Change regex library to pcre2 for cross-platform compilation and better regex support.

include/ksr/regex.h

@@ -1,13 +1,14 @@
 #pragma once
 
 #include <jemalloc/jemalloc.h>
-#include <regex.h>
+#define PCRE2_CODE_UNIT_WIDTH 8
+#include <pcre2.h>
 
 /**
  * Regex structure.
  */
 typedef struct {
-	regex_t expr;
+	pcre2_code *expr;
 	unsigned nmatches;
 	char **matches;
 } ksregex;
@@ -16,17 +17,17 @@ typedef struct {
  * Create a ksregex from an expression.
  * @param expression - the expression to compile.
  * @param nmatches - the number of matches groups to get, 0 if you want none.
- * @param flags - regex flags.
+ * @param options - PCRE2 primary option bits.
  * @return - the created regex, NULL if an error happen.
  */
-ksregex* ksregex_new(const char *expression, unsigned nmatches, int flags);
+ksregex *ksregex_new(PCRE2_SPTR expression, unsigned nmatches, int options);
 /**
  * Create a ksregex from an expression with 0 match group.
  * @param expression - the expression to compile.
- * @param flags - regex flags.
+ * @param options - PCRE2 primary option bits.
  */
-static inline ksregex* ksregex_new_nogroup(const char *expression, int flags)
+static inline ksregex *ksregex_new_nogroup(PCRE2_SPTR expression, int options)
-{ return ksregex_new(expression, 0, flags); }
+{ return ksregex_new(expression, 0, options); }
 
 /**
  * Try to match the regex with a tested string.
@@ -59,5 +60,8 @@ static inline void ksregex_free(ksregex *regex)
 	// free matches groups.
 	_ksregex_free_matches(regex);
 
+	// free pcre expr.
+	pcre2_code_free(regex->expr);
+
 	free(regex); // free regex data.
 }

meson.build

@@ -5,6 +5,8 @@ c = meson.get_compiler('c')
 deps = [ dependency('jemalloc') ]
 # add pthread dependency
 deps += c.find_library('pthread')
+# add pcre2 dependency
+deps += c.find_library('pcre2-8')
 
 # includes
 include_dirs = include_directories([ 'include' ])

src/regex.c

@@ -1,35 +1,41 @@
 #include <ksr/regex.h>
-#include <regex.h>
 #include <string.h>
 
-ksregex* ksregex_new(const char *expression, unsigned nmatches, int flags)
+ksregex *ksregex_new(PCRE2_SPTR expression, unsigned nmatches, int options)
 {
 	ksregex *regex = malloc(sizeof(ksregex));
 	regex->nmatches = nmatches;
 	regex->matches = NULL;
 
-	if (regcomp(&regex->expr, expression, // compile expression in regex.
+	int error_code;
-			flags | REG_EXTENDED | (nmatches == 0 ? REG_NOSUB : 0)) != 0)
+	PCRE2_SIZE error_offset;
+	// compile expression in regex.
+	regex->expr = pcre2_compile(expression, strlen(expression), options, &error_code, &error_offset, NULL);
+
+	if (error_code < 0)
 		return NULL; // an error happened, returning NULL.
 
 	return regex; // return created regex.
 }
 
-void ksregex_save_string_matches(ksregex *regex, regmatch_t *raw_matches, const char *source)
+void ksregex_save_string_matches(ksregex *regex, pcre2_match_data *match_data)
 {
 	// allocate string matches array.
 	regex->matches = malloc(sizeof(char *) * regex->nmatches);
 
-	// save full string match.
+	for(unsigned i = 0; i < regex->nmatches; i++)
-	regex->matches[0] = strdup(source);
-
-	for(unsigned i = 1; i < regex->nmatches; i++)
 	{ // for each group match, converting it to an independent string.
-		regoff_t current_match_length = raw_matches[i].rm_eo - raw_matches[i].rm_so;
+		// get buffer of current pcre2 match.
-		regex->matches[i] = malloc(current_match_length + 1); // allocate the current group match string.
+		PCRE2_UCHAR* pcre2_buffer;
-		// copying the string part from source to the current match string.
+		PCRE2_SIZE pcre2_buffer_len;
-		strncpy(regex->matches[i], &source[raw_matches[i].rm_so], current_match_length);
+		pcre2_substring_get_bynumber(match_data, i, &pcre2_buffer, &pcre2_buffer_len);
-		regex->matches[i][current_match_length] = 0; // set end of string.
+
+		regex->matches[i] = malloc(pcre2_buffer_len + 1); // allocate the current group match string.
+		// copying the current buffer to the current match string.
+		strncpy(regex->matches[i], pcre2_buffer, pcre2_buffer_len);
+		regex->matches[i][pcre2_buffer_len] = 0; // set end of string.
+
+		pcre2_substring_free(pcre2_buffer); // free pcre2 buffer.
 	}
 }
 
@@ -38,21 +44,21 @@ bool ksregex_matches(ksregex *regex, const char *tested)
 	// if there was matches, free them.
 	_ksregex_free_matches(regex);
 
-	// allocate raw matches array.
+	// create match data.
-	regmatch_t *raw_matches = malloc(sizeof(regmatch_t) * regex->nmatches);
+	pcre2_match_data *match_data = pcre2_match_data_create(regex->nmatches, NULL);
 
-	if (regexec(&regex->expr, tested, regex->nmatches, raw_matches, 0) != 0)
+	if (pcre2_match(regex->expr, tested, strlen(tested), 0, 0, match_data, NULL) <= 0)
-	{ // an error happened, freeing results then returning NULL.
+	{ // an error happened, freeing results then returning false.
-		free(raw_matches); // free raw groups matches.
+		pcre2_match_data_free(match_data); // free match data.
 		return false;
 	}
 
 	if (regex->nmatches > 0)
 	{ // there are matches, getting them.
-		ksregex_save_string_matches(regex, raw_matches, tested);
+		ksregex_save_string_matches(regex, match_data);
 	}
 
-	free(raw_matches); // free raw groups matches.
+	pcre2_match_data_free(match_data); // free match data.
 
 	return true; // no error, return true.
 }

tests/ksregex.c

@@ -10,7 +10,7 @@ int main(void)
 	assert(reg1->matches == NULL);
 
 	// try a more complex regex with matches groups.
-	ksregex *reg2 = ksregex_new("^([0-9])([a-z])$", 3, REG_ICASE);
+	ksregex *reg2 = ksregex_new("^([0-9])([a-z])$", 3, PCRE2_CASELESS);
 	assert(ksregex_matches(reg2, "1a"));
 	assert(!ksregex_matches(reg2, "H8"));
 	assert(!ksregex_matches(reg2, "4MM"));