findkey.cpp

小型搜索引擎,用C/C++编写,属于全文搜索引擎

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

using namespace std;

extern int FindUrl(const char* url, char **content);

/*
 * Find results pages containing query words.
 * Return the number of results.
 */
int FindKey(const char* key)
{
	FILE *fpfile;
	int flag=2,fpos,c,count=0;
	char fbuffer[URL_LEN],buffer[URL_LEN],url[URL_LEN];

	fpfile = fopen(DATA_FILE_NAME.c_str(),"r");
	if(fpfile == NULL){
		return -1;
	}	
	
	while( (c=fgetc(fpfile)) != EOF ){
		int b;

		if(c=='<'){
			flag=1;
		}
		if(c=='>'){
			flag=2;
		}

		// for Chinese words
		b=key[0]<0 ? 256+key[0]: key[0];

		if(b!=c) continue;

		fpos = ftell(fpfile);
		memset(fbuffer, 0, URL_LEN);
		fread(fbuffer,1,strlen(key)-1,fpfile);
		if( memcmp(key+1,fbuffer,strlen(key)-1) !=0 ){
			fseek(fpfile,fpos,SEEK_SET);
			continue;
		}

		if(flag==1) continue;		// <,in <>

		fseek(fpfile,-25,SEEK_CUR);
		memset(buffer, 0, URL_LEN);
		fread(buffer,1,50,fpfile);
		fseek(fpfile,-25,SEEK_CUR);

		//make sure not to meet the end of file here
		for(;;){
			if(feof(fpfile)) return -1;

			if(fgetc(fpfile)!=1) continue;

			if(feof(fpfile)) return -1;

			if(fgetc(fpfile)!=1) continue;

			if(feof(fpfile)) return -1;

			if(fgetc(fpfile)!=1) continue;

			if(feof(fpfile)) return -1;

			if(fgetc(fpfile)!=91) continue;

			int i=0;
			int charVal;
			char* charPos;

			memset(url, 0, URL_LEN);
			while( (charVal = fgetc(fpfile)) ){
				if(charVal==93) break;

				url[i]=charVal;
				i++;
			}
			url[i]=0;
			printf("###[%s] contains the key.\n",url);
			printf("......");

			i=0;
			while(buffer[i]==0) i++;

			charPos = buffer+i;
			charPos[49] = '\0';
			printf("%s",charPos);
			printf("......\n");

			count++;
			break;
		}
		cout << "What to do next?(c:continue/q:quit/s:"
			<< "show the correspanding web page)." << endl;
		for(;;){
			string c;
			for(;;){
				cin >> c;
				if( c == "c" || c == "q" || c == "s" ) break;
			}
			
			if( c == "c" ) break;
			if( c == "q" ) {
				fclose(fpfile);
				return count;
			}
			if( c == "s" ){
				char *sContentBuf=NULL;

				if( FindUrl(url, &sContentBuf) == -1 ){
					printf("Haven't found the url.\n");
				}else{
					for(unsigned int i=512; (i<strlen(sContentBuf))&&(i<600); i++){
						if((sContentBuf[i] == '\n') || (sContentBuf[i] == '\r')){
							sContentBuf[i] = '\0';
							break;
						}
						if(i == 599){
							sContentBuf[i] = '\0';
						}
					}

					cout << endl << sContentBuf << endl << "......" << endl << endl;

					if( sContentBuf) free(sContentBuf);
				}

				cout << "What to do next?(c:continue/q:quit/s:"
					<< "show the correspanding web page)." << endl;

			}	
		}
	}

	fclose(fpfile);
	return count;
}