") ||
 121                 !strcmp(p_entry->psz_key, "dur") )
 122             {
 123                 b_timed_node = true;
 124                 /* will remove duration */
 125                 continue;
 126             }
 127             else if( !strcmp(p_entry->psz_key, "timeContainer") )
 128             {
 129                 /* also remove sequential timings info (all abs now) */
 130                 continue;
 131             }
 132             else
 133             {
 134                 psz_value = (char const*)p_entry->p_value;
 135             }
 136
 137             if( psz_value == NULL )
 138                 continue;
 139
 140             vlc_memstream_printf( p_stream, " %s=\"%s\"",
 141                                   p_entry->psz_key, psz_value );
 142         }
 143     }
 144
 145     if( b_timed_node )
 146     {
 147         if( tt_time_Valid( &p_node->timings.begin ) )
 148         {
 149             char *psz = tt_genTiming( p_node->timings.begin );
 150             vlc_memstream_printf( p_stream, " begin=\"%s\"", psz );
 151             free( psz );
 152         }
 153
 154         if( tt_time_Valid( &p_node->timings.end ) )
 155         {
 156             char *psz = tt_genTiming( p_node->timings.end );
 157             vlc_memstream_printf( p_stream, " end=\"%s\"", psz );
 158             free( psz );
 159         }
 160     }
 161 }
 162
 163 static void tt_node_ToText( struct vlc_memstream *p_stream, const tt_basenode_t *p_basenode,
 164                             const tt_time_t *playbacktime )
 165 {
 166     if( p_basenode->i_type == TT_NODE_TYPE_ELEMENT )
 167     {
 168         const tt_node_t *p_node = (const tt_node_t *) p_basenode;
 169
 170         if( tt_time_Valid( playbacktime ) &&
 171            !tt_timings_Contains( &p_node->timings, playbacktime ) )
 172             return;
 173
 174         vlc_memstream_putc( p_stream, '<' );
 175         vlc_memstream_puts( p_stream, p_node->psz_node_name );
 176
 177         tt_node_AttributesToText( p_stream, p_node );
 178
 179         if( tt_node_HasChild( p_node ) )
 180         {
 181             vlc_memstream_putc( p_stream, '>' );
 182
 183 #ifdef TTML_DEMUX_DEBUG
 184             vlc_memstream_printf( p_stream, "<!-- starts %ld ends %ld -->",
 185                                   tt_time_Convert( &p_node->timings.begin ),
 186                                   tt_time_Convert( &p_node->timings.end ) );
 187 #endif
 188
 189             for( const tt_basenode_t *p_child = p_node->p_child;
 190                                    p_child; p_child = p_child->p_next )
 191             {
 192                 tt_node_ToText( p_stream, p_child, playbacktime );
 193             }
 194
 195             vlc_memstream_printf( p_stream, "</%s>", p_node->psz_node_name );
 196         }
 197         else
 198             vlc_memstream_puts( p_stream, "/>" );
 199     }
 200     else
 201     {
 202         const tt_textnode_t *p_textnode = (const tt_textnode_t *) p_basenode;
 203         vlc_memstream_puts( p_stream, p_textnode->psz_text );
 204     }
 205 }
 206
 207 static int Control( demux_t* p_demux, int i_query, va_list args )
 208 {
 209     demux_sys_t *p_sys = p_demux->p_sys;
 210     vlc_tick_t i64;
 211     double *pf, f;
 212     bool b;
 213
 214     switch( i_query )
 215     {
 216         case DEMUX_CAN_SEEK:
 217             *va_arg( args, bool * ) = true;
 218             return VLC_SUCCESS;
 219         case DEMUX_GET_TIME:
 220             *va_arg( args, vlc_tick_t * ) = p_sys->i_next_demux_time;
 221             return VLC_SUCCESS;
 222         case DEMUX_SET_TIME:
 223             if( p_sys->times.i_count )
 224             {
 225                 tt_time_t t = tt_time_Create( va_arg( args, vlc_tick_t ) - VLC_TICK_0 );
 226                 size_t i_index = tt_timings_FindLowerIndex( p_sys->times.p_array,
 227                                                             p_sys->times.i_count, t, &b );
