Consistent return and notices for substr type of functions

ext/iconv/iconv.c

@@ -639,23 +639,38 @@ static php_iconv_err_t _php_iconv_substr(smart_str *pretval,
 	}
 
 	if (len < 0) {
-		if ((len += (total_len - offset)) < 0) {
-			return PHP_ICONV_ERR_SUCCESS;
+		len += (total_len - offset);
+		if (len < 0) {
+			/* Argument 2 is offsetn and argument 1 is source string */
+			php_error_docref(NULL, E_NOTICE, "Argument #2 ($%s) is not contained in argument #1 ($%s)",
+				get_active_function_arg_name(2), get_active_function_arg_name(1));
+			len = 0;
 		}
 	}
 
 	if (offset < 0) {
-		if ((offset += total_len) < 0) {
-			return PHP_ICONV_ERR_SUCCESS;
+		offset += total_len;
+		if (offset < 0) {
+			/* Argument 2 is offsetn and argument 1 is source string */
+			php_error_docref(NULL, E_NOTICE, "Argument #2 ($%s) is not contained in argument #1 ($%s)",
+				get_active_function_arg_name(2), get_active_function_arg_name(1));
+			offset = 0;
 		}
 	}
 
 	if((size_t)len > total_len) {
+		// TODO Emit notice/ValueError
 		len = total_len;
 	}
 
 
 	if ((size_t)offset > total_len) {
+		/* Argument 2 is offsetn and argument 1 is source string */
+		php_error_docref(NULL, E_NOTICE, "Argument #2 ($%s) is not contained in argument #1 ($%s)",
+			get_active_function_arg_name(2), get_active_function_arg_name(1));
+		/* Return an empty string */
+		smart_str_appendl(pretval, "", 0);
+		smart_str_0(pretval);
 		return PHP_ICONV_ERR_SUCCESS;
 	}
 

ext/iconv/tests/iconv_substr.phpt

@@ -40,13 +40,17 @@ bar("This is a test", 0, -100000);
 bar("This is a test", -9, -100000);
 var_dump(iconv("ISO-2022-JP", "EUC-JP", iconv_substr(iconv("EUC-JP", "ISO-2022-JP", "����ˤ��� ISO-2022-JP"), 3, 8, "ISO-2022-JP")));
 ?>
---EXPECT--
+--EXPECTF--
 666768696a6b6c
 666768696a6b6c
 a6a4a8a4aaa4ab
 a4aba4ada4afa4b1a4b3a4b5a4b7
-bool(false)
-bool(false)
+
+Notice: substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
+string(0) ""
+
+Notice: iconv_substr(): Argument #2 ($offset) is not contained in argument #1 ($str) in %s on line %d
+string(0) ""
 string(14) "This is a test"
 string(14) "This is a test"
 string(3) "est"
@@ -56,7 +60,11 @@ string(3) "est"
 string(5) "This "
 string(5) "This "
 bool(false)
+
+Notice: iconv_substr(): Argument #2 ($offset) is not contained in argument #1 ($str) in %s on line %d
+string(0) ""
 bool(false)
-bool(false)
-bool(false)
+
+Notice: iconv_substr(): Argument #2 ($offset) is not contained in argument #1 ($str) in %s on line %d
+string(0) ""
 string(10) "���� ISO-2"

ext/mbstring/mbstring.c

@@ -2160,6 +2160,8 @@ PHP_FUNCTION(mb_substr)
 	} else if (-from < mblen) {
 		real_from = mblen + from;
 	} else {
+		php_error_docref(NULL, E_NOTICE, "Argument #2 ($%s) is not contained in argument #1 ($%s)",
+			get_active_function_arg_name(2), get_active_function_arg_name(1));
 		real_from = 0;
 	}
 
@@ -2173,6 +2175,7 @@ PHP_FUNCTION(mb_substr)
 	} else if (real_from < mblen && -len < mblen - real_from) {
 		real_len = (mblen - real_from) + len;
 	} else {
+		// TODO Emit notice?
 		real_len = 0;
 	}
 
