better logic

Author: LamentXU123

NEWS

@@ -23,9 +23,9 @@ PHP                                                                        NEWS
 
 - Intl:
   . Fixed Collator::sort(), collator_sort(), Collator::asort(), and
-    collator_asort() to report string conversion errors through the intl error
-    handler instead of emitting a warning and continuing with an empty string.
-    (Weilin Du)
+    collator_asort() to report UTF-8/UTF-16 conversion errors through the intl
+    error handler instead of emitting a warning and continuing with an empty
+    string. (Weilin Du)
   . Fixed grammatical issues in Normalizer invalid form and IntlCalendar time
     zone offset error messages. (Weilin Du)
 

UPGRADING

@@ -77,11 +77,10 @@ PHP 8.6 UPGRADE NOTES
     when the offset argument is not of type int instead of silently converting
     the value.
   . Collator::sort(), collator_sort(), Collator::asort(), and
-    collator_asort() now report string conversion failures during comparison
-    through the intl error mechanism and return false. With
-    intl.use_exceptions enabled, these failures throw IntlException.
-    Previously, these paths emitted a warning and compared the value as an
-    empty string.
+    collator_asort() now report UTF-8/UTF-16 conversion failures during
+    comparison through the intl error mechanism and return false. With
+    intl.use_exceptions enabled, these failures throw IntlException. Previously,
+    these paths emitted a warning and compared the value as an empty string.
 
 - PCNTL:
   . pcntl_alarm() now raises a ValueError if the seconds argument is

ext/intl/collator/collator_convert.cpp

@@ -45,7 +45,11 @@ static void collator_set_conversion_error(UErrorCode status, const char *message
 	}
 
 	ZEND_ASSERT(INTL_G(current_collator_error) != nullptr);
-	intl_error_set(INTL_G(current_collator_error), status, message);
+	intl_error *err = INTL_G(current_collator_error);
+	intl_error_reset(err);
+	err->code = status;
+	err->custom_error_message = zend_string_init(
+		message, strlen(message), false);
 }
 
 /* {{{ collator_convert_hash_item_from_utf8_to_utf16 */
@@ -223,19 +227,19 @@ U_CFUNC zval* collator_convert_object_to_string( zval* obj, zval *rv )
 		/* cast_object failed => bail out. */
 		zval_ptr_dtor( zstr );
 		ZVAL_NULL( zstr );
-		if( !EG(exception) ) {
-			zend_throw_error(NULL, "Object of class %s could not be converted to string", ZSTR_VAL(Z_OBJCE_P(obj)->name));
+		if( EG(exception) ) {
+			return nullptr;
 		}
-		return nullptr;
+		COLLATOR_CONVERT_RETURN_FAILED( obj );
 	}
 
 	/* Object wasn't successfully converted => bail out. */
 	if( zstr == nullptr )
 	{
-		if( !EG(exception) ) {
-			zend_throw_error(NULL, "Object of class %s could not be converted to string", ZSTR_VAL(Z_OBJCE_P(obj)->name));
+		if( EG(exception) ) {
+			return nullptr;
 		}
-		return nullptr;
+		COLLATOR_CONVERT_RETURN_FAILED( obj );
 	}
 
 	/* Convert the string to UTF-16. */

ext/intl/collator/collator_sort.cpp

@@ -50,22 +50,14 @@ static const size_t DEF_SORT_KEYS_INDX_BUF_INCREMENT = 1048576;
 
 static const size_t DEF_UTF16_BUF_SIZE = 1024;
 
-static void collator_copy_error(intl_error *dst, const intl_error *src)
-{
-	intl_error_reset(dst);
-	dst->code = src->code;
-	if (src->custom_error_message) {
-		dst->custom_error_message = zend_string_copy(src->custom_error_message);
-	}
-}
-
 static void collator_report_sort_error(Collator_object *co, const intl_error *sort_error)
 {
-	collator_copy_error(COLLATOR_ERROR_P(co), sort_error);
-	collator_copy_error(&INTL_G(g_error), sort_error);
-
-	if (INTL_G(use_exceptions) && INTL_G(g_error).custom_error_message) {
-		zend_throw_error_exception(IntlException_ce_ptr, INTL_G(g_error).custom_error_message, 0, 0);
+	if (sort_error->custom_error_message) {
+		intl_errors_set(
+			COLLATOR_ERROR_P(co), sort_error->code,
+			ZSTR_VAL(sort_error->custom_error_message));
+	} else {
+		intl_errors_set_code(COLLATOR_ERROR_P(co), sort_error->code);
 	}
 }
 

