1
0
Fork 0
forked from fun/fun

Refactored the PCRE2 extension plus adding some regex examples using this extension. (0.41.7)

This commit is contained in:
Johannes Findeisen 2026-05-07 21:54:05 +02:00
commit be590414af
17 changed files with 637 additions and 404 deletions

View file

@ -60,68 +60,7 @@ case OP_PCRE2_FINDALL: {
push_value(vm, make_array_from_values(NULL, 0));
break;
}
#ifndef PCRE2_CODE_UNIT_WIDTH
#define PCRE2_CODE_UNIT_WIDTH 8
#endif
#include <pcre2.h>
int errorcode;
PCRE2_SIZE erroff;
uint32_t opt = 0;
if (flags & 1) opt |= PCRE2_CASELESS; /* I */
if (flags & 2) opt |= PCRE2_MULTILINE; /* M */
if (flags & 4) opt |= PCRE2_DOTALL; /* S */
if (flags & 8) opt |= PCRE2_UTF; /* U */
if (flags & 16) opt |= PCRE2_EXTENDED; /* X */
pcre2_code *re = pcre2_compile((PCRE2_SPTR)pattern, PCRE2_ZERO_TERMINATED, opt, &errorcode, &erroff, NULL);
if (!re) {
free(pattern);
free(subject);
push_value(vm, make_array_from_values(NULL, 0));
break;
}
pcre2_match_data *mdata = pcre2_match_data_create_from_pattern(re, NULL);
Value out = make_array_from_values(NULL, 0);
size_t subj_len = strlen(subject);
size_t start_off = 0;
int gcount = 0;
while (1) {
int rc = pcre2_match(re, (PCRE2_SPTR)subject, (PCRE2_SIZE)subj_len, start_off, 0, mdata, NULL);
if (rc <= 0) break;
PCRE2_SIZE *ov = pcre2_get_ovector_pointer(mdata);
int s0 = (int)ov[0];
int e0 = (int)ov[1];
/* result map for this match */
Value res = make_map_empty();
char *full = string_substr(subject, s0, e0 - s0);
(void)map_set(&res, "full", make_string(full ? full : ""));
if (full) free(full);
(void)map_set(&res, "start", make_int(s0));
(void)map_set(&res, "end", make_int(e0));
Value groups = make_array_from_values(NULL, 0);
for (int i = 1; i < rc; ++i) {
int s = (int)ov[2 * i];
int e = (int)ov[2 * i + 1];
char *gstr = (s >= 0 && e >= s) ? string_substr(subject, s, e - s) : NULL;
Value gv = make_string(gstr ? gstr : "");
if (gstr) free(gstr);
(void)array_push(&groups, gv);
}
(void)map_set(&res, "groups", groups);
(void)array_push(&out, res);
/* advance start offset; guard against empty match */
if (e0 == s0) {
if ((size_t)e0 < subj_len) {
start_off = e0 + 1;
} else {
break;
}
} else {
start_off = e0;
}
gcount = rc;
}
pcre2_match_data_free(mdata);
pcre2_code_free(re);
Value out = fun_pcre2_findall(pattern, subject, flags);
free(pattern);
free(subject);
push_value(vm, out);

View file

@ -58,58 +58,7 @@ case OP_PCRE2_MATCH: {
push_value(vm, make_nil());
break;
}
#ifndef PCRE2_CODE_UNIT_WIDTH
#define PCRE2_CODE_UNIT_WIDTH 8
#endif
#include <pcre2.h>
int errorcode;
PCRE2_SIZE erroff;
uint32_t opt = 0;
if (flags & 1) opt |= PCRE2_CASELESS; /* I */
if (flags & 2) opt |= PCRE2_MULTILINE; /* M */
if (flags & 4) opt |= PCRE2_DOTALL; /* S */
if (flags & 8) opt |= PCRE2_UTF; /* U */
if (flags & 16) opt |= PCRE2_EXTENDED; /* X */
pcre2_code *re = pcre2_compile((PCRE2_SPTR)pattern, PCRE2_ZERO_TERMINATED, opt, &errorcode, &erroff, NULL);
if (!re) {
free(pattern);
free(subject);
push_value(vm, make_nil());
break;
}
pcre2_match_data *mdata = pcre2_match_data_create_from_pattern(re, NULL);
int rc = pcre2_match(re, (PCRE2_SPTR)subject, (PCRE2_SIZE)strlen(subject), 0, 0, mdata, NULL);
if (rc <= 0) {
pcre2_match_data_free(mdata);
pcre2_code_free(re);
free(pattern);
free(subject);
push_value(vm, make_nil());
break;
}
PCRE2_SIZE *ov = pcre2_get_ovector_pointer(mdata);
/* Build result map */
Value res = make_map_empty();
int start0 = (int)ov[0];
int end0 = (int)ov[1];
char *full = string_substr(subject, start0, end0 - start0);
(void)map_set(&res, "full", make_string(full ? full : ""));
if (full) free(full);
(void)map_set(&res, "start", make_int(start0));
(void)map_set(&res, "end", make_int(end0));
/* groups array (excluding group 0) */
Value groups = make_array_from_values(NULL, 0);
for (int i = 1; i < rc; ++i) {
int s = (int)ov[2 * i];
int e = (int)ov[2 * i + 1];
char *gstr = (s >= 0 && e >= s) ? string_substr(subject, s, e - s) : NULL;
Value gv = make_string(gstr ? gstr : "");
if (gstr) free(gstr);
(void)array_push(&groups, gv);
}
(void)map_set(&res, "groups", groups);
pcre2_match_data_free(mdata);
pcre2_code_free(re);
Value res = fun_pcre2_match(pattern, subject, flags);
free(pattern);
free(subject);
push_value(vm, res);

View file

@ -53,32 +53,10 @@ case OP_PCRE2_TEST: {
push_value(vm, make_int(0));
break;
}
#ifndef PCRE2_CODE_UNIT_WIDTH
#define PCRE2_CODE_UNIT_WIDTH 8
#endif
#include <pcre2.h>
int errorcode;
PCRE2_SIZE erroff;
uint32_t opt = 0;
if (flags & 1) opt |= PCRE2_CASELESS; /* I */
if (flags & 2) opt |= PCRE2_MULTILINE; /* M */
if (flags & 4) opt |= PCRE2_DOTALL; /* S */
if (flags & 8) opt |= PCRE2_UTF; /* U */
if (flags & 16) opt |= PCRE2_EXTENDED; /* X */
pcre2_code *re = pcre2_compile((PCRE2_SPTR)pattern, PCRE2_ZERO_TERMINATED, opt, &errorcode, &erroff, NULL);
if (!re) {
free(pattern);
free(subject);
push_value(vm, make_int(0));
break;
}
pcre2_match_data *mdata = pcre2_match_data_create_from_pattern(re, NULL);
int rc = pcre2_match(re, (PCRE2_SPTR)subject, (PCRE2_SIZE)strlen(subject), 0, 0, mdata, NULL);
pcre2_match_data_free(mdata);
pcre2_code_free(re);
int rc = fun_pcre2_test(pattern, subject, flags);
free(pattern);
free(subject);
push_value(vm, make_int(rc >= 0 ? 1 : 0));
push_value(vm, make_int(rc));
#else
/* pop args and return 0 when PCRE2 disabled */
Value a = pop_value(vm);