ss_sbm.c

harvest是一个下载html网页得机器人

/*
 * (c) Copyright 1992, 1993 by Panagiotis Tsirigotis
 * All rights reserved.  The file named COPYRIGHT specifies the terms 
 * and conditions for redistribution.
 */

static char RCSid[] = "ss_sbm.c,v 1.1.1.1 1994/04/26 19:08:11 hardy Exp" ;

char *malloc() ;

#include "ss_impl.h"
#include "ss_sbm.h"

PRIVATE int sbm_setup() ;
PRIVATE char *sbm_match() ;
PRIVATE void sbm_done() ;


struct ss_ops __strs_sbmops = { sbm_setup, sbm_match, sbm_done } ;


PRIVATE int sbm_setup( hp )
	header_s *hp ;
{
	last_int *last_occurrence ;
	register int i ;
	int patlen = SS_PATLEN( hp ) ;
	char *pattern = SS_PATTERN( hp ) ;

	last_occurrence = (last_int *) malloc( ALPHABET_SIZE * sizeof( last_int ) ) ;
	if ( last_occurrence == (last_int *)0 )
		return( SS_ERR ) ;
	
	for ( i = 0 ; i < ALPHABET_SIZE ; i++ )
		last_occurrence[ i ] = -1 ;
	for ( i = 0 ; i < patlen ; i++ )
		last_occurrence[ (unsigned char) pattern[ i ] ] = i ;

	SBM_HEADER( hp )->last_occurrence = last_occurrence ;
	return( SS_OK ) ;
}



PRIVATE char *sbm_match( hp, str, len )
	header_s *hp ;
	char *str ;
	int len ;
{
	register int s = 0 ;
	char *pattern = SS_PATTERN( hp ) ;
	int patlen = SS_PATLEN( hp ) ;
	last_int *last_occurrence = SBM_HEADER( hp )->last_occurrence ;
	register int j ;

	while ( s <= len - patlen )
	{
		register char c ;
		last_int lo ;

		/*
		 * Try matching pattern right-to-left
		 */
		for ( j = patlen-1 ;; )
		{
			c = SS_MAP( hp, str[ s+j ] ) ;
			if ( pattern[ j ] == c )
				if ( j )
					j-- ;
				else
					return( &str[ s ] ) ;
			else
				break ;
		}
		lo = last_occurrence[ (unsigned char) c ] ;
		if ( j > lo )
			s += j - lo ;
		else
			s++ ;
	}
	return( CHAR_NULL ) ;
}


PRIVATE void sbm_done( hp )
	header_s *hp ;
{
	(void) free( (char *) SBM_HEADER( hp )->last_occurrence ) ;
}