Implement GH-24074: Locale::acceptFromHttp() available locales list

UPGRADING

@@ -49,6 +49,14 @@ PHP 8.7 UPGRADE NOTES
 5. Changed Functions
 ========================================
 
+- Intl:
+  . Locale::acceptFromHttp() and locale_accept_from_http() now accept an
+    optional $availableLocales array argument. When given, the
+    Accept-Language header is matched against these locales instead of all
+    locales known to ICU, and the matching entry is returned as passed.
+    A ValueError is thrown if the array is empty or contains an invalid
+    locale, and a TypeError if it contains a non-string value.
+
 ========================================
 6. New Functions
 ========================================

ext/intl/locale/locale.stub.php

@@ -142,7 +142,7 @@ public static function canonicalize(string $locale): ?string {}
      * @tentative-return-type
      * @alias locale_accept_from_http
      */
-    public static function acceptFromHttp(string $header): string|false {}
+    public static function acceptFromHttp(string $header, ?array $availableLocales = null): string|false {}
 
     /**
      * @alias locale_is_right_to_left

ext/intl/locale/locale_methods.cpp

@@ -22,6 +22,8 @@
 #include <unicode/udata.h>
 #include <unicode/putil.h>
 #include <unicode/ures.h>
+#include <unicode/uenum.h>
+#include <unicode/locid.h>
 
 extern "C" {
 #include "php_intl.h"
@@ -1661,7 +1663,7 @@ PHP_INTL_FUNCTION_WITH_ERROR_RESET(locale_lookup)
 /* }}} */
 
 /* {{{ Tries to find out best available locale based on HTTP "Accept-Language" header */
-U_CFUNC PHP_FUNCTION(locale_accept_from_http)
+PHP_INTL_FUNCTION_WITH_ERROR_RESET(locale_accept_from_http)
 {
 	UEnumeration *available;
 	char *http_accept = nullptr;
@@ -1670,10 +1672,52 @@ U_CFUNC PHP_FUNCTION(locale_accept_from_http)
 	int len;
 	char resultLocale[INTL_MAX_LOCALE_LEN+1];
 	UAcceptResult outResult;
+	HashTable *available_locales = nullptr;
+	std::unique_ptr<const char *[]> available_list;
+	std::unique_ptr<icu::Locale[]> available_canonical;
+	uint32_t available_count = 0;
 
-	ZEND_PARSE_PARAMETERS_START(1, 1)
+	ZEND_PARSE_PARAMETERS_START(1, 2)
 		Z_PARAM_STRING(http_accept, http_accept_len)
+		Z_PARAM_OPTIONAL
+		Z_PARAM_ARRAY_HT_OR_NULL(available_locales)
 	ZEND_PARSE_PARAMETERS_END();
+
+	if (available_locales) {
+		uint32_t i = 0;
+		zval *entry;
+
+		available_count = zend_hash_num_elements(available_locales);
+		if (available_count == 0) {
+			zend_argument_must_not_be_empty_error(2);
+			RETURN_THROWS();
+		}
+
+		available_list.reset(new const char *[available_count]);
+		available_canonical.reset(new icu::Locale[available_count]);
+		ZEND_HASH_FOREACH_VAL(available_locales, entry) {
+			ZVAL_DEREF(entry);
+			if (Z_TYPE_P(entry) != IS_STRING) {
+				zend_argument_type_error(2, "must only contain string values");
+				RETURN_THROWS();
+			}
+			if (zend_str_has_nul_byte(Z_STR_P(entry))) {
+				zend_argument_value_error(2, "must not contain any null bytes");
+				RETURN_THROWS();
+			}
+			if (Z_STRLEN_P(entry) > INTL_MAX_LOCALE_LEN) {
+				zend_argument_value_error(2, "must not contain locales longer than %d characters", INTL_MAX_LOCALE_LEN);
+				RETURN_THROWS();
+			}
+			available_canonical[i] = icu::Locale(Z_STRVAL_P(entry));
+			if (available_canonical[i].isBogus()) {
+				zend_argument_value_error(2, "must only contain valid locales");
+				RETURN_THROWS();
+			}
+			available_list[i++] = Z_STRVAL_P(entry);
+		} ZEND_HASH_FOREACH_END();
+	}
+
 	if (UNEXPECTED(http_accept_len > ULOC_FULLNAME_CAPACITY)) {
 		/* check each fragment, if any bigger than capacity, can't do it due to bug #72533 */
 		char *start = http_accept;
@@ -1693,7 +1737,11 @@ U_CFUNC PHP_FUNCTION(locale_accept_from_http)
 		} while(end != nullptr);
 	}
 
