lib.c

C语言的朴素贝叶斯分类器代码

/* THIS FILE MAINLY CONTAINS THE CODE FOR READING DATA FROM AN EXTERNAL DATA 
                      FILE AND GENERATING RANDOM NUMBERS                     */


#include "header.h"


/*----------------------------- Global variables ----------------------------*/
char **DataSet;                                            /* dataset buffer */
long ROW, COL;                      /* Number of row and column in a dataset */

/*-------------------------Function references-------------------------------*/
long GetNumberofAttributes(char *);
long GetNumberofTuples(char *);
void AllocateDataSpace( long, long );
void GetDataIntoBuffer(char *);
void DisplayDataBuffer();
/*---------------------------------------------------------------------------*/
/*--------------------Reads data from the file-------------------------------*/
/*---------------------------------------------------------------------------*/
ReadData(char *filename){
FILE *fp;

                                                  /* checking file existance */

  if ((fp = fopen(filename, "r")) == NULL){
  	printf ("\n %s not found in this path", filename);
	exit(0);
  }
  fclose( fp );

                       /* get number of tuples and attrib  from the data file*/

  COL = GetNumberofAttributes( filename );
  ROW = GetNumberofTuples( filename );

                   /* allocate memory for the data set in the DataSet buffer */

  AllocateDataSpace( ROW, COL );

                           /* Collect data into DataSet buffer from the file */

  GetDataIntoBuffer( filename );
}
/*---------------------------------------------------------------------------*/
long GetNumberofTuples(char *filename){
FILE *pfp;
char command[STRBUF];

  sprintf(command, "cat %s | wc -l", filename);    /* use wc system command  */
  pfp = popen(command, "r");	       	           /* open a pipe to read    */
  fgets(command, STRBUF, pfp);		           /* get the number of line */
  pclose (pfp);
  return atol(command);
}
/*---------------------------------------------------------------------------*/
long GetNumberofAttributes(char *filename){
long i, field;
FILE *fp;
char buffer[STRBUF];

  fp = fopen(filename, "r");
  fgets( buffer, STRBUF, fp);
  fclose (fp);
  field = 0;
  for ( i = strlen(buffer); i; i--){
	if ( SEPARATOR(buffer[i]) )  
		field ++;
  }
  return field + 1;                    /* number of field is = separator + 1 */
}
/*---------------------------------------------------------------------------*/
void AllocateDataSpace(long row, long col){

 DataSet = (char**)malloc(sizeof(char*) * row);
 SUCCESS( DataSet );
 while(row){
        DataSet[ row - 1] = (char *)malloc(sizeof(char) * col);
	SUCCESS( DataSet[ row - 1 ] );
        row --;
  }
}
/*---------------------------------------------------------------------------*/
void GetDataIntoBuffer(char *database){
FILE *fp;
long tuples, field, i, j;
char buffer[STRBUF], tempBuffer[STRBUF];


  fp = fopen (database, "r");

  for (tuples = 0; tuples < ROW; tuples ++){
	fgets( buffer, STRBUF, fp);                     /* collect one tuple */
	i = 0;
	for ( field = 0; field < COL; field ++ ){    /* collect attrib value */
	        strcpy(tempBuffer, buffer);
		for ( ; i < strlen (buffer) && ! SEPARATOR(buffer[i]); i++);
		tempBuffer[i++] = '\0';
		for (j = strlen(tempBuffer)-1; j >= 0 &&
			                     ! SEPARATOR(tempBuffer[j]); j--);

		DataSet[tuples][field] = (char)atoi(tempBuffer + j + 1); 
	}
  }
  fclose (fp);
}
/*---------------------------------------------------------------------------*/
void FreeDataBuffer(){

  free( DataSet );
}
/*---------------------------------------------------------------------------*/
void DisplayDataBuffer(){
int i = ROW;
int j = COL;

  for(i = 0 ; i < ROW; i ++){
	printf("%d", DataSet[i][0]);
	for(j = 1; j < COL; j ++)
		printf(",%d", DataSet[i][j]);
	printf("\n");
  }
}
/*---------------------------------------------------------------------------*/