publib.h

求2个或多个文本间的公共子串

#include <unistd.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <errno.h>
#include <fcntl.h>
#include <malloc.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#ifndef NULL
#define NULL (void *)0
#endif

#ifndef smch_t
#define smch_t unsigned short 
#endif

#ifndef _PUBLIB_HEADER_
#define _PUBLIB_HEADER_

/*Define a structure containing file useful infomation
*********/
typedef struct Fileinfo{
	char pPath[256];  /*file name*/
	/* FILE * pFile;
        *******/       
	char * pStr;        /*file content pointer*/
	size_t nLen;	    /*file length*/	
}Fileinfo;


/*Get file info through the file name
intput: the non-NULL string pStr is the file name
output: value-result variable pInfo points to the file info struct

return: 0 indicate successful, else error happened
**********/
int GetFileInfo(char * pStr, Fileinfo * pInfo);


/*Put every character to a memory of fixed size
 * Here we assume 2 bytes is enough for any encode method
 * In the process, we can construct the lookup table for each string;
 * we assume prime number of 1333 is a good candidate for the number of common characters 
 *
 * input:
 * 	pOrgStr: the string may contain differnt encode characters.
 * 	orgLen: the length
 *
 * output:
 * 	**ppTranStr: transfered string with each elements consumming the same memory 
 * 	pTranLen: value-result variable for the length of the transfered string 
 * 	lookup[1333]: lookup table for 1333 character
 * 	
 * 	return: 0 is ok; -1 is wrong
**********/
int TransferString(char * pOrgStr, const size_t orgLen, smch_t ** ppTranStr, size_t * pTranLen);


typedef struct CharIndex{
	smch_t smch;
	size_t * location;
	size_t memLen;
	size_t memUse;
	struct CharIndex * pNext;
}CharIndex;

/* Consruct a index for a string. Then it is easy to locate each kinds of character
 * Input:
 * 	pConStr: point to element
 * 	nLen: size of element
 * 	nIdxLen: Index length
 * output:
 * 	*ppIndexArray: to record the index information
 * return: 0 is ok, else return -1
 * 
 * I think it is better to let nIdxLen to be a prime number between 1000 to 6000
**********/
int IndexCharForStr(smch_t * pConStr, const size_t nStrLen, CharIndex ** ppIndexArray, const size_t nIdxLen);

/* Get the positions for a character
 * Input:
 * 	ppIndexArray: the character index array
 *	nIdxLen: the array length
 * output:
 * 	ppListPos: the list beginning position
 * 	pListLen: list length
 * return:
 * 	0 is ok, -1 indicates failure
 * ********************/
int GetCharPosList(const smch_t smch, CharIndex ** ppIndexArray, const size_t nIdxLen, size_t ** ppListPos, size_t * pListLen);

#endif