@@ -2218,6 +2221,8 @@ PHP_FUNCTION(mb_strcut)
 	if (from < 0) {
 		from = string.len + from;
 		if (from < 0) {
+			php_error_docref(NULL, E_NOTICE, "Argument #2 ($%s) is not contained in argument #1 ($%s)",
+				get_active_function_arg_name(2), get_active_function_arg_name(1));
 			from = 0;
 		}
 	}
@@ -2233,6 +2238,8 @@ PHP_FUNCTION(mb_strcut)
 	}
 
 	if (from > string.len) {
+		php_error_docref(NULL, E_NOTICE, "Argument #2 ($%s) is not contained in argument #1 ($%s)",
+			get_active_function_arg_name(2), get_active_function_arg_name(1));
 		RETURN_EMPTY_STRING();
 	}
 

ext/mbstring/tests/bug49354.phpt

@@ -13,10 +13,12 @@ var_dump(mb_strcut($crap, 12, 100, 'UTF-8'));
 var_dump(mb_strcut($crap, 13, 100, 'UTF-8'));
 
 ?>
---EXPECT--
+--EXPECTF--
 string(12) "AåBäCöDü"
 string(11) "åBäCöDü"
 string(11) "åBäCöDü"
 string(9) "BäCöDü"
 string(0) ""
+
+Notice: mb_strcut(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(0) ""

ext/mbstring/tests/mb_strcut.phpt

@@ -55,12 +55,16 @@ print MBStringChars(mb_strcut($utf16le, 1, 3, 'UTF-16LE'), 'UTF-16LE') . "\n";
 print MBStringChars(mb_strcut($utf16le, 1, 4, 'UTF-16LE'), 'UTF-16LE') . "\n";
 
 ?>
---EXPECT--
+--EXPECTF--
 == EUC-JP ==
 [a4ce cab8]
 [a4b3 a4ce]
 [30 31 32 33 a4b3 a4ce cab8 bbfa cef3 a4cf c6fc cbdc b8ec a4c7 a4b9 a1a3 45 55 43 2d 4a 50 a4f2 bbc8 a4c3 a4c6 a4a4 a4de a4b9 a1a3 c6fc cbdc b8ec a4cf cccc c5dd bdad a4a4 a1a3]
+
+Notice: mb_strcut(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 OK
+
+Notice: mb_strcut(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 OK
 == UTF-8 ==
 []

ext/mbstring/tests/mb_substr.phpt

@@ -24,8 +24,10 @@ $str = mb_substr($euc_jp, -100, 10,'EUC-JP');
 ($str !== "") ? print "4 OK: ".bin2hex($str)."\n" : print "NG: ".bin2hex($str)."\n";
 
 ?>
---EXPECT--
+--EXPECTF--
 1: c6fccbdcb8eca4c7a4b9a1a34555432d
 2: 30313233a4b3a4cecab8bbfacef3a4cfc6fccbdcb8eca4c7a4b9a1a34555432d4a50a4f2bbc8a4c3a4c6a4a4a4dea4b9a1a3c6fccbdcb8eca4cfccccc5ddbdada4a4a1a3
 3 OK
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 4 OK: 30313233a4b3a4cecab8bbfacef3a4cf

ext/mbstring/tests/mb_substr_variation4.phpt

@@ -44,31 +44,47 @@ for ($i = -60; $i <= 60; $i += 10) {
 
 echo "Done";
 ?>
---EXPECT--
+--EXPECTF--
 *** Testing mb_substr() : usage variations ***
 
 **-- Offset is: -60 --**
 -- ASCII String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(8) "K0lzIA=="
 --Multibyte String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(16) "5pel5pys6Kqe44OG"
 
 **-- Offset is: -50 --**
 -- ASCII String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(8) "K0lzIA=="
 --Multibyte String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(16) "5pel5pys6Kqe44OG"
 
 **-- Offset is: -40 --**
 -- ASCII String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(8) "K0lzIA=="
 --Multibyte String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(16) "5pel5pys6Kqe44OG"
 
 **-- Offset is: -30 --**
 -- ASCII String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(8) "K0lzIA=="
 --Multibyte String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(16) "5pel5pys6Kqe44OG"
 
