static int
is_srt(const char *buf, size_t len)
{
linereader_t lr;
int n;
int64_t start, stop;
srt_skip_preamble(&buf, &len);
linereader_init(&lr, buf, len);
if(linereader_next(&lr) < 0)
return 0;
if(get_int(&lr, &n))
return 0;
if(linereader_next(&lr) < 0)
return 0;
if(get_srt_timestamp(&lr, &start, &stop))
return 0;
if(stop < start)
return 0;
return 1;
}
static ext_subtitles_t *
load_srt(const char *url, const char *buf, size_t len)
{
int n;
size_t tlen = 0;
int64_t start, stop, pstart = -1, pstop = -1;
linereader_t lr;
ext_subtitles_t *es = calloc(1, sizeof(ext_subtitles_t));
char *txt = NULL;
size_t txtoff = 0;
const int tag_flags = TEXT_PARSE_HTML_TAGS | TEXT_PARSE_HTML_ENTITIES |
TEXT_PARSE_SLOPPY_TAGS | TEXT_PARSE_SUB_TAGS;
srt_skip_preamble(&buf, &len);
TAILQ_INIT(&es->es_entries);
linereader_init(&lr, buf, len);
while(1) {
if((n = linereader_next(&lr)) < 0)
break;
if(get_srt_timestamp(&lr, &start, &stop) == 0) {
if(txt != NULL && pstart != -1 && pstop != -1) {
txt[txtoff] = 0;
es_insert_text(es, txt, pstart, pstop, tag_flags);
free(txt);
txt = NULL;
tlen = 0;
txtoff = 0;
}
pstart = start;
pstop = stop;
continue;
}
if(pstart == -1)
continue;
txt = realloc(txt, tlen + lr.ll + 1);
memcpy(txt + tlen, lr.buf, lr.ll);
txt[tlen + lr.ll] = 0x0a;
if(lr.ll == 0 && tlen > 0)
txtoff = tlen - 1;
tlen += lr.ll + 1;
}
if(txt != NULL && pstart != -1 && pstop != -1) {
txt[txtoff] = 0;
es_insert_text(es, txt, pstart, pstop, tag_flags);
}
free(txt);
return es;
}
int
main( int argc, char *argv[] )
{
FILE *out_a, *out_b;
char *h1, *h2, *s, *q;
char *id = NULL;
char *outname_a = NULL, *outname_b = NULL, *outdirname = NULL;
int i;
struct options *o = options_new( );
struct tokenset *t = tokenset_new( );
struct fqreader *fq = fqreader_new( NULL );
/* Get the command line options */
options_cmdline( o, argc, argv );
if ( _IS_NULL( o->outname ) || strlen( o->outname ) == 0 ) {
o->outname = realloc( o->outname, 4 * sizeof ( char ) );
strcpy( o->outname, "FQU" );
}
/* Create the output directory if necessary */
outdirname = realloc( outdirname, sizeof ( char ) * ( 2 + strlen( o->outname ) ) );
strcpy( outdirname, o->outname );
if ( strchr( outdirname, '/' ) ) { /* not the current directory */
dirname( outdirname );
mkdirp( outdirname, S_IRWXU | S_IRWXG | S_IRWXO );
}
/* Read the id files */
for ( i = o->optind; i < argc; i++ ) {
char *line;
struct linereader *z = linereader_new( );
linereader_init( z, argv[i] );
if ( _IS_NULL( z ) ) {
fprintf( stderr, "[ERROR] %s: Cannot open input file \"%s\"\n", _I_AM, argv[i] );
exit( 1 );
}
while ( ( line = ( char * ) linereader_next( z ) ) ) {
unsigned len;
stru_trim( line );
if ( line[0] == '#' || stru_is_ws( line ) )
continue;
len = strcspn( line, "\f\n\r\t\v " );
line[len] = '\0';
tokenset_add( t, line );
}
linereader_free( z );
}
outname_a = realloc( outname_a, sizeof ( char ) * ( 6 + strlen( o->outname ) ) );
strcpy( outname_a, o->outname );
strcat( outname_a, "_A.fq" );
out_a = fopen( outname_a, "wb" );
if ( _IS_NULL( out_a ) ) {
fprintf( stderr, "Could not open first output file \"%s\"\n", outname_a );
exit( 1 );
}
outname_b = realloc( outname_b, sizeof ( char ) * ( 6 + strlen( o->outname ) ) );
strcpy( outname_b, o->outname );
strcat( outname_b, "_B.fq" );
out_b = fopen( outname_b, "wb" );
if ( _IS_NULL( out_b ) ) {
fprintf( stderr, "Could not open second output file \"%s\"\n", outname_b );
exit( 1 );
}
while ( fqreader_next( fq, &h1, &h2, &s, &q ) ) {
utils_extract_id( &id, h1 ); /* get the identifier from header 1 */
if ( tokenset_exists( t, id ) )
fprintf( out_a, "@%s\n%s\n+%s\n%s\n", h1, s, h2, q );
else
fprintf( out_b, "@%s\n%s\n+%s\n%s\n", h1, s, h2, q );
}
fclose( out_a );
fclose( out_b );
_FREE( id );
_FREE( outname_a );
_FREE( outname_b );
fqreader_free( fq );
options_free( o );
tokenset_free( t );
return 0;
}