 228                 p_sys->times.i_current = i_index;
 229                 p_sys->b_first_time = true;
 230                 return VLC_SUCCESS;
 231             }
 232             break;
 233         case DEMUX_SET_NEXT_DEMUX_TIME:
 234             p_sys->i_next_demux_time = va_arg( args, vlc_tick_t );
 235             p_sys->b_slave = true;
 236             return VLC_SUCCESS;
 237         case DEMUX_GET_LENGTH:
 238             if( p_sys->times.i_count )
 239             {
 240                 tt_time_t t = tt_time_Sub( p_sys->times.p_array[p_sys->times.i_count - 1],
 241                                            p_sys->temporal_extent.begin );
 242                 *va_arg( args, vlc_tick_t * ) = tt_time_Convert( &t );
 243                 return VLC_SUCCESS;
 244             }
 245             break;
 246         case DEMUX_GET_POSITION:
 247             pf = va_arg( args, double * );
 248             if( p_sys->times.i_current >= p_sys->times.i_count )
 249             {
 250                 *pf = 1.0;
 251             }
 252             else if( p_sys->times.i_count > 0 )
 253             {
 254                 i64 = tt_time_Convert( &p_sys->times.p_array[p_sys->times.i_count - 1] );
 255                 *pf = (double) p_sys->i_next_demux_time / (i64 + VLC_TICK_FROM_MS(500));
 256             }
 257             else
 258             {
 259                 *pf = 0.0;
 260             }
 261             return VLC_SUCCESS;
 262         case DEMUX_SET_POSITION:
 263             f = va_arg( args, double );
 264             if( p_sys->times.i_count )
 265             {
 266                 i64 = f * tt_time_Convert( &p_sys->times.p_array[p_sys->times.i_count - 1] );
 267                 tt_time_t t = tt_time_Create( i64 );
 268                 size_t i_index = tt_timings_FindLowerIndex( p_sys->times.p_array,
 269                                                             p_sys->times.i_count, t, &b );
 270                 p_sys->times.i_current = i_index;
 271                 p_sys->b_first_time = true;
 272                 return VLC_SUCCESS;
 273             }
 274             break;
 275         case DEMUX_CAN_PAUSE:
 276         case DEMUX_SET_PAUSE_STATE:
 277         case DEMUX_CAN_CONTROL_PACE:
 278             return demux_vaControlHelper( p_demux->s, 0, -1, 0, 1, i_query, args );
 279
 280         case DEMUX_GET_PTS_DELAY:
 281         case DEMUX_GET_FPS:
 282         case DEMUX_GET_META:
 283         case DEMUX_GET_ATTACHMENTS:
 284         case DEMUX_GET_TITLE_INFO:
 285         case DEMUX_HAS_UNSUPPORTED_META:
 286         case DEMUX_CAN_RECORD:
 287         default:
 288             break;
 289     }
 290
 291     return VLC_EGENERIC;
 292 }
 293
 294 static int ReadTTML( demux_t* p_demux )
 295 {
 296     demux_sys_t* p_sys = p_demux->p_sys;
 297     const char* psz_node_name;
 298
 299     do
 300     {
 301         int i_type = xml_ReaderNextNode( p_sys->p_reader, &psz_node_name );
 302         bool b_empty = xml_ReaderIsEmptyElement( p_sys->p_reader );
 303
 304         if( i_type <= XML_READER_NONE )
 305             break;
 306
 307         switch(i_type)
 308         {
 309             default:
 310                 break;
 311
 312             case XML_READER_STARTELEM:
 313                 if( tt_node_NameCompare( psz_node_name, "tt" ) ||
 314                     p_sys->p_rootnode != NULL )
 315                     return VLC_EGENERIC;
 316
 317                 p_sys->p_rootnode = tt_node_New( p_sys->p_reader, NULL, psz_node_name );
 318                 if( b_empty )
 319                     break;
 320                 if( !p_sys->p_rootnode ||
 321                     tt_nodes_Read( p_sys->p_reader, p_sys->p_rootnode ) != VLC_SUCCESS )
 322                     return VLC_EGENERIC;
 323                 break;
 324
 325             case XML_READER_ENDELEM:
 326                 if( !p_sys->p_rootnode ||
 327                     tt_node_NameCompare( psz_node_name, p_sys->p_rootnode->psz_node_name ) )
