Merge pull request libgit2#3953 from arthurschreiber/arthur/fix-regcomp-locale-issues

Make sure we use the `C` locale for `regcomp` on macOS.

Author: carlosmn

CMakeLists.txt

@@ -20,6 +20,7 @@ SET(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} "${CMAKE_CURRENT_SOURCE_DIR}/cmake/Mo
 
 INCLUDE(CheckLibraryExists)
 INCLUDE(CheckFunctionExists)
+INCLUDE(CheckSymbolExists)
 INCLUDE(CheckStructHasMember)
 INCLUDE(AddCFlagIfSupported)
 INCLUDE(FindPkgConfig)
@@ -507,6 +508,11 @@ ELSE ()
 	ENDIF ()
 ENDIF()
 
+CHECK_SYMBOL_EXISTS(regcomp_l "xlocale.h" HAVE_REGCOMP_L)
+IF (HAVE_REGCOMP_L)
+	ADD_DEFINITIONS(-DHAVE_REGCOMP_L)
+ENDIF ()
+
 CHECK_FUNCTION_EXISTS(futimens HAVE_FUTIMENS)
 IF (HAVE_FUTIMENS)
 	ADD_DEFINITIONS(-DHAVE_FUTIMENS)

src/config.c

@@ -478,7 +478,7 @@ int git_config_iterator_glob_new(git_config_iterator **out, const git_config *cf
 	iter = git__calloc(1, sizeof(all_iter));
 	GITERR_CHECK_ALLOC(iter);
 
-	if ((result = regcomp(&iter->regex, regexp, REG_EXTENDED)) != 0) {
+	if ((result = p_regcomp(&iter->regex, regexp, REG_EXTENDED)) != 0) {
 		giterr_set_regex(&iter->regex, result);
 		git__free(iter);
 		return -1;
@@ -512,7 +512,7 @@ int git_config_backend_foreach_match(
 	int error = 0;
 
 	if (regexp != NULL) {
-		if ((error = regcomp(&regex, regexp, REG_EXTENDED)) != 0) {
+		if ((error = p_regcomp(&regex, regexp, REG_EXTENDED)) != 0) {
 			giterr_set_regex(&regex, error);
 			regfree(&regex);
 			return -1;
@@ -1003,7 +1003,7 @@ int git_config_multivar_iterator_new(git_config_iterator **out, const git_config
 		goto on_error;
 
 	if (regexp != NULL) {
-		error = regcomp(&iter->regex, regexp, REG_EXTENDED);
+		error = p_regcomp(&iter->regex, regexp, REG_EXTENDED);
 		if (error != 0) {
 			giterr_set_regex(&iter->regex, error);
 			error = -1;

src/config_file.c

@@ -570,7 +570,7 @@ static int config_set_multivar(
 	if ((result = git_config__normalize_name(name, &key)) < 0)
 		return result;
 
-	result = regcomp(&preg, regexp, REG_EXTENDED);
+	result = p_regcomp(&preg, regexp, REG_EXTENDED);
 	if (result != 0) {
 		giterr_set_regex(&preg, result);
 		result = -1;
@@ -657,7 +657,7 @@ static int config_delete_multivar(git_config_backend *cfg, const char *name, con
 
 	refcounted_strmap_free(map);
 
-	result = regcomp(&preg, regexp, REG_EXTENDED);
+	result = p_regcomp(&preg, regexp, REG_EXTENDED);
 	if (result != 0) {
 		giterr_set_regex(&preg, result);
 		result = -1;
@@ -1957,4 +1957,3 @@ static int config_write(diskfile_backend *cfg, const char *key, const regex_t *p
 	git_buf_free(&reader->buffer);
 	return result;
 }
-

src/diff_driver.c

@@ -114,7 +114,7 @@ static int diff_driver_add_patterns(
 		if (error < 0)
 			break;
 
