[pull] master from php:master

NEWS

@@ -13,6 +13,8 @@ PHP                                                                        NEWS
     className/classList writes and attribute removals. (Ilia Alshanetsky)
 
 - Intl:
+  . Fixed grapheme_strpos() and grapheme_strrpos() with an empty needle
+    returning UTF-16 offsets instead of grapheme offsets. (Ilia Alshanetsky)
   . Fixed a memory leak when dumping IntlCalendar instances. (Ilia Alshanetsky)
   . Fixed Collator::sortWithSortKeys() allocating fixed 2MiB buffers
     regardless of array size. (Ilia Alshanetsky)
@@ -22,6 +24,13 @@ PHP                                                                        NEWS
     read. (iliaal)
   . Fixed a use-after-free when IntlRuleBasedBreakIterator is constructed
     from compiled rules. (iliaal)
+  . Fixed Spoofchecker methods not recording the ICU error code when an ICU
+    call fails. (Ilia Alshanetsky)
+
+- MBString:
+  . Fixed mb_ereg_replace() emitting a NUL or out-of-bounds bytes in the
+    replacement when a \k<name> backref has no closing delimiter.
+    (Ilia Alshanetsky)
 
 - PDO_PGSQL:
   . Added Pdo\Pgsql::ATTR_CHUNK_SIZE to fetch a result set in chunks of the

ext/date/php_date.c

@@ -3258,7 +3258,12 @@ PHP_FUNCTION(date_format)
 }
 /* }}} */
 
-static bool php_date_modify(zval *object, char *modify, size_t modify_len) /* {{{ */
+typedef enum {
+	PHP_DATE_MODIFY_WARNING,
+	PHP_DATE_MODIFY_THROW
+} php_date_modify_error_mode;
+
+static bool php_date_modify(zval *object, char *modify, size_t modify_len, const php_date_modify_error_mode error_mode) /* {{{ */
 {
 	php_date_obj *dateobj;
 	timelib_time *tmp_time;
@@ -3278,10 +3283,22 @@ static bool php_date_modify(zval *object, char *modify, size_t modify_len) /* {{
 
 	if (err && err->error_count) {
 		/* spit out the first library error message, at least */
-		php_error_docref(NULL, E_WARNING, "Failed to parse time string (%s) at position %d (%c): %s", modify,
-			err->error_messages[0].position,
-			err->error_messages[0].character ? err->error_messages[0].character : ' ',
-			err->error_messages[0].message);
+		if (error_mode == PHP_DATE_MODIFY_THROW) {
+			zend_string *func_name = get_active_function_or_method_name();
+			zend_throw_exception_ex(date_ce_date_malformed_string_exception, 0,
+				"%s(): Failed to parse time string (%s) at position %d (%c): %s",
+				ZSTR_VAL(func_name),
+				modify,
+				err->error_messages[0].position,
+				err->error_messages[0].character ? err->error_messages[0].character : ' ',
+				err->error_messages[0].message);
+			zend_string_release_ex(func_name, false);
+		} else {
+			php_error_docref(NULL, E_WARNING, "Failed to parse time string (%s) at position %d (%c): %s", modify,
+				err->error_messages[0].position,
+				err->error_messages[0].character ? err->error_messages[0].character : ' ',
+				err->error_messages[0].message);
+		}
 		timelib_time_dtor(tmp_time);
 		return false;
 	}
@@ -3350,7 +3367,7 @@ PHP_FUNCTION(date_modify)
 		RETURN_THROWS();
 	}
 
-	if (!php_date_modify(object, modify, modify_len)) {
+	if (!php_date_modify(object, modify, modify_len, PHP_DATE_MODIFY_WARNING)) {
 		RETURN_FALSE;
 	}
 
