一个文本搜索引擎的实现

文本搜索引擎，c++实现，南大数据结构期末大作业

#include “SearchEngine.h“
#include “iostream“

using namespace std;

//extern linkTable *lkhead;
double *matrix_mul（SMatrix *sm double *pr）;
bool matrix_equal（double * double *）;

static int sm_sum = 0;
double average = 1.0 / FILE_SUM;

SMatrix *create_sm（linkNodeTable *node） {
	if （node == NULL）
		return NULL;
	int count = 0;
	linkTable *pnode = node;
	while （pnode） {
		sm_sum += pnode->indegree;
		pnode = pnode->next;
	}
	SMatrix *sm = new SMatrix[sm_sum];
	int outdegree[FILE_SUM];
	for （int i = 0; i < FILE_SUM; ++i）
		outdegree[i] = 0;
	while （node） {
		for （int i = 0; i < node->indegree; ++i） {
			sm[count].row = node->index;
			sm[count].col = node->link_text[i];
			outdegree[sm[count].col]++;
			count++;
		}
		node = node->next;
	}
	for （int i = 0; i < sm_sum; ++i）
		sm[i].value = 1 / outdegree[sm[i].col];
	return sm;
}

double *page_rank（linkTable *lkh） {
	SMatrix *smatrix = create_sm（lkh）;
	double *PRresult = new double[FILE_SUM];
	for （int i = 0; i < FILE_SUM; ++i）
		PRresult[i] = average;
	double *newPRresult = matrix_mul（smatrix PRresult）;
	while （!matrix_equal（PRresult newPRresult）） {
		double *tmp = newPRresult;
		newPRresult = NULL;
		delete tmp;
		newPRresult = matrix_mul（smatrix PRresult）;
	}
	delete newPRresult;
	return PRresult;
}

double *matrix_mul（SMatrix *sm double *pr） {
	double *result = new double[FILE_SUM];
	for （int i = 0; i < FILE_SUM; ++i）
		result[i] = 0;
	for （int i = 0; i < sm_sum; ++i）
		result[sm[i].row] = result[sm[i].row] + 0.85 * sm[i].value * pr[sm[i].col] + 0.15 * average;
	return result;
}

bool matrix_equal（double *mdest double *msrc） {
	bool equal = true;
	for （int i = 0; i < FILE_SUM; ++i）
		if （mdest[i] != msrc[i]） {
			equal = false;
			mdest[i] = msrc[i];
		}
	return equal;
}

 属性            大小     日期    时间   名称
----------- ---------  ---------- -----  ----

     文件     116224  2016-01-10 04:12  SearchEngine\Debug\SearchEngine.exe

     文件     887024  2016-01-10 04:12  SearchEngine\Debug\SearchEngine.ilk

     文件    1249280  2016-01-10 04:12  SearchEngine\Debug\SearchEngine.pdb

     文件     112316  2016-01-10 04:01  SearchEngine\SearchEngine\Debug\PageRank.obj

     文件     244515  2016-01-10 04:08  SearchEngine\SearchEngine\Debug\ReadDoc.obj

     文件     147224  2016-01-10 04:12  SearchEngine\SearchEngine\Debug\RunSearch.obj

     文件       1619  2016-01-10 04:12  SearchEngine\SearchEngine\Debug\SearchEngine.log

     文件     106508  2016-01-10 04:01  SearchEngine\SearchEngine\Debug\SearchEngine.obj

     文件       3974  2016-01-10 04:12  SearchEngine\SearchEngine\Debug\SearchEngine.tlog\cl.command.1.tlog

     文件      22132  2016-01-10 04:12  SearchEngine\SearchEngine\Debug\SearchEngine.tlog\CL.read.1.tlog

     文件       4572  2016-01-10 04:12  SearchEngine\SearchEngine\Debug\SearchEngine.tlog\CL.write.1.tlog

     文件       1986  2016-01-10 04:12  SearchEngine\SearchEngine\Debug\SearchEngine.tlog\link.command.1.tlog

     文件       3606  2016-01-10 04:12  SearchEngine\SearchEngine\Debug\SearchEngine.tlog\link.read.1.tlog

     文件       1090  2016-01-10 04:12  SearchEngine\SearchEngine\Debug\SearchEngine.tlog\link.write.1.tlog

     文件        185  2016-01-10 04:12  SearchEngine\SearchEngine\Debug\SearchEngine.tlog\SearchEngine.lastbuildstate

     文件     125037  2016-01-10 04:01  SearchEngine\SearchEngine\Debug\SearchTreeFunc.obj

     文件     125159  2016-01-10 04:07  SearchEngine\SearchEngine\Debug\SEMain.obj

     文件     379904  2016-01-10 04:12  SearchEngine\SearchEngine\Debug\vc120.idb

     文件     413696  2016-01-10 04:12  SearchEngine\SearchEngine\Debug\vc120.pdb

     文件   22408777  2015-12-14 15:44  SearchEngine\SearchEngine\documents.txt

     文件     409930  2015-12-14 15:47  SearchEngine\SearchEngine\link.txt

     文件       1829  2016-01-10 03:50  SearchEngine\SearchEngine\PageRank.cpp

     文件       3247  2016-01-10 04:08  SearchEngine\SearchEngine\ReadDoc.cpp

     文件       7346  2016-01-10 04:12  SearchEngine\SearchEngine\RunSearch.cpp

     文件        968  2016-01-10 03:31  SearchEngine\SearchEngine\SearchEngine.h

     文件       4309  2016-01-10 03:21  SearchEngine\SearchEngine\SearchEngine.vcxproj

     文件       1415  2016-01-10 03:21  SearchEngine\SearchEngine\SearchEngine.vcxproj.filters

     文件       1596  2016-01-09 14:45  SearchEngine\SearchEngine\SearchTreeFunc.cpp

     文件        911  2016-01-10 04:05  SearchEngine\SearchEngine\SEMain.cpp

     文件    8192000  2016-01-10 04:18  SearchEngine\SearchEngine.sdf

............此处省略10个文件信息