ext/intl/tests/collator_sort_conversion_error_level.phpt

@@ -0,0 +1,29 @@
+--TEST--
+Collator::sort() reports comparison conversion errors through intl.error_level
+--EXTENSIONS--
+intl
+--FILE--
+<?php
+class BadErrorLevelString {
+    public function __toString(): string {
+        return "\xFF";
+    }
+}
+
+ini_set("intl.error_level", E_WARNING);
+ini_set("intl.use_exceptions", 0);
+
+$coll = new Collator('en_US');
+$array = ['b', new BadErrorLevelString(), 'a'];
+
+var_dump($coll->sort($array, Collator::SORT_STRING));
+var_dump($coll->getErrorCode() === U_INVALID_CHAR_FOUND);
+echo $coll->getErrorMessage(), PHP_EOL;
+?>
+--EXPECTF--
+Deprecated: ini_set(): Using a value different than 0 for intl.error_level is deprecated, as the intl.error_level INI setting is deprecated. Instead the intl.use_exceptions INI setting should be enabled to throw exceptions on errors or intl_get_error_code()/intl_get_error_message() should be used to manually deal with errors in %s on line %d
+
+Warning: Collator::sort(): Error converting string from UTF-8 to UTF-16 in %s on line %d
+bool(false)
+bool(true)
+Collator::sort(): Error converting string from UTF-8 to UTF-16: U_INVALID_CHAR_FOUND

ext/intl/tests/collator_sort_error_reset_reentrancy.phpt

@@ -1,41 +1,36 @@
 --TEST--
-Collator::sort() conversion failure must survive a re-entrant Collator call
+Collator::sort() conversion failure must survive a re-entrant Collator call in __toString()
 --EXTENSIONS--
 intl
 --FILE--
 <?php
-class Bad {
-    public function __toString(): string {
-        return "\xFF";
-    }
-}
-
-class Toucher {
+class BadAfterReentry {
     public static Collator $coll;
+    public static bool $called = false;
 
     public function __toString(): string {
-        if (self::$coll->getErrorCode() !== U_ZERO_ERROR) {
-            self::$coll->getLocale(Locale::VALID_LOCALE);
-        }
-        return 'z';
+        self::$called = true;
+        self::$coll->getLocale(Locale::VALID_LOCALE);
+        return "\xFF";
     }
 }
 
 $coll = new Collator('en_US');
-Toucher::$coll = $coll;
+BadAfterReentry::$coll = $coll;
 
-$array = ['a', new Bad(), 'b', new Toucher()];
+$array = ['a', new BadAfterReentry(), 'b'];
 
 var_dump($coll->sort($array, Collator::SORT_STRING));
+var_dump(BadAfterReentry::$called);
 var_dump($coll->getErrorCode() === U_INVALID_CHAR_FOUND);
 echo $coll->getErrorMessage(), PHP_EOL;
-var_dump($array[0], $array[1] instanceof Bad, $array[2], $array[3] instanceof Toucher);
+var_dump($array[0], $array[1] instanceof BadAfterReentry, $array[2]);
 ?>
 --EXPECT--
 bool(false)
 bool(true)
+bool(true)
 Collator::sort(): Error converting string from UTF-8 to UTF-16: U_INVALID_CHAR_FOUND
 string(1) "a"
 bool(true)
 string(1) "b"
-bool(true)

ext/intl/tests/collator_sort_tostring_throws.phpt

@@ -31,7 +31,7 @@ foreach ([Collator::SORT_REGULAR, Collator::SORT_STRING] as $flag) {
 
 $array = ['b', new NoToString(), 'a'];
 try {
-    var_dump($coll->sort($array, Collator::SORT_REGULAR));
+    var_dump($coll->sort($array, Collator::SORT_STRING));
 } catch (Throwable $e) {
     echo get_class($e), ': ', $e->getMessage(), PHP_EOL;
 }