@@ -3364,21 +3381,16 @@ PHP_METHOD(DateTime, modify)
 	zval                *object;
 	char                *modify;
 	size_t               modify_len;
-	zend_error_handling  zeh;
 
 	object = ZEND_THIS;
 	ZEND_PARSE_PARAMETERS_START(1, 1)
 		Z_PARAM_STRING(modify, modify_len)
 	ZEND_PARSE_PARAMETERS_END();
 
-	zend_replace_error_handling(EH_THROW, date_ce_date_malformed_string_exception, &zeh);
-	if (!php_date_modify(object, modify, modify_len)) {
-		zend_restore_error_handling(&zeh);
+	if (!php_date_modify(object, modify, modify_len, PHP_DATE_MODIFY_THROW)) {
 		RETURN_THROWS();
 	}
 
-	zend_restore_error_handling(&zeh);
-
 	RETURN_OBJ_COPY(Z_OBJ_P(object));
 }
 /* }}} */
@@ -3389,7 +3401,6 @@ PHP_METHOD(DateTimeImmutable, modify)
 	zval *object, new_object;
 	char *modify;
 	size_t   modify_len;
-	zend_error_handling zeh;
 
 	object = ZEND_THIS;
 	ZEND_PARSE_PARAMETERS_START(1, 1)
@@ -3398,15 +3409,11 @@ PHP_METHOD(DateTimeImmutable, modify)
 
 	date_clone_immutable(object, &new_object);
 
-	zend_replace_error_handling(EH_THROW, date_ce_date_malformed_string_exception, &zeh);
-	if (!php_date_modify(&new_object, modify, modify_len)) {
+	if (!php_date_modify(&new_object, modify, modify_len, PHP_DATE_MODIFY_THROW)) {
 		zval_ptr_dtor(&new_object);
-		zend_restore_error_handling(&zeh);
 		RETURN_THROWS();
 	}
 
-	zend_restore_error_handling(&zeh);
-
 	RETURN_OBJ(Z_OBJ(new_object));
 }
 /* }}} */
@@ -3463,7 +3470,7 @@ PHP_METHOD(DateTimeImmutable, add)
 }
 /* }}} */
 
-static void php_date_sub(zval *object, zval *interval, zval *return_value) /* {{{ */
+static void php_date_sub(zval *object, zval *interval, zval *return_value, const bool should_throw) /* {{{ */
 {
 	php_date_obj     *dateobj;
 	php_interval_obj *intobj;
@@ -3475,7 +3482,15 @@ static void php_date_sub(zval *object, zval *interval, zval *return_value) /* {{
 	DATE_CHECK_INITIALIZED(intobj->initialized, Z_OBJCE_P(interval));
 
 	if (intobj->diff->have_weekday_relative || intobj->diff->have_special_relative) {
-		php_error_docref(NULL, E_WARNING, "Only non-special relative time specifications are supported for subtraction");
+		if (should_throw) {
+			zend_string *func_name = get_active_function_or_method_name();
+			zend_throw_exception_ex(date_ce_date_invalid_operation_exception, 0,
+				"%s(): Only non-special relative time specifications are supported for subtraction",
+				ZSTR_VAL(func_name));
+			zend_string_release_ex(func_name, false);
+		} else {
+			php_error_docref(NULL, E_WARNING, "Only non-special relative time specifications are supported for subtraction");
+		}
 		return;
 	}
 
@@ -3497,7 +3512,7 @@ PHP_FUNCTION(date_sub)
 		RETURN_THROWS();
 	}
 
-	php_date_sub(object, interval, return_value);
+	php_date_sub(object, interval, return_value, false);
 	RETURN_OBJ_COPY(Z_OBJ_P(object));
 }
 /* }}} */
@@ -3506,15 +3521,12 @@ PHP_FUNCTION(date_sub)
 PHP_METHOD(DateTime, sub)
 {
 	zval *object, *interval;
-	zend_error_handling zeh;
 
 	if (zend_parse_method_parameters(ZEND_NUM_ARGS(), getThis(), "OO", &object, date_ce_date, &interval, date_ce_interval) == FAILURE) {
 		RETURN_THROWS();
 	}
 
