filemanager.cpp

主元分析PCA的C代码

#include "filemanager.h"
#include <io.h>
#include <FCNTL.H>
#include <memory.h>
#include <string.h>

FileManager::FileManager()
{
	this->content=NULL;
}

FileManager::~FileManager()
{
	if(this->content)
		delete []this->content;
}

/*****************************************************
从指定文件读取一行
*****************************************************/
void FileManager::ReadLine(FILE *fp, char *buf)
{
	if(fgets(buf,MAX_LINE_CHAR,fp)==NULL)return;
	for(int i=0;i<MAX_LINE_CHAR;i++)
	{
		if(buf[i]==0x0a || buf[i]==0x0d)
		{
			buf[i]='\0';
			break;
		}
	}
}

/*****************************************************
以空格或制表符为间隔解析一个数据
*****************************************************/
int FileManager::ParseData(int pos,const char *buf,char *szRes)
{
	int i=pos,j=0;

	while((buf[i]==32 || buf[i]==9 || buf[i]==',') && buf[i])i++;
	while(!(buf[i]==32 || buf[i]==9 || buf[i]==',') && buf[i])
	{
		szRes[j]=buf[i];
		i++;
		j++;
	}
	szRes[j]=0;
	return i;
}

/*****************************************************
读入文件的所有内容，文件超过指定尺寸
*****************************************************/
void FileManager::LoadFileContent(const char *file_name)
{
	int buf_len;
	int fh=_open(file_name, _O_RDWR);
	if(fh!=-1)
	{
		buf_len=_filelength(fh)+1;
		_close(fh);
	}
	else
		buf_len=MAX_CONTENT_CHAR;
	this->content=new char[buf_len];
	memset(this->content,0,buf_len);

	FILE *fp=fopen(file_name,"r");
	if(fp==NULL)
		return;
	fread(this->content, sizeof(char), buf_len-1,fp);
	fclose(fp);
}

/*****************************************************
释放载入的文件内容
*****************************************************/
void FileManager::ReleaseFileContent()
{
	if(this->content)
		delete []this->content;
	this->content=NULL;
}

/*****************************************************
解析读取设置内容的一个项
*****************************************************/
char *FileManager::ParseItemValue(const char *ItemName)
{
	if(!this->content || !ItemName)return NULL;
	
	char *str=strstr(this->content,ItemName);
	int i;
	char *res;

	if(!str)return NULL;
	str=strstr(str,"=");
	for(i=1; str[i]; i++)
	{
		if(str[i]==0x0a)
		{
			i--;
			break;
		}
	}
	res=new char[i+1];
	memcpy(res,str+1,sizeof(char)*i);
	res[i]=0;
	return res;
}

/*****************************************************
读入训练或者测试数据
*****************************************************/
/*double *FileManager::ReadData(int number_of_inputs,int &number_of_data_points,char *file_name)
{
	char buf[MAX_LINE_CHAR];
	char str[100];
	FILE *fp = fopen(file_name,"r");
	double *res;
	int i,j,p;
	int nPos;
	
	this->ReadLine(fp,buf);
	number_of_data_points = atoi(buf);
	res = new double[number_of_data_points*(number_of_inputs+1)];
	for(i = 0, p = 0; i < number_of_data_points; i++)
	{
		this->ReadLine(fp,buf);
		nPos=0;
		for(j=0;j<number_of_inputs+1;j++,p++)
		{
			nPos=this->ParseData(nPos,buf,str);
			res[p]=atof(str);
		}
	}

	fclose(fp);
	return res;
}*/

double *FileManager::ReadData(int number_of_inputs,int &number_of_data_points, char *file_name)
{
	int i,j;
	FILE *fp;	      
	char buf[MAX_LINE_CHAR];
	double *res=NULL;
	fp = fopen(file_name, "r");
    
	this->ReadLine(fp,buf);
	number_of_data_points = atoi(buf);
	res = new double[number_of_data_points*(number_of_inputs+1)];
	//res=(double *)malloc(number_of_data_points*(number_of_inputs+1)*sizeof(double));
	if(!res)
		return NULL;
	memset(res, 0, number_of_data_points*(number_of_inputs+1)*sizeof(double));
	for(i=0; i<number_of_data_points;i++)
	{
		for(j=0; j<number_of_inputs+1;j++)
		{
			fscanf(fp, "%lf", &res[i*(number_of_inputs+1)+j]);
		}
	}
	fclose(fp);
	return res;
}

/*****************************************************
在指定文件中追加入一行
*****************************************************/
void FileManager::WriteFileLine(const char *str,char *file_name)
{
	FILE *fp=fopen(file_name,"a");
	if(!fp)return;
	fprintf(fp,"%s\n",str);
	fclose(fp);
}

/*****************************************************
读取交叉验证训练数据
*****************************************************/
double *FileManager::ReadCrossTrnData(char *dataset_name, int set_id, int number_crossset, int number_of_inputs, int &number_of_data_points)
{
	char str[255];
	int i,j,k,ch;
	int *count=(int *)malloc(sizeof(int)*number_crossset);
	int number_sample=0;
	
	for(i=0;i<number_crossset;i++)
	{
		if(i==set_id)continue;
		sprintf(str,"./dat/%s/%s_%d.txt",dataset_name,dataset_name,i);
		FILE *fp=fopen(str,"r");
		if(!fp)
		{
			free(count);
			return NULL;
		}
		count[i]=0;
		while((ch=fgetc(fp))!=EOF)
			if(ch=='\n')
				count[i]++;
		fclose(fp);
		number_sample+=count[i];
	}
	number_of_data_points=number_sample;
	
	//AllocData(trn_data);
	//double *res=(double *)malloc(number_sample*(number_of_inputs+1)*sizeof(double));
	double *res=new double[number_sample*(number_of_inputs+1)*sizeof(double)];
	double *p=res;
	for(i=0;i<number_crossset;i++)
	{
		if(i==set_id)continue;
		sprintf(str,"./dat/%s/%s_%d.txt",dataset_name,dataset_name,i);
		FILE *fp=fopen(str,"r");
		for(j=0;j<count[i];j++)
		{
			for(k=0;k<number_of_inputs+1;k++, p++)
				fscanf(fp,"%lf", p);
		}
		fclose(fp);
	}
	free(count);
	return res;
}

/*****************************************************
读取交叉验证测试数据
*****************************************************/
double *FileManager::ReadCrossTstData(char *dataset_name, int set_id, int number_crossset, int number_of_inputs, int &number_of_data_points)
{
	char str[255];
	int j,k,ch;
	int number_sample=0;
	
	sprintf(str,"./dat/%s/%s_%d.txt",dataset_name,dataset_name,set_id);
	FILE *fp=fopen(str,"r");
	while((ch=fgetc(fp))!=EOF)
		if(ch=='\n')
			number_sample++;
	fclose(fp);
	number_of_data_points=number_sample;
	
	//double *res=(double *)malloc(number_sample*(number_of_inputs+1)*sizeof(double));
	double *res=new double[number_sample*(number_of_inputs+1)*sizeof(double)];
	double *p=res;
	sprintf(str,"./dat/%s/%s_%d.txt",dataset_name,dataset_name,set_id);
	fp=fopen(str,"r");
	for(j=0;j<number_sample;j++)
	{
		for(k=0;k<number_of_inputs+1;k++, p++)
			fscanf(fp,"%lf", p);
	}
	fclose(fp);
	return res;
}