|
Packit |
89ede9 |
/*
|
|
Packit |
89ede9 |
* pages.c
|
|
Packit |
89ede9 |
*
|
|
Packit |
89ede9 |
* Copyright (c) Chris Putnam 2016-2018
|
|
Packit |
89ede9 |
*
|
|
Packit |
89ede9 |
* Program and source code released under GPL verison 2
|
|
Packit |
89ede9 |
*/
|
|
Packit |
89ede9 |
#include <stdio.h>
|
|
Packit |
89ede9 |
#include <stdlib.h>
|
|
Packit |
89ede9 |
#include <string.h>
|
|
Packit |
89ede9 |
#include "is_ws.h"
|
|
Packit |
89ede9 |
#include "utf8.h"
|
|
Packit |
89ede9 |
#include "pages.h"
|
|
Packit |
89ede9 |
|
|
Packit |
89ede9 |
/* extract_range()
|
|
Packit |
89ede9 |
*
|
|
Packit |
89ede9 |
* Handle input strings like:
|
|
Packit |
89ede9 |
*
|
|
Packit |
89ede9 |
* "1-15"
|
|
Packit |
89ede9 |
* " 1 - 15 "
|
|
Packit |
89ede9 |
* " 1000--- 1500"
|
|
Packit |
89ede9 |
* " 1 <<em-dash>> 10"
|
|
Packit |
89ede9 |
* " 107 111"
|
|
Packit |
89ede9 |
*/
|
|
Packit |
89ede9 |
static void
|
|
Packit |
89ede9 |
extract_range( str *input, str *begin, str *end )
|
|
Packit |
89ede9 |
{
|
|
Packit |
89ede9 |
/* -30 is the first character of a UTF8 em-dash and en-dash */
|
|
Packit |
89ede9 |
const char terminators[] = { ' ', '-', '\t', '\r', '\n', -30, '\0' };
|
|
Packit |
89ede9 |
char *p;
|
|
Packit |
89ede9 |
|
|
Packit |
89ede9 |
str_empty( begin );
|
|
Packit |
89ede9 |
str_empty( end );
|
|
Packit |
89ede9 |
|
|
Packit |
89ede9 |
if ( input->len==0 ) return;
|
|
Packit |
89ede9 |
|
|
Packit |
89ede9 |
p = skip_ws( input->data );
|
|
Packit |
89ede9 |
while ( *p && !strchr( terminators, *p ) )
|
|
Packit |
89ede9 |
str_addchar( begin, *p++ );
|
|
Packit |
89ede9 |
|
|
Packit |
89ede9 |
p = skip_ws( p );
|
|
Packit |
89ede9 |
|
|
Packit |
89ede9 |
while ( *p=='-' ) p++;
|
|
Packit |
89ede9 |
while ( utf8_is_emdash( p ) ) p+=3;
|
|
Packit |
89ede9 |
while ( utf8_is_endash( p ) ) p+=3;
|
|
Packit |
89ede9 |
|
|
Packit |
89ede9 |
p = skip_ws( p );
|
|
Packit |
89ede9 |
|
|
Packit |
89ede9 |
while ( *p && !strchr( terminators, *p ) )
|
|
Packit |
89ede9 |
str_addchar( end, *p++ );
|
|
Packit |
89ede9 |
}
|
|
Packit |
89ede9 |
|
|
Packit |
89ede9 |
int
|
|
Packit |
89ede9 |
pages_add( fields *bibout, char *outtag, str *invalue, int level )
|
|
Packit |
89ede9 |
{
|
|
Packit |
89ede9 |
int fstatus, status = 1;
|
|
Packit |
89ede9 |
str start, stop;
|
|
Packit |
89ede9 |
|
|
Packit |
89ede9 |
str_init( &start );
|
|
Packit |
89ede9 |
str_init( &stop );
|
|
Packit |
89ede9 |
|
|
Packit |
89ede9 |
extract_range( invalue, &start, &stop );
|
|
Packit |
89ede9 |
|
|
Packit |
89ede9 |
if ( str_memerr( &start ) || str_memerr( &stop ) ) {
|
|
Packit |
89ede9 |
status = 0;
|
|
Packit |
89ede9 |
goto out;
|
|
Packit |
89ede9 |
}
|
|
Packit |
89ede9 |
|
|
Packit |
89ede9 |
if ( start.len>0 ) {
|
|
Packit |
89ede9 |
fstatus = fields_add( bibout, "PAGES:START", start.data, level );
|
|
Packit |
89ede9 |
if ( fstatus!=FIELDS_OK ) {
|
|
Packit |
89ede9 |
status = 0;
|
|
Packit |
89ede9 |
goto out;
|
|
Packit |
89ede9 |
}
|
|
Packit |
89ede9 |
}
|
|
Packit |
89ede9 |
|
|
Packit |
89ede9 |
if ( stop.len>0 ) {
|
|
Packit |
89ede9 |
fstatus = fields_add( bibout, "PAGES:STOP", stop.data, level );
|
|
Packit |
89ede9 |
if ( fstatus!=FIELDS_OK ) status = 0;
|
|
Packit |
89ede9 |
}
|
|
Packit |
89ede9 |
|
|
Packit |
89ede9 |
out:
|
|
Packit |
89ede9 |
str_free( &start );
|
|
Packit |
89ede9 |
str_free( &stop );
|
|
Packit |
89ede9 |
return status;
|
|
Packit |
89ede9 |
}
|
|
Packit |
89ede9 |
|