feat: add library method to validate and canonicalize locale ID

php-ecma-intl · Oct 12, 2023 · f79edc5 · f79edc5
1 parent 1670424
commit f79edc5
Show file tree

Hide file tree

Showing 4 changed files with 113 additions and 0 deletions.
diff --git a/config.m4 b/config.m4
@@ -161,6 +161,7 @@ if test "$PHP_ECMA_INTL" != "no"; then
       tests/criterion/ecma402/locale_min_max_test.c                            \
       tests/criterion/ecma402/locale_options_test.c                            \
       tests/criterion/ecma402/locale_test.c                                    \
+      tests/criterion/ecma402/locale_validation_test.cpp                       \
       tests/criterion/ecma402/numbering_system_test.c                          \
       tests/criterion/ecma402/time_zone_test.c                                 \
       tests/criterion/ecma402/unit_test.c                                      \

diff --git a/src/ecma402/locale.cpp b/src/ecma402/locale.cpp
@@ -608,6 +608,34 @@ int ecma402_minimize(const char *localeId, char *minimized, ecma402_errorStatus
 	return getMaxOrMin(MINIMIZE, localeId, minimized, status, isCanonicalized);
 }
 
+int ecma402_validateAndCanonicalizeUnicodeLocaleId(const char *localeId, char *canonicalized,
+                                                   ecma402_errorStatus *status)
+{
+	char **available, *bestAvailable, *tmp;
+	size_t total, length, resultLength = -1;
+
+	available = (char **)malloc(sizeof(char *) * uloc_countAvailable());
+	bestAvailable = (char *)malloc(sizeof(char) * ULOC_FULLNAME_CAPACITY);
+	total = ecma402_intlAvailableLocales(available);
+
+	if (ecma402_bestAvailableLocale(available, total, localeId, bestAvailable, false) > 0) {
+		tmp = (char *)malloc(sizeof(char) * ULOC_FULLNAME_CAPACITY);
+		length = languageTagForLocaleId(localeId, tmp, status);
+
+		if (!ecma402_hasError(status) && length > 0) {
+			strcpy(canonicalized, tmp);
+			resultLength = length;
+		}
+
+		free(tmp);
+	}
+
+	free(bestAvailable);
+	free(available);
+
+	return resultLength;
+}
+
 namespace {
 
 	int getHourCyclesForLocale(char *localeId, const char **values)

diff --git a/src/ecma402/locale.h b/src/ecma402/locale.h
@@ -498,6 +498,27 @@ int ecma402_maximize(const char *localeId, char *maximized, ecma402_errorStatus
  */
 int ecma402_minimize(const char *localeId, char *minimized, ecma402_errorStatus *status, bool isCanonicalized);
 
+/**
+ * Returns the Unicode canonicalized locale identifier form of the locale ID
+ * after validating the locale is supported by this implementation (using
+ * ecma402_bestAvailableLocale()).
+ *
+ * The result parameter should already be allocated on the stack with
+ * enough memory to store the buffer. Typically, this should use
+ * ULOC_FULLNAME_CAPACITY. For example:
+ *
+ *     malloc(sizeof(char) * ULOC_FULLNAME_CAPACITY)
+ *
+ * @param localeId The locale identifier to canonicalize.
+ * @param canonicalized A buffer in which to store the canonicalized name.
+ * @param status A status object to pass error messages back to the caller.
+ *
+ * @return The length of the string stored to the canonicalized buffer, or -1 if
+ * the localeId cannot be validated or canonicalized.
+ */
+int ecma402_validateAndCanonicalizeUnicodeLocaleId(const char *localeId, char *canonicalized,
+                                                   ecma402_errorStatus *status);
+
 #ifdef __cplusplus
 }
 #endif

diff --git a/tests/criterion/ecma402/locale_validation_test.cpp b/tests/criterion/ecma402/locale_validation_test.cpp
@@ -0,0 +1,63 @@
+#include "../test.h"
+
+#include "src/ecma402/locale.h"
+
+#include <unicode/uloc.h>
+
+#define TEST_SUITE ecma402LocaleValidation
+
+// NOLINTBEGIN(cert-err58-cpp, misc-const-correctness,
+//             misc-use-anonymous-namespace)
+
+using string = std::basic_string<char, std::char_traits<char>, criterion::allocator<char>>;
+
+struct localeIdValidationTest {
+	string localeId;
+	string expected;
+
+	localeIdValidationTest(string localeId, string expected) : localeId(localeId), expected(expected) {}
+};
+
+ParameterizedTestParameters(TEST_SUITE, validateAndCanonicalizeUnicodeLocaleId)
+{
+	static criterion::parameters<struct localeIdValidationTest> tests;
+
+	tests.emplace_back("en-US", "en-US");
+	tests.emplace_back("en_US", "en-US");
+	tests.emplace_back("en-Latn-US", "en-Latn-US");
+	tests.emplace_back("en-US-u-nu-latn-ca-gregory", "en-US-u-ca-gregory-nu-latn");
+	tests.emplace_back("foobar", "-1");
+	tests.emplace_back("zz-ZZ", "-1");
+	tests.emplace_back("und-u-va-posix", "-1");
+	tests.emplace_back("en-US-POSIX", "en-US-u-va-posix");
+	tests.emplace_back("en_US_POSIX", "en-US-u-va-posix");
+
+	return tests;
+}
+
+ParameterizedTest(struct localeIdValidationTest *test, TEST_SUITE, validateAndCanonicalizeUnicodeLocaleId)
+{
+	char *result;
+	size_t resultLength;
+	ecma402_errorStatus *status;
+
+	status = ecma402_initErrorStatus();
+	result = (char *)malloc(sizeof(char) * ULOC_FULLNAME_CAPACITY);
+	resultLength = ecma402_validateAndCanonicalizeUnicodeLocaleId(test->localeId.c_str(), result, status);
+
+	cr_assert(eq(i8, ecma402_hasError(status), 0));
+
+	if (test->expected == "-1") {
+		cr_expect(eq(i8, resultLength, -1));
+	} else {
+		cr_expect(eq(str, result, test->expected.c_str()),
+		          "Expected canonicalized value of \"%s\" for language tag \"%s\"; got \"%s\" instead",
+		          test->expected.c_str(), test->localeId.c_str(), result);
+		cr_expect(eq(i8, resultLength, test->expected.length()));
+	}
+
+	free(result);
+}
+
+// NOLINTEND(cert-err58-cpp, misc-const-correctness,
+//           misc-use-anonymous-namespace)