source.c

朴素贝叶斯c代码

#include "source.h"

Source::Source( char delim[], char *fn)
{
	if(fn== NULL)
	{
		fname = new char[1024];
//		printf("Please enter training file name:");
		scanf("%s", fname);
#ifdef BDG
		cout << fname <<endl;
#endif
	}
	else {
		fname = new char[strlen(fn)+1];
		strcpy(fname, fn);
	}

	handle = fopen(fname, "r");
	ERROR2(handle, "File %s cannot be opened!", fname);
	strcpy(delimiter, delim);
	n_Instances    = countInstances();
	n_Descriptions = countDescriptions();
	buffer = NULL;
}

void Source::init()
{
	fseek (handle, 0L, SEEK_SET);
}

Description Source::readDescription()
{
	if(buffer==NULL) {
		where = buffer = new char[MAXLEN];
		fgets(buffer, 2048, handle);
	}

	if(where == NULL) {
		fgets(buffer, 2048, handle);
		where = buffer;
	}
	if(where) {
		char *curr = where;
		where = strstr(where,delimiter);
		if(where) *where = 0;
//		cout << curr<<(where!=NULL)?(*where):"";
		Description val = atof(curr);
//		cout << val <<"\n";
		if(where) where++; // to next value
		return val;
	}
}

int Source::countInstances()
{
	n_Instances = 0;
	char tmp[MAXLEN];
	long int pos = ftell(handle);
	init();

	while(!feof(handle)) {
		if(fgets(tmp, MAXLEN, handle) != NULL)
			n_Instances++;
	}

	fseek (handle, pos, SEEK_SET);
//	n_Instances --;
//	cout << "Instance :"<<n_Instances;
	return n_Instances;
}

int Source::countDescriptions()
{
	n_Descriptions = 0;
	char tmp[MAXLEN];
	long int pos = ftell(handle);
	init();

	if(fgets(tmp, MAXLEN, handle) != NULL) { 
		n_Descriptions++;

		char *pstr = tmp;
    while((pstr = strstr(pstr, delimiter)) != NULL)
		{
			pstr++;
			n_Descriptions++;
		}
	}

	fseek (handle, pos, SEEK_SET);

#ifdef DBG2
	cout << "Source::countDescriptions: "<<n_Descriptions<<endl;
#endif
	return n_Descriptions;
}

Description *Source::readOneInstance()
{
	char tmp[MAXLEN];

	Description *oneInst;
	oneInst = new Description [n_Descriptions+1];

	for(int i=0; i<n_Descriptions; i++)
	{
		oneInst[i] = readDescription();
#ifdef DBG2
		cout << oneInst[i] <<",";
#endif

	}
#ifdef DBG2
	cout <<endl;
#endif

	return oneInst;
}