 328                     return VLC_EGENERIC;
 329                 break;
 330         }
 331
 332     } while( 1 );
 333
 334     if( p_sys->p_rootnode == NULL )
 335         return VLC_EGENERIC;
 336
 337     return VLC_SUCCESS;
 338 }
 339
 340 static int Demux( demux_t* p_demux )
 341 {
 342     demux_sys_t* p_sys = p_demux->p_sys;
 343
 344     /* Last one must be an end time */
 345     while( p_sys->times.i_current + 1 < p_sys->times.i_count &&
 346            tt_time_Convert( &p_sys->times.p_array[p_sys->times.i_current] ) <= p_sys->i_next_demux_time )
 347     {
 348         const vlc_tick_t i_playbacktime =
 349                 tt_time_Convert( &p_sys->times.p_array[p_sys->times.i_current] );
 350         const vlc_tick_t i_playbackendtime =
 351                 tt_time_Convert( &p_sys->times.p_array[p_sys->times.i_current + 1] ) - 1;
 352
 353         if ( !p_sys->b_slave && p_sys->b_first_time )
 354         {
 355             es_out_SetPCR( p_demux->out, VLC_TICK_0 + i_playbacktime );
 356             p_sys->b_first_time = false;
 357         }
 358
 359         struct vlc_memstream stream;
 360
 361         if( vlc_memstream_open( &stream ) )
 362             return VLC_DEMUXER_EGENERIC;
 363
 364         tt_node_ToText( &stream, (tt_basenode_t *) p_sys->p_rootnode,
 365                         &p_sys->times.p_array[p_sys->times.i_current] );
 366
 367         if( vlc_memstream_close( &stream ) == VLC_SUCCESS )
 368         {
 369             block_t* p_block = block_heap_Alloc( stream.ptr, stream.length );
 370             if( p_block )
 371             {
 372                 p_block->i_dts =
 373                     p_block->i_pts = VLC_TICK_0 + i_playbacktime;
 374                 p_block->i_length = i_playbackendtime - i_playbacktime;
 375
 376                 es_out_Send( p_demux->out, p_sys->p_es, p_block );
 377             }
 378         }
 379
 380         p_sys->times.i_current++;
 381     }
 382
 383     if ( !p_sys->b_slave )
 384     {
 385         es_out_SetPCR( p_demux->out, VLC_TICK_0 + p_sys->i_next_demux_time );
 386         p_sys->i_next_demux_time += VLC_TICK_FROM_MS(125);
 387     }
 388
 389     if( p_sys->times.i_current + 1 >= p_sys->times.i_count )
 390         return VLC_DEMUXER_EOF;
 391
 392     return VLC_DEMUXER_SUCCESS;
 393 }
 394
 395 int tt_OpenDemux( vlc_object_t* p_this )
 396 {
 397     demux_t     *p_demux = (demux_t*)p_this;
 398     demux_sys_t *p_sys;
 399
 400     const uint8_t *p_peek;
 401     ssize_t i_peek = vlc_stream_Peek( p_demux->s, &p_peek, 2048 );
 402     if( unlikely( i_peek <= 32 ) )
 403         return VLC_EGENERIC;
 404
 405     const char *psz_xml = (const char *) p_peek;
 406     size_t i_xml  = i_peek;
 407
 408     /* Try to probe without xml module/loading the full document */
 409     char *psz_alloc = NULL;
 410     switch( GetQWBE(p_peek) )
 411     {
 412         /* See RFC 3023 Part 4 */
 413         case UINT64_C(0xFFFE3C003F007800): /* UTF16 BOM<? */
 414         case UINT64_C(0xFFFE3C003F007400): /* UTF16 BOM<t */
 415         case UINT64_C(0xFEFF003C003F0078): /* UTF16 BOM<? */
 416         case UINT64_C(0xFEFF003C003F0074): /* UTF16 BOM<t */
 417             psz_alloc = FromCharset( "UTF-16", p_peek, i_peek );
 418             break;
 419         case UINT64_C(0x3C003F0078006D00): /* UTF16-LE <?xm */
 420         case UINT64_C(0x3C003F0074007400): /* UTF16-LE <tt */
 421             psz_alloc = FromCharset( "UTF-16LE", p_peek, i_peek );
 422             break;
 423         case UINT64_C(0x003C003F0078006D): /* UTF16-BE <?xm */
 424         case UINT64_C(0x003C003F00740074): /* UTF16-BE <tt */
 425             psz_alloc = FromCharset( "UTF-16BE", p_peek, i_peek );
 426             break;
 427         case UINT64_C(0xEFBBBF3C3F786D6C): /* UTF8 BOM<?xml */
 428         case UINT64_C(0x3C3F786D6C207665): /* UTF8 <?xml ve */
 429         case UINT64_C(0xEFBBBF3C74742078): /* UTF8 BOM<tt x*/
 430             break;
 431         default:
