libhb/decsrtsub.c

   1 /*
   2    This file is part of the HandBrake source code.
   3    Homepage: <http://handbrake.fr/>.
   4    It may be used under the terms of the GNU General Public License. */
   5
   6 #include <stdlib.h>
   7 #include <stdio.h>
   8 #include <string.h>
   9 #include <iconv.h>
  10 #include <errno.h>
  11 #include "hb.h"
  12
  13 struct start_and_end {
  14     unsigned long start, end;
  15 };
  16
  17 enum
  18 {
  19     k_state_inEntry,
  20     k_state_potential_new_entry,
  21     k_state_timecode,
  22 };
  23
  24 typedef struct srt_entry_s {
  25     long offset, duration;
  26     long start, stop;
  27     char text[1024];
  28 } srt_entry_t;
  29
  30 /*
  31  * Store all context in the work private struct,
  32  */
  33 struct hb_work_private_s
  34 {
  35     hb_job_t *job;
  36     FILE *file;
  37     unsigned long current_time;
  38     unsigned long number_of_entries;
  39     unsigned long current_state;
  40     srt_entry_t current_entry;
  41     iconv_t *iconv_context;
  42     hb_subtitle_t *subtitle;
  43     uint64_t start_time;              // In HB time
  44     uint64_t stop_time;               // In HB time
  45 };
  46
  47 static struct start_and_end read_time_from_string( const char* timeString )
  48 {
  49     // for ex. 00:00:15,248 --> 00:00:16,545
  50
  51     long houres1, minutes1, seconds1, milliseconds1,
  52         houres2, minutes2, seconds2, milliseconds2;
  53
  54     sscanf(timeString, "%ld:%ld:%ld,%ld --> %ld:%ld:%ld,%ld\n", &houres1, &minutes1, &seconds1, &milliseconds1,
  55            &houres2, &minutes2, &seconds2, &milliseconds2);
  56
  57     struct start_and_end result = {
  58         milliseconds1 + seconds1*1000 + minutes1*60*1000 + houres1*60*60*1000,
  59         milliseconds2 + seconds2*1000 + minutes2*60*1000 + houres2*60*60*1000};
  60     return result;
  61 }
  62
  63 /*
  64  * Read the SRT file and put the entries into the subtitle fifo for all to read
  65  */
  66 static hb_buffer_t *srt_read( hb_work_private_t *pv )
  67 {
  68
  69     char line_buffer[1024];
  70
  71     if( !pv->file )
  72     {
  73         return NULL;
  74     }
  75
  76     while( fgets( line_buffer, sizeof( line_buffer ), pv->file ) )
  77     {
  78         switch (pv->current_state)
  79         {
  80         case k_state_timecode:
  81         {
  82             struct start_and_end timing = read_time_from_string( line_buffer );
  83             pv->current_entry.duration = timing.end - timing.start;
  84             pv->current_entry.offset = timing.start - pv->current_time;
  85
  86             pv->current_time = timing.end;
  87
  88             pv->current_entry.start = timing.start;
  89             pv->current_entry.stop = timing.end;
  90
  91             pv->current_state = k_state_inEntry;
  92             continue;
  93         }
  94
  95         case k_state_inEntry:
  96         {
  97             char *p, *q;
  98             size_t in_size;
  99             size_t out_size;
 100             size_t retval;
 101
 102             // If the current line is empty, we assume this is the
 103             //  seperation betwene two entries. In case we are wrong,
 104             //  the mistake is corrected in the next state.
 105             if (strcmp(line_buffer, "\n") == 0 || strcmp(line_buffer, "\r\n") == 0) {
 106                 pv->current_state = k_state_potential_new_entry;
 107                 continue;
 108             }
 109
 110
 111             for( q = pv->current_entry.text; (q < pv->current_entry.text+1024) && *q; q++);
 112
 113             p = line_buffer;
 114
 115             in_size = strlen(line_buffer);
 116             out_size = (pv->current_entry.text+1024) - q;
 117
 118             retval = iconv( pv->iconv_context, &p, &in_size, &q, &out_size);
 119             *q = '\0';
 120
 121             if( ( retval == -1 ) && ( errno == EINVAL ) )
 122             {
 123                 hb_error( "Invalid shift sequence" );
 124             } else if ( ( retval == -1 ) && ( errno == EILSEQ ) )
 125             {
 126                 hb_error( "Invalid byte for codeset in input, %"PRId64" bytes discarded", (int64_t)in_size);
 127             } else if ( ( retval == -1 ) && ( errno == E2BIG ) )
 128             {
 129                 hb_error( "Not enough space in output buffer");
 130             }
 131
 132             break;
 133         }
 134
 135         case k_state_potential_new_entry:
 136         {
 137             const char endpoint[] = "\0";
 138             const unsigned long potential_entry_number = strtol(line_buffer, (char**)&endpoint, 10);
 139             hb_buffer_t *buffer = NULL;
 140             /*
 141              * Is this really new next entry begin?