-	zend_replace_error_handling(EH_THROW, date_ce_date_invalid_operation_exception, &zeh);
-	php_date_sub(object, interval, return_value);
-	zend_restore_error_handling(&zeh);
+	php_date_sub(object, interval, return_value, true);
 
 	RETURN_OBJ_COPY(Z_OBJ_P(object));
 }
@@ -3524,7 +3536,6 @@ PHP_METHOD(DateTime, sub)
 PHP_METHOD(DateTimeImmutable, sub)
 {
 	zval *object, *interval, new_object;
-	zend_error_handling zeh;
 
 	object = ZEND_THIS;
 	ZEND_PARSE_PARAMETERS_START(1, 1)
@@ -3533,9 +3544,7 @@ PHP_METHOD(DateTimeImmutable, sub)
 
 	date_clone_immutable(object, &new_object);
 
-	zend_replace_error_handling(EH_THROW, date_ce_date_invalid_operation_exception, &zeh);
-	php_date_sub(&new_object, interval, return_value);
-	zend_restore_error_handling(&zeh);
+	php_date_sub(&new_object, interval, return_value, true);
 
 	RETURN_OBJ(Z_OBJ(new_object));
 }

ext/hash/hash.c

@@ -17,7 +17,6 @@
 #include <config.h>
 #endif
 
-#include <math.h>
 #include "php_hash.h"
 #include "ext/standard/info.h"
 #include "ext/standard/file.h"
@@ -1033,10 +1032,10 @@ PHP_FUNCTION(hash_pbkdf2)
 	}
 	digest_length = length;
 	if (!raw_output) {
-		digest_length = (zend_long) ceil((float) length / 2.0);
+		digest_length = length / 2 + (length % 2);
 	}
 
-	loops = (zend_long) ceil((float) digest_length / (float) ops->digest_size);
+	loops = (digest_length - 1) / ops->digest_size + 1;
 
 	result = safe_emalloc(loops, ops->digest_size, 0);
 

ext/hash/tests/hash_pbkdf2_large_length.phpt

@@ -0,0 +1,19 @@
+--TEST--
+Hash: hash_pbkdf2() function : large output length
+--FILE--
+<?php
+
+$length = 33554433;
+$hash = hash_pbkdf2('md5', 'password', 'salt', 1, $length);
+
+/* The last hexit comes from the first byte of the final PBKDF2 block. */
+$block = intdiv(intdiv($length + 1, 2) - 1, 16) + 1;
+$expected = bin2hex(hash_hmac('md5', 'salt' . pack('N', $block), 'password', true));
+
+var_dump(strlen($hash));
+var_dump($hash[$length - 1] === $expected[0]);
+
+?>
+--EXPECT--
+int(33554433)
+bool(true)

ext/hash/tests/hash_pbkdf2_max_length.phpt

@@ -0,0 +1,17 @@
+--TEST--
+Hash: hash_pbkdf2() function : output length of PHP_INT_MAX
+--SKIPIF--
+<?php
+if (PHP_INT_SIZE == 4) die("skip this test is not for 32bit platforms");
+if (getenv("USE_ZEND_ALLOC") === "0") die("skip Zend MM disabled");
+?>
+--INI--
+memory_limit=128M
+--FILE--
+<?php
+
+hash_pbkdf2('md5', 'password', 'salt', 1, PHP_INT_MAX);
+
+?>
+--EXPECTF--
+Fatal error: Allowed memory size of 134217728 bytes exhausted%s(tried to allocate 4611686018427387904 bytes) in %s on line %d

