unsigned char *BM(const unsigned char *s, int slen, const unsigned char *wanted, int wlen)
{
int i;
int j = 0;
int len = slen - wlen;
int bmGs[BM_XSIZE];
int bmBc[BM_ASIZE];
if (NULL == wanted || 0 > wlen || NULL == s || slen < wlen) return NULL;
if (-1 == preBmGs(wanted, wlen, bmGs) || -1 == preBmBc(wanted, wlen, bmBc)) return NULL;
while (j <= len) {
for (i = wlen - 1; i >= 0 && wanted[i] == s[i+j]; i--);
if (i < 0) {
return (unsigned char *)(s + j);
} else {
#if BM_VERSION == 1
j += MAX(bmGs[i], i - bmBc[s[i+j]]);
#else
j += MAX(bmGs[i], bmBc[s[i+j]] - wlen + 1 + i);
#endif
}
}
return NULL;
}
int
bm_init(BM *bmp,
unsigned char *x,
int m,
int icase)
{
int i;
memset(bmp, 0, sizeof(bmp));
bmp->icase = icase;
bmp->bmGs = (int *) calloc(sizeof(int), m);
if (bmp->bmGs == NULL)
return -1;
bmp->saved_m = m;
bmp->saved_x = (unsigned char *) malloc(m);
if (bmp->saved_x == NULL)
return -2;
for (i = 0; i < m; i++)
bmp->saved_x[i] = icase ? tolower(x[i]) : x[i];
/* Preprocessing */
if (preBmGs(bmp->saved_x, m, bmp->bmGs) < 0)
return -3;
preBmBc((unsigned char *) bmp->saved_x, m, bmp->bmBc);
return 0;
}
//T为文本串，P为模式串，成功返回找到次数，失败返回0
int bmsearch(char *T, char *P){
int i, j, bmGs[PSIZE], bmBc[ASIZE];
//初始化坏字符数组和好前缀数组
preBmGs(P, bmGs);
preBmBc(P, bmBc);
int m = strlen(P);
int tLen = strlen(T);
int ret = 0;
j = 0;
while(j <= tLen - m){ //j从前往后遍历文本串
for(i = m - 1; i >= 0; --i){ //i从后往前遍历模式串
cmp_count++;
if(T[j + i] != P[i]){
break;
}
}
if(i == -1){ //表示匹配成功
found[ret++] = j;
j += m;
}else{
j += (bmGs[i] > (bmBc[P[i]] - m + 1 + i) ? bmGs[i] : (bmBc[P[i]] - m + 1 + i));
}
}
return ret;
}
// x - pointer to pattern
// m - len(x)
// y - pointer to string to search
// n - len(y)
// modified to return location of first match, or -1
int
boyer_moore(
unsigned char *x,
int m,
unsigned char *y,
int n)
{
int i, j, bmBc[ASIZE];
int *bmGs = safe_malloc(m * sizeof(int));
/* Preprocessing */
preBmGs(x, m, bmGs);
preBmBc(x, m, bmBc);
/* Searching */
j = 0;
while (j <= n - m) {
for (i = m - 1; i >= 0 && x[i] == y[i + j]; --i) ;
if (i < 0) {
free(bmGs);
return j;
//j += bmGs[0]; // just returning the first match
}
else {
j += MAX(bmGs[i], bmBc[y[i + j]] - m + 1 + i);
}
}
free(bmGs);
return -1;
}
HB_ISIZ hb_strAtTBM( const char * needle, HB_ISIZ m, const char * haystack, HB_ISIZ n )
{
HB_ISIZ r = 0;
HB_ISIZ bcShift, j, shift, u, v, turboShift;
HB_ISIZ bmBc[ ASIZE ];
HB_ISIZ * bmGs;
bmGs = ( HB_ISIZ * ) hb_xgrab( m * sizeof( HB_ISIZ ) );
/* Preprocessing */
preBmGs( needle, m, bmGs );
preBmBc( needle, m, bmBc );
/* Searching */
j = u = 0;
shift = m;
while( j <= n - m )
{
HB_ISIZ i = m - 1;
while( i >= 0 && needle[ i ] == haystack[ i + j ] )
{
--i;
if( u != 0 && i == m - 1 - shift )
i -= u;
}
if( i < 0 )
{
r = j + 1;
break;
#if 0 /* To continue search */
shift = bmGs[ 0 ];
u = m - shift;
#endif
}
else
{
v = m - 1 - i;
turboShift = u - v;
bcShift = bmBc[ ( HB_UCHAR ) haystack[ i + j ] ] - m + 1 + i;
shift = HB_MAX( turboShift, bcShift );
shift = HB_MAX( shift, bmGs[ i ] );
if( shift == bmGs[ i ] )
u = HB_MIN( m - shift, v );
else
{
if( turboShift < bcShift )
shift = HB_MAX( shift, u + 1 );
u = 0;
}
}
j += shift;
}
hb_xfree( bmGs );
return r;
}
/**
* @brief string matching with Boyer-Moore algorithm.