-	available = ures_openAvailableLocales(nullptr, &status);
+	if (available_list) {
+		available = uenum_openCharStringsEnumeration(available_list.get(), available_count, &status);
+	} else {
+		available = ures_openAvailableLocales(nullptr, &status);
+	}
 	INTL_CHECK_STATUS(status, "failed to retrieve locale list");
 	len = uloc_acceptLanguageFromHTTP(resultLocale, INTL_MAX_LOCALE_LEN,
 						&outResult, http_accept, available, &status);
@@ -1702,6 +1750,12 @@ U_CFUNC PHP_FUNCTION(locale_accept_from_http)
 	if (UNEXPECTED(len < 0 || outResult == ULOC_ACCEPT_FAILED)) {
 		RETURN_FALSE;
 	}
+	resultLocale[len] = '\0';
+	for (uint32_t i = 0; i < available_count; i++) {
+		if (strcmp(available_list[i], resultLocale) == 0 || strcmp(available_canonical[i].getName(), resultLocale) == 0) {
+			RETURN_STRING(available_list[i]);
+		}
+	}
 	RETURN_STRINGL(resultLocale, len);
 }
 /* }}} */

ext/intl/php_intl.stub.php

@@ -503,7 +503,7 @@ function locale_canonicalize(string $locale): ?string {}
 
 function locale_lookup(array $languageTag, string $locale, bool $canonicalize = false, ?string $defaultLocale = null): ?string {}
 
-function locale_accept_from_http(string $header): string|false {}
+function locale_accept_from_http(string $header, ?array $availableLocales = null): string|false {}
 
 function locale_is_right_to_left(string $locale): bool {}
 

ext/intl/tests/gh24074.phpt

@@ -0,0 +1,60 @@
+--TEST--
+GH-24074 (Locale::acceptFromHttp() with a list of available locales)
+--EXTENSIONS--
+intl
+--SKIPIF--
+<?php if (version_compare(INTL_ICU_VERSION, '67.1') < 0) die('skip for ICU >= 67.1'); ?>
+--FILE--
+<?php
+
+$header = 'fr-CH, fr;q=0.9, en;q=0.8, de;q=0.7';
+
+var_dump(Locale::acceptFromHttp($header, ['en_US', 'de_DE']));
+var_dump(Locale::acceptFromHttp($header, ['de', 'en']));
+var_dump(Locale::acceptFromHttp($header, ['de']));
+var_dump(Locale::acceptFromHttp('de-CH, en;q=0.5', ['en', 'de']));
+var_dump(Locale::acceptFromHttp('en-us', ['de', 'en_US']));
+var_dump(Locale::acceptFromHttp('en-us', ['de', 'en-US']));
+var_dump(Locale::acceptFromHttp('ja', ['en', 'de']));
+var_dump(Locale::acceptFromHttp('en-us,en;q=0.5', null));
+var_dump(locale_accept_from_http($header, ['de', 'en']));
+
+var_dump(Locale::acceptFromHttp(str_repeat('a', 200), ['en']));
+var_dump(Locale::acceptFromHttp('en', ['en']));
+var_dump(intl_get_error_message());
+
+$invalid = [
+    [],
+    [1],
+    ["e\0n"],
+    [str_repeat('a', 200)],
+    ['en', str_repeat('a', 20)],
+];
+
+foreach ($invalid as $locales) {
+    try {
+        Locale::acceptFromHttp('en', $locales);
+    } catch (\Throwable $e) {
+        echo $e::class, ': ', $e->getMessage(), PHP_EOL;
+    }
+}
+
+?>
+--EXPECT--
+string(5) "en_US"
+string(2) "en"
+string(2) "de"
+string(2) "de"
+string(5) "en_US"
+string(5) "en-US"
+bool(false)
+string(5) "en_US"
+string(2) "en"
+bool(false)
+string(2) "en"
+string(12) "U_ZERO_ERROR"
+ValueError: Locale::acceptFromHttp(): Argument #2 ($availableLocales) must not be empty
+TypeError: Locale::acceptFromHttp(): Argument #2 ($availableLocales) must only contain string values
+ValueError: Locale::acceptFromHttp(): Argument #2 ($availableLocales) must not contain any null bytes
+ValueError: Locale::acceptFromHttp(): Argument #2 ($availableLocales) must not contain locales longer than 156 characters
+ValueError: Locale::acceptFromHttp(): Argument #2 ($availableLocales) must only contain valid locales

ext/intl/tests/locale_accept_from_http_error_reset.phpt

@@ -0,0 +1,32 @@
+--TEST--
+Locale::acceptFromHttp() resets intl error before argument parsing
+--EXTENSIONS--
+intl
+--INI--
+intl.error_level=0
+intl.use_exceptions=0
+--FILE--
+<?php
+foreach (['Locale::acceptFromHttp', 'locale_accept_from_http'] as $func) {
+    $func(str_repeat('a', 200));
+    var_dump(intl_get_error_code() === U_ILLEGAL_ARGUMENT_ERROR);
+
+    try {
+        $func('en', 123);
+    } catch (\Throwable $e) {
+        echo $e::class, "\n";
+    }
+
+    var_dump(intl_get_error_code() === U_ZERO_ERROR);
+    var_dump(intl_get_error_message());
+}
+?>
+--EXPECT--
+bool(true)
+TypeError
+bool(true)
+string(12) "U_ZERO_ERROR"
+bool(true)
+TypeError
+bool(true)
+string(12) "U_ZERO_ERROR"

ext/intl/tests/locale_bug74439.phpt

@@ -33,7 +33,7 @@ foreach ($methods as $method) {
 ?>
 --EXPECT--
 getDefault: 0, 0
-acceptFromHttp: 1, 1
+acceptFromHttp: 2, 1
 canonicalize: 1, 1
 composeLocale: 1, 1
 getAllVariants: 1, 1