ext/hash/tests/hash_pbkdf2_max_length_raw.phpt

@@ -0,0 +1,17 @@
+--TEST--
+Hash: hash_pbkdf2() function : raw output length of PHP_INT_MAX
+--SKIPIF--
+<?php
+if (PHP_INT_SIZE == 4) die("skip this test is not for 32bit platforms");
+if (getenv("USE_ZEND_ALLOC") === "0") die("skip Zend MM disabled");
+?>
+--INI--
+memory_limit=128M
+--FILE--
+<?php
+
+hash_pbkdf2('md5', 'password', 'salt', 1, PHP_INT_MAX, true);
+
+?>
+--EXPECTF--
+Fatal error: Allowed memory size of 134217728 bytes exhausted%s(tried to allocate 9223372036854775808 bytes) in %s on line %d

ext/intl/grapheme/grapheme_util.cpp

@@ -139,7 +139,11 @@ U_CFUNC int32_t grapheme_strpos_utf16(char *haystack, size_t haystack_len, char
 			ret_pos = -1;
 			goto finish;
 		}
-		ret_pos = last && offset >= 0 ? uhaystack_len : offset_pos;
+		if (last && offset >= 0) {
+			ret_pos = grapheme_count_graphemes(bi, uhaystack, uhaystack_len);
+		} else {
+			ret_pos = grapheme_count_graphemes(bi, uhaystack, offset_pos);
+		}
 		goto finish;
 	}
 

ext/intl/spoofchecker/spoofchecker_main.cpp

@@ -46,6 +46,7 @@ U_CFUNC PHP_METHOD(Spoofchecker, isSuspicious)
 	ret = intl_icu_compat_uspoof_check_utf8(co->uspoof, ZSTR_VAL(text), ZSTR_LEN(text), co->uspoofres, SPOOFCHECKER_ERROR_CODE_P(co));
 
 	if (U_FAILURE(SPOOFCHECKER_ERROR_CODE(co))) {
+		intl_error_set_code(NULL, SPOOFCHECKER_ERROR_CODE(co));
 		php_error_docref(NULL, E_WARNING, "(%d) %s", SPOOFCHECKER_ERROR_CODE(co), u_errorName(SPOOFCHECKER_ERROR_CODE(co)));
 
 		if (intl_icu_compat_uspoof_check_result_mismatch(co->uspoofres, ret, &errmask, SPOOFCHECKER_ERROR_CODE_P(co))) {
@@ -83,6 +84,7 @@ U_CFUNC PHP_METHOD(Spoofchecker, areConfusable)
 		ret = uspoof_areConfusableUTF8(co->uspoof, ZSTR_VAL(s1), (int32_t)ZSTR_LEN(s1), ZSTR_VAL(s2), (int32_t)ZSTR_LEN(s2), SPOOFCHECKER_ERROR_CODE_P(co));
 	}
 	if (U_FAILURE(SPOOFCHECKER_ERROR_CODE(co))) {
+		intl_error_set_code(NULL, SPOOFCHECKER_ERROR_CODE(co));
 		php_error_docref(NULL, E_WARNING, "(%d) %s", SPOOFCHECKER_ERROR_CODE(co), u_errorName(SPOOFCHECKER_ERROR_CODE(co)));
 		RETURN_TRUE;
 	}
@@ -109,6 +111,7 @@ U_CFUNC PHP_METHOD(Spoofchecker, setAllowedLocales)
 	uspoof_setAllowedLocales(co->uspoof, ZSTR_VAL(locales), SPOOFCHECKER_ERROR_CODE_P(co));
 
 	if (U_FAILURE(SPOOFCHECKER_ERROR_CODE(co))) {
+		intl_error_set_code(NULL, SPOOFCHECKER_ERROR_CODE(co));
 		php_error_docref(NULL, E_WARNING, "(%d) %s", SPOOFCHECKER_ERROR_CODE(co), u_errorName(SPOOFCHECKER_ERROR_CODE(co)));
 		return;
 	}
@@ -130,6 +133,7 @@ U_CFUNC PHP_METHOD(Spoofchecker, setChecks)
 	uspoof_setChecks(co->uspoof, checks, SPOOFCHECKER_ERROR_CODE_P(co));
 
