findurl.cpp

小型搜索引擎,用C/C++编写,属于全文搜索引擎

/*
 * If content is NULL, find out wheather INDEX_FILE includes the URL.
 * Otherwise, read the corresponding page of the URL.
 *
 */

#include "Tse.h"
#include "Url.h"

int FindUrl(const char *url, char **content)
{
	FILE *fpfile,*fpindex;
	char *buffer,file[URL_LEN+1];
	int offsett, offset2;

	if(content){
		fpfile = fopen(DATA_FILE_NAME.c_str(),"r");
		if(fpfile == NULL) return -1;
	}

	fpindex = fopen(INDEX_FILE_NAME.c_str(),"r");
	if(fpindex == NULL) return -1;

	while(!feof(fpindex)){
		fscanf(fpindex, "%d %s", &offsett, file);

		//cout << "offset: " << offsett << endl;
		//cout << "url: " << file << endl;
		
		if( strlen(url) > strlen(file) ){
			if( strncmp(url, file, strlen(url)) ){
				continue;
			}
		}else{
			if( strncmp(url, file, strlen(file)) ){
				continue;
			}
		}
		
		if(content == NULL){
			fclose(fpindex);
			return 0;
		}

		fseek( fpfile, offsett, 0 );
		fscanf( fpindex, "%d", &offset2 );
		offset2 = offset2 - offsett;

		buffer = ( char* )malloc( offset2+1 );
		if(buffer == NULL){
			fclose(fpfile);
			fclose(fpindex);
			return -1;
		}
		memset(buffer,0,offset2+1);
		fread(buffer, 1, offset2, fpfile);

		if(content == NULL)
			free(buffer);
		else
			*content = buffer;

		if(content) fclose(fpfile);
		fclose(fpindex);

		return 0;
	}
	
	if(content) fclose(fpfile);
	fclose(fpindex);

	return -1;
}