 432             if(GetDWBE(p_peek) != UINT32_C(0x3C747420)) /* tt node without xml document marker */
 433                 return VLC_EGENERIC;
 434     }
 435
 436     if( psz_alloc )
 437     {
 438         psz_xml = psz_alloc;
 439         i_xml = strlen( psz_alloc );
 440     }
 441
 442     /* Simplified probing. Valid TTML must have a namespace declaration */
 443     const char *psz_tt = strnstr( psz_xml, "tt", i_xml );
 444     if( !psz_tt || psz_tt == psz_xml ||
 445         ((size_t)(&psz_tt[2] - (const char*)p_peek)) == i_xml || isalpha(psz_tt[2]) ||
 446         (psz_tt[-1] != ':' && psz_tt[-1] != '<') )
 447     {
 448         free( psz_alloc );
 449         return VLC_EGENERIC;
 450     }
 451     else
 452     {
 453         const char * const rgsz[] =
 454         {
 455             "=\"http://www.w3.org/ns/ttml\"",
 456             "=\"http://www.w3.org/2004/11/ttaf1\"",
 457             "=\"http://www.w3.org/2006/04/ttaf1\"",
 458             "=\"http://www.w3.org/2006/10/ttaf1\"",
 459         };
 460         const char *psz_ns = NULL;
 461         for( size_t i=0; i<ARRAY_SIZE(rgsz) && !psz_ns; i++ )
 462         {
 463             psz_ns = strnstr( psz_xml, rgsz[i],
 464                               i_xml - (psz_tt - psz_xml) );
 465         }
 466         free( psz_alloc );
 467         if( !psz_ns )
 468             return VLC_EGENERIC;
 469     }
 470
 471     p_demux->p_sys = p_sys = calloc( 1, sizeof( *p_sys ) );
 472     if( unlikely( p_sys == NULL ) )
 473         return VLC_ENOMEM;
 474
 475     p_sys->b_first_time = true;
 476     p_sys->temporal_extent.i_type = TT_TIMINGS_PARALLEL;
 477     tt_time_Init( &p_sys->temporal_extent.begin );
 478     tt_time_Init( &p_sys->temporal_extent.end );
 479     tt_time_Init( &p_sys->temporal_extent.dur );
 480     p_sys->temporal_extent.begin.base = 0;
 481
 482     p_sys->p_xml = xml_Create( p_demux );
 483     if( !p_sys->p_xml )
 484         goto error;
 485
 486     p_sys->p_reader = xml_ReaderCreate( p_sys->p_xml, p_demux->s );
 487     if( !p_sys->p_reader )
 488         goto error;
 489
 490 #ifndef TTML_DEMUX_DEBUG
 491     p_sys->p_reader->obj.flags |= OBJECT_FLAGS_QUIET;
 492 #endif
 493
 494     if( ReadTTML( p_demux ) != VLC_SUCCESS )
 495         goto error;
 496
 497     tt_timings_Resolve( (tt_basenode_t *) p_sys->p_rootnode, &p_sys->temporal_extent,
 498                         &p_sys->times.p_array, &p_sys->times.i_count );
 499
 500 #ifdef TTML_DEMUX_DEBUG
 501     {
 502         struct vlc_memstream stream;
 503
 504         if( vlc_memstream_open( &stream ) )
 505             goto error;
 506
 507         tt_time_t t;
 508         tt_time_Init( &t );
 509         tt_node_ToText( &stream, (tt_basenode_t*)p_sys->p_rootnode, &t /* invalid */ );
 510
 511         vlc_memstream_putc( &stream, '\0' );
 512
 513         if( vlc_memstream_close( &stream ) == VLC_SUCCESS )
 514         {
 515             msg_Dbg( p_demux, "%s", stream.ptr );
 516             free( stream.ptr );
 517         }
 518     }
 519 #endif
 520
 521     p_demux->pf_demux = Demux;
 522     p_demux->pf_control = Control;
 523
 524     es_format_t fmt;
 525     es_format_Init( &fmt, SPU_ES, VLC_CODEC_TTML );
 526     p_sys->p_es = es_out_Add( p_demux->out, &fmt );
 527     if( !p_sys->p_es )
 528         goto error;
 529
 530     es_format_Clean( &fmt );
 531
 532     return VLC_SUCCESS;
 533
 534 error:
 535     tt_CloseDemux( p_demux );
 536
 537     return VLC_EGENERIC;
 538 }
 539
 540 void tt_CloseDemux( demux_t* p_demux )
 541 {
 542     demux_sys_t* p_sys = p_demux->p_sys;
 543
 544     if( p_sys->p_rootnode )
 545         tt_node_RecursiveDelete( p_sys->p_rootnode );
 546
 547     if( p_sys->p_es )
 548         es_out_Del( p_demux->out, p_sys->p_es );
 549
 550     if( p_sys->p_reader )
 551         xml_ReaderDelete( p_sys->p_reader );
 552
 553     if( p_sys->p_xml )
 554         xml_Delete( p_sys->p_xml );
 555
 556     free( p_sys->times.p_array );
 557
 558     free( p_sys );
 559 }