 	if (U_FAILURE(SPOOFCHECKER_ERROR_CODE(co))) {
+		intl_error_set_code(NULL, SPOOFCHECKER_ERROR_CODE(co));
 		php_error_docref(NULL, E_WARNING, "(%d) %s", SPOOFCHECKER_ERROR_CODE(co), u_errorName(SPOOFCHECKER_ERROR_CODE(co)));
 	}
 }
@@ -220,6 +224,7 @@ U_CFUNC PHP_METHOD(Spoofchecker, setAllowedChars)
 	efree(upattern);
 
 	if (U_FAILURE(SPOOFCHECKER_ERROR_CODE(co))) {
+		intl_error_set_code(NULL, SPOOFCHECKER_ERROR_CODE(co));
 		php_error_docref(NULL, E_WARNING, "(%d) %s", SPOOFCHECKER_ERROR_CODE(co), u_errorName(SPOOFCHECKER_ERROR_CODE(co)));
 	}
 }
@@ -355,6 +360,7 @@ U_CFUNC PHP_METHOD(Spoofchecker, areBidiConfusable)
 		ret = uspoof_areBidiConfusableUTF8(co->uspoof, (UBiDiDirection)direction, ZSTR_VAL(s1), (int32_t)ZSTR_LEN(s1), ZSTR_VAL(s2), (int32_t)ZSTR_LEN(s2), SPOOFCHECKER_ERROR_CODE_P(co));
 	}
 	if (U_FAILURE(SPOOFCHECKER_ERROR_CODE(co))) {
+		intl_error_set_code(NULL, SPOOFCHECKER_ERROR_CODE(co));
 		php_error_docref(NULL, E_WARNING, "(%d) %s", SPOOFCHECKER_ERROR_CODE(co), u_errorName(SPOOFCHECKER_ERROR_CODE(co)));
 		RETURN_TRUE;
 	}

ext/intl/tests/grapheme_empty_offset_multibyte.phpt

@@ -0,0 +1,34 @@
+--TEST--
+grapheme_strpos() family with empty needle and offset on multi-code-unit graphemes
+--EXTENSIONS--
+intl
+--FILE--
+<?php
+
+var_dump(grapheme_strpos("😀x", ""));
+var_dump(grapheme_strpos("😀x", "", 0));
+var_dump(grapheme_strpos("😀x", "", 1));
+var_dump(grapheme_stripos("😀x", "", 1));
+var_dump(grapheme_strpos("😀x", "", -1));
+var_dump(grapheme_strrpos("😀x", ""));
+var_dump(grapheme_strrpos("😀x", "", 1));
+var_dump(grapheme_strripos("😀x", "", 1));
+var_dump(grapheme_strrpos("😀x", "", -1));
+try {
+    var_dump(grapheme_strpos("😀x", "", 5));
+} catch (Throwable $e) {
+    echo $e::class, ': ', $e->getMessage(), "\n";
+}
+
+?>
+--EXPECT--
+int(0)
+int(0)
+int(1)
+int(1)
+int(1)
+int(2)
+int(2)
+int(2)
+int(1)
+ValueError: grapheme_strpos(): Argument #3 ($offset) must be contained in argument #1 ($haystack)

ext/intl/tests/spoofchecker_setchecks_error_code.phpt

@@ -0,0 +1,18 @@
+--TEST--
+Spoofchecker::setChecks() records the ICU error code
+--EXTENSIONS--
+intl
+--SKIPIF--
+<?php if (!class_exists("Spoofchecker")) print 'skip'; ?>
+--FILE--
+<?php
+
+$s = new Spoofchecker();
+$s->setChecks(1 << 20);
+var_dump(intl_get_error_code(), intl_get_error_message());
+
+?>
+--EXPECTF--
+Warning: Spoofchecker::setChecks(): (1) U_ILLEGAL_ARGUMENT_ERROR in %s on line %d
+int(1)
+string(24) "U_ILLEGAL_ARGUMENT_ERROR"