 142              */
 143             if (potential_entry_number == pv->number_of_entries + 1) {
 144                 /*
 145                  * We found the next entry - or a really rare error condition
 146                  */
 147                 if( *pv->current_entry.text )
 148                 {
 149                     long length;
 150                     char *p;
 151                     uint64_t start_time = ( pv->current_entry.start +
 152                                             pv->subtitle->config.offset ) * 90;
 153                     uint64_t stop_time = ( pv->current_entry.stop +
 154                                            pv->subtitle->config.offset ) * 90;
 155
 156                     if( !( start_time > pv->start_time && stop_time < pv->stop_time ) )
 157                     {
 158                         hb_deep_log( 3, "Discarding SRT at time start %"PRId64", stop %"PRId64, start_time, stop_time);
 159                         memset( &pv->current_entry, 0, sizeof( srt_entry_t ) );
 160                         ++(pv->number_of_entries);
 161                         pv->current_state = k_state_timecode;
 162                         continue;
 163                     }
 164
 165                     length = strlen( pv->current_entry.text );
 166
 167                     for( p = pv->current_entry.text; *p; p++)
 168                     {
 169                         if( *p == '\n' || *p == '\r' )
 170                         {
 171                             *p = ' ';
 172                         }
 173                     }
 174
 175                     buffer = hb_buffer_init( length + 1 );
 176
 177                     if( buffer )
 178                     {
 179                         buffer->start = start_time - pv->start_time;
 180                         buffer->stop = stop_time - pv->start_time;
 181
 182                         memcpy( buffer->data, pv->current_entry.text, length + 1 );
 183                     }
 184                 }
 185                 memset( &pv->current_entry, 0, sizeof( srt_entry_t ) );
 186                 ++(pv->number_of_entries);
 187                 pv->current_state = k_state_timecode;
 188                 if( buffer )
 189                 {
 190                     return buffer;
 191                 }
 192                 continue;
 193             } else {
 194                 /*
 195                  * Well.. looks like we are in the wrong mode.. lets add the
 196                  * newline we misinterpreted...
 197                  */
 198                 strncat(pv->current_entry.text, " ", 1024);
 199                 pv->current_state = k_state_inEntry;
 200             }
 201
 202             break;
 203         }
 204         }
 205     }
 206
 207     return NULL;
 208 }
 209
 210 static int decsrtInit( hb_work_object_t * w, hb_job_t * job )
 211 {
 212     int retval = 1;
 213     hb_work_private_t * pv;
 214     hb_buffer_t *buffer;
 215     int i;
 216     hb_chapter_t * chapter;
 217     hb_title_t *title = job->title;
 218
 219     pv = calloc( 1, sizeof( hb_work_private_t ) );
 220     if( pv )
 221     {
 222         w->private_data = pv;
 223
 224         pv->job = job;
 225
 226         buffer = hb_buffer_init( 0 );
 227         hb_fifo_push( w->fifo_in, buffer);
 228
 229         pv->file = fopen( w->subtitle->config.src_filename, "r" );
 230
 231         pv->current_state = k_state_potential_new_entry;
 232         pv->number_of_entries = 0;
 233         pv->current_time = 0;
 234         pv->subtitle = w->subtitle;
 235
 236         /*
 237          * Figure out the start and stop times from teh chapters being
 238          * encoded - drop subtitle not in this range.
 239          */
 240         pv->start_time = 0;
 241         for( i = 1; i < job->chapter_start; ++i )
 242         {
 243             chapter = hb_list_item( title->list_chapter, i - 1 );
 244             if( chapter )
 245             {
 246                 pv->start_time += chapter->duration;
 247             } else {
 248                 hb_error( "Could not locate chapter %d for SRT start time", i );
 249                 retval = 0;
 250             }
 251         }
 252         chapter = hb_list_item( title->list_chapter, i - 1 );
 253
 254         if( chapter )
 255         {
 256             pv->stop_time = pv->start_time + chapter->duration;
 257         } else {
 258             hb_error( "Could not locate chapter %d for SRT stop time", i );
 259             retval = 0;
 260         }
 261
 262         hb_deep_log( 3, "SRT Start time %"PRId64", stop time %"PRId64, pv->start_time, pv->stop_time);
 263
 264         pv->iconv_context = iconv_open( "utf8", pv->subtitle->config.src_codeset );
 265
 266
 267         if( pv->iconv_context == (iconv_t) -1 )
 268         {
 269             hb_error("Could not open the iconv library with those file formats\n");
 270
 271         } else {
 272             memset( &pv->current_entry, 0, sizeof( srt_entry_t ) );
 273
 274             pv->file = fopen( w->subtitle->config.src_filename, "r" );
 275
 276             if( !pv->file )
 277             {
 278                 hb_error("Could not open the SRT subtitle file '%s'\n",
 279                          w->subtitle->config.src_filename);
 280             } else {
 281                 retval = 0;
 282             }
 283         }
 284     }
 285
 286     return retval;
 287 }
 288
 289 static int decsrtWork( hb_work_object_t * w, hb_buffer_t ** buf_in,
 290                        hb_buffer_t ** buf_out )
 291 {
 292     hb_work_private_t * pv = w->private_data;
 293     hb_buffer_t * in = *buf_in;
 294     hb_buffer_t * out = NULL;
 295
 296     out = srt_read( pv );
 297
 298     if( out )
 299     {
 300         /*
 301          * Keep a buffer in our input fifo so that we get run.
 302          */
 303         hb_fifo_push( w->fifo_in, in);
 304         *buf_in = NULL;
 305         *buf_out = out;
 306     } else {
 307         *buf_out = NULL;
 308         return HB_WORK_OK;
 309     }
 310
 311     return HB_WORK_OK;
 312 }
 313
 314 static void decsrtClose( hb_work_object_t * w )
 315 {
 316     hb_work_private_t * pv = w->private_data;
 317     fclose( pv->file );
 318     iconv_close(pv->iconv_context);
 319     free( w->private_data );
 320 }
 321
 322 hb_work_object_t hb_decsrtsub =
 323 {
 324     WORK_DECSRTSUB,
 325     "SRT Subtitle Decoder",
 326     decsrtInit,
 327     decsrtWork,
 328     decsrtClose
 329 };