-		if ((error = regcomp(&pat->re, buf.ptr, regex_flags)) != 0) {
+		if ((error = p_regcomp(&pat->re, buf.ptr, regex_flags)) != 0) {
 			/*
 			 * TODO: issue a warning
 			 */
@@ -210,7 +210,7 @@ static int git_diff_driver_builtin(
 		goto done;
 
 	if (ddef->words &&
-		(error = regcomp(
+		(error = p_regcomp(
 			&drv->word_pattern, ddef->words, ddef->flags | REG_EXTENDED)))
 	{
 		error = giterr_set_regex(&drv->word_pattern, error);
@@ -314,7 +314,7 @@ static int git_diff_driver_load(
 		goto done;
 	if (!ce || !ce->value)
 		/* no diff.<driver>.wordregex, so just continue */;
-	else if (!(error = regcomp(&drv->word_pattern, ce->value, REG_EXTENDED)))
+	else if (!(error = p_regcomp(&drv->word_pattern, ce->value, REG_EXTENDED)))
 		found_driver = true;
 	else {
 		/* TODO: warn about bad regex instead of failure */
@@ -519,4 +519,3 @@ void git_diff_find_context_clear(git_diff_find_context_payload *payload)
 		payload->driver = NULL;
 	}
 }
-

src/revparse.c

@@ -50,7 +50,7 @@ static int build_regex(regex_t *regex, const char *pattern)
 		return GIT_EINVALIDSPEC;
 	}
 
-	error = regcomp(regex, pattern, REG_EXTENDED);
+	error = p_regcomp(regex, pattern, REG_EXTENDED);
 	if (!error)
 		return 0;
 

src/unix/posix.h

@@ -80,4 +80,14 @@ GIT_INLINE(int) p_futimes(int f, const struct p_timeval t[2])
 # define p_futimes futimes
 #endif
 
+#ifdef HAVE_REGCOMP_L
+#include <xlocale.h>
+GIT_INLINE(int) p_regcomp(regex_t *preg, const char *pattern, int cflags)
+{
+	return regcomp_l(preg, pattern, cflags, (locale_t) 0);
+}
+#else
+# define p_regcomp regcomp
+#endif
+
 #endif

src/win32/posix.h

@@ -57,4 +57,7 @@ extern int p_lstat_posixly(const char *filename, struct stat *buf);
 extern struct tm * p_localtime_r(const time_t *timer, struct tm *result);
 extern struct tm * p_gmtime_r(const time_t *timer, struct tm *result);
 
+/* Use the bundled regcomp */
+#define p_regcomp regcomp
+
 #endif

tests/core/posix.c

@@ -11,6 +11,7 @@
 
 #include "clar_libgit2.h"
 #include "posix.h"
+#include "userdiff.h"
 
 void test_core_posix__initialize(void)
 {
@@ -39,7 +40,7 @@ void test_core_posix__inet_pton(void)
 	struct in_addr addr;
 	struct in6_addr addr6;
 	size_t i;
-	
+
 	struct in_addr_data {
 		const char *p;
 		const uint8_t n[4];
@@ -146,3 +147,27 @@ void test_core_posix__utimes(void)
 
 	p_unlink("foo");
 }
+
+void test_core_posix__p_regcomp_compile_single_byte_regexps(void)
+{
+	regex_t preg;
+
+	cl_must_pass(p_regcomp(&preg, "[\xc0-\xff][\x80-\xbf]", REG_EXTENDED));
+
+	regfree(&preg);
+}
+
+void test_core_posix__p_regcomp_compile_userdiff_regexps(void)
+{
+	regex_t preg;
+	size_t idx;
+
+	for (idx = 0; idx < ARRAY_SIZE(builtin_defs); ++idx) {
+		git_diff_driver_definition ddef = builtin_defs[idx];
+
+		cl_must_pass(p_regcomp(&preg, ddef.fns, REG_EXTENDED | ddef.flags));
+		cl_must_pass(p_regcomp(&preg, ddef.words, REG_EXTENDED));
+	}
+
+	regfree(&preg);
+}