* Output the offset begin with pattern, otherwise nothing output means not any
* pattern find in the search string.
* Note: if do not want to use malloc, can use stack memory like ASIZE make a
* int[] array but not a int* array, and when m >= XSIZE then use strstr(). This
* application in a project name "QDBM" database code.
*
* @param x pattern.
* @param m pattern length.
* @param y search string.
* @param n search string length.
*/
void string_matching_with_bm_algorithm(const char* x, int m,
const char* y, int n) {
int i = 0, j = 0, bmBc[ASIZE] = {0}, *bmGs = NULL, XSIZE = m+1;
if (!(bmGs = (int*) malloc(sizeof(int)*XSIZE))) {
fprintf(stderr, "malloc err!\n"); return;
}
preBmGs(x, m, bmGs, XSIZE); preBmBc(x, m, bmBc); // preprocessing
while (j <= n-m) { // searching
for (i = m-1; i >= 0 && x[i] == y[j+i]; i --) ;
if (i < 0) { OUTPUT(j); j += bmGs[0]; }
else { j += MAX(bmGs[i], bmBc[(int)(y[j+i])]-((m-1)-i)); }
}
free(bmGs);
}
void *
boyer_moore_init(
unsigned char *x,
int m)
{
boyer_moore_data_t *bm = safe_malloc(sizeof(boyer_moore_data_t));
bm->x = safe_malloc(m * sizeof(*x));
memcpy(bm->x, x, m * sizeof(*x));
bm->m = m;
bm->bmGs = safe_malloc(m * sizeof(int));
// Pre-process
preBmGs(x, m, bm->bmGs);
preBmBc(x, m, bm->bmBc);
return (void *) bm;
}
int search(unsigned char *x, int m, unsigned char *y, int n) {
int j, bmBc[SIGMA], qsBc[SIGMA], count;
/* Preprocessing */
BEGIN_PREPROCESSING
preBmBc(x, m, bmBc);
preQsBc(x, m, qsBc);
END_PREPROCESSING
count = 0;
/* Searching */
BEGIN_SEARCHING
j = 0;
while (j<= n - m) {
if (memcmp(x, y + j, m) == 0)
OUTPUT(j);
j += MAX(bmBc[y[j + m - 1]], qsBc[y[j + m]]);
}
END_SEARCHING
return count;
}
char* Ssmith2(char * textt,char *patt,int n, int m)//smith ╦сие
{
int j,bmBc[ASIZE],qsBc[ASIZE];
unsigned char * text,*pat;
text = (unsigned char*)textt;
pat = (unsigned char*)patt;
if(*pat == '\0') { OUTPUT(0);return textt;};
/* preprocessing */
preBmBc((char*)pat,m,bmBc);
preQsBc((char*)pat,m,qsBc);
/* searching */
j=0;
while (j<=n-m)
{
if (memcmp(pat,text+j,m)==0)
OUTPUT(j);
j+=MAX(bmBc[text[j+m-1]],qsBc[text[j+m]]);
}
SRET(j);
}