10 int ksprintf(kstring_t *s, const char *fmt, ...)
15 l = vsnprintf(s->s + s->l, s->m - s->l, fmt, ap); // This line does not work with glibc 2.0. See `man snprintf'.
17 if (l + 1 > s->m - s->l) {
20 s->s = (char*)realloc(s->s, s->m);
22 l = vsnprintf(s->s + s->l, s->m - s->l, fmt, ap);
29 char *kstrtok(const char *str, const char *sep, ks_tokaux_t *aux)
31 const char *p, *start;
32 if (sep) { // set up the table
33 if (str == 0 && (aux->tab[0]&1)) return 0; // no need to set up if we have finished
37 aux->tab[0] = aux->tab[1] = aux->tab[2] = aux->tab[3] = 0;
38 for (p = sep; *p; ++p) aux->tab[*p>>6] |= 1ull<<(*p&0x3f);
39 } else aux->sep = sep[0];
41 if (aux->finished) return 0;
42 else if (str) aux->p = str - 1, aux->finished = 0;
44 for (p = start = aux->p + 1; *p; ++p)
45 if (aux->tab[*p>>6]>>(*p&0x3f)&1) break;
47 for (p = start = aux->p + 1; *p; ++p)
48 if (*p == aux->sep) break;
50 aux->p = p; // end of token
51 if (*p == 0) aux->finished = 1; // no more tokens
55 // s MUST BE a null terminated string; l = strlen(s)
56 int ksplit_core(char *s, int delimiter, int *_max, int **_offsets)
58 int i, n, max, last_char, last_start, *offsets, l;
59 n = 0; max = *_max; offsets = *_offsets;
62 #define __ksplit_aux do { \
66 max = max? max<<1 : 2; \
67 offsets = (int*)realloc(offsets, sizeof(int) * max); \
69 offsets[n++] = last_start; \
73 for (i = 0, last_char = last_start = 0; i <= l; ++i) {
75 if (isspace(s[i]) || s[i] == 0) {
76 if (isgraph(last_char)) __ksplit_aux; // the end of a field
78 if (isspace(last_char) || last_char == 0) last_start = i;
81 if (s[i] == delimiter || s[i] == 0) {
82 if (last_char != 0 && last_char != delimiter) __ksplit_aux; // the end of a field
84 if (last_char == delimiter || last_char == 0) last_start = i;
89 *_max = max; *_offsets = offsets;
93 /**********************
94 * Boyer-Moore search *
95 **********************/
97 typedef unsigned char ubyte_t;
99 // reference: http://www-igm.univ-mlv.fr/~lecroq/string/node14.html
100 static int *ksBM_prep(const ubyte_t *pat, int m)
102 int i, *suff, *prep, *bmGs, *bmBc;
103 prep = calloc(m + 256, sizeof(int));
104 bmGs = prep; bmBc = prep + m;
106 for (i = 0; i < 256; ++i) bmBc[i] = m;
107 for (i = 0; i < m - 1; ++i) bmBc[pat[i]] = m - i - 1;
109 suff = calloc(m, sizeof(int));
114 for (i = m - 2; i >= 0; --i) {
115 if (i > g && suff[i + m - 1 - f] < i - g)
116 suff[i] = suff[i + m - 1 - f];
120 while (g >= 0 && pat[g] == pat[g + m - 1 - f]) --g;
127 for (i = 0; i < m; ++i) bmGs[i] = m;
128 for (i = m - 1; i >= 0; --i)
129 if (suff[i] == i + 1)
130 for (; j < m - 1 - i; ++j)
133 for (i = 0; i <= m - 2; ++i)
134 bmGs[m - 1 - suff[i]] = m - 1 - i;
140 void *kmemmem(const void *_str, int n, const void *_pat, int m, int **_prep)
142 int i, j, *prep = 0, *bmGs, *bmBc;
143 const ubyte_t *str, *pat;
144 str = (const ubyte_t*)_str; pat = (const ubyte_t*)_pat;
145 prep = (_prep == 0 || *_prep == 0)? ksBM_prep(pat, m) : *_prep;
146 if (_prep && *_prep == 0) *_prep = prep;
147 bmGs = prep; bmBc = prep + m;
150 for (i = m - 1; i >= 0 && pat[i] == str[i+j]; --i);
152 int max = bmBc[str[i+j]] - m + 1 + i;
153 if (max < bmGs[i]) max = bmGs[i];
155 } else return (void*)(str + j);
157 if (_prep == 0) free(prep);
161 char *kstrstr(const char *str, const char *pat, int **_prep)
163 return (char*)kmemmem(str, strlen(str), pat, strlen(pat), _prep);
166 char *kstrnstr(const char *str, const char *pat, int n, int **_prep)
168 return (char*)kmemmem(str, n, pat, strlen(pat), _prep);
171 /***********************
172 * The main() function *
173 ***********************/
183 s = (kstring_t*)calloc(1, sizeof(kstring_t));
185 ksprintf(s, " abcdefg: %d ", 100);
186 printf("'%s'\n", s->s);
188 fields = ksplit(s, 0, &n);
189 for (i = 0; i < n; ++i)
190 printf("field[%d] = '%s'\n", i, s->s + fields[i]);
193 for (p = kstrtok("ab:cde:fg/hij::k", ":/", &aux); p; p = kstrtok(0, 0, &aux)) {
194 kputsn(p, aux.p - p, s);
199 free(s->s); free(s); free(fields);
202 static char *str = "abcdefgcdgcagtcakcdcd";
203 static char *pat = "cd";
206 while ((ret = kstrstr(s, pat, &prep)) != 0) {
207 printf("match: %s\n", ret);