 **-- Offset is: -20 --**

ext/mbstring/tests/mb_substr_variation6.phpt

@@ -49,31 +49,47 @@ for ($i = -60; $i <= 60; $i += 10) {
 
 echo "Done";
 ?>
---EXPECT--
+--EXPECTF--
 *** Testing mb_substr() : usage variations ***
 
 **-- Offset is: -60 --**
 -- ASCII String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(8) "2b497320"
 --Multibyte String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(24) "e697a5e69cace8aa9ee38386"
 
 **-- Offset is: -50 --**
 -- ASCII String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(8) "2b497320"
 --Multibyte String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(24) "e697a5e69cace8aa9ee38386"
 
 **-- Offset is: -40 --**
 -- ASCII String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(8) "2b497320"
 --Multibyte String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(24) "e697a5e69cace8aa9ee38386"
 
 **-- Offset is: -30 --**
 -- ASCII String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(8) "2b497320"
 --Multibyte String --
+
+Notice: mb_substr(): Argument #2 ($start) is not contained in argument #1 ($str) in %s on line %d
 string(24) "e697a5e69cace8aa9ee38386"
 
 **-- Offset is: -20 --**

ext/opcache/Optimizer/sccp.c

@@ -968,12 +968,6 @@ static inline int ct_eval_func_call(
 				return FAILURE;
 			}
 			/* pass */
-		} else if (zend_string_equals_literal(name, "substr")) {
-			if (Z_TYPE_P(args[0]) != IS_STRING
-					|| Z_TYPE_P(args[1]) != IS_LONG) {
-				return FAILURE;
-			}
-			/* pass */
 		} else if (zend_string_equals_literal(name, "pow")) {
 			if ((Z_TYPE_P(args[0]) != IS_LONG && Z_TYPE_P(args[0]) != IS_DOUBLE)
 					|| (Z_TYPE_P(args[1]) != IS_LONG && Z_TYPE_P(args[1]) != IS_DOUBLE)) {

ext/standard/basic_functions.stub.php

@@ -543,9 +543,9 @@ function bin2hex(string $data): string {}
 
 function hex2bin(string $data): string|false {}
 
-function strspn(string $str, string $mask, int $start = 0, ?int $len = null): int|false {}
+function strspn(string $str, string $mask, int $start = 0, ?int $len = null): int {}
 
-function strcspn(string $str, string $mask, int $start = 0, ?int $len = null): int|false {}
+function strcspn(string $str, string $mask, int $start = 0, ?int $len = null): int {}
 
 #if HAVE_NL_LANGINFO
 function nl_langinfo(int $item): string|false {}
@@ -610,7 +610,7 @@ function chunk_split(string $str, int $chunklen = 76, string $ending = "\r\n"):
 
 function substr(string $str, int $start, ?int $length = null): string|false {}
 
-function substr_replace(array|string $str, array|string $replace, array|int $start, array|int|null $length = null): string|array|false {}
+function substr_replace(array|string $str, array|string $replace, array|int $start, array|int|null $length = null): string|array {}
 
 function quotemeta(string $str): string {}
 
@@ -688,7 +688,7 @@ function str_split(string $str, int $split_length = 1): array {}
 
 function strpbrk(string $haystack, string $char_list): string|false {}
 
-function substr_compare(string $main_str, string $str, int $offset, ?int $length = null, bool $case_insensitivity = false): int|false {}
+function substr_compare(string $main_str, string $str, int $offset, ?int $length = null, bool $case_insensitivity = false): int {}
 
 function utf8_encode(string $data): string {}
 

ext/standard/basic_functions_arginfo.h

@@ -1,5 +1,5 @@
 /* This is a generated file, edit the .stub.php file instead.
- * Stub hash: 0224dc521c4a8bd49fbcfd26cddb01a2e571cf74 */
+ * Stub hash: 3937e2a532217fd78bc2ec94014a86a3e570d001 */
 
 ZEND_BEGIN_ARG_WITH_RETURN_TYPE_INFO_EX(arginfo_set_time_limit, 0, 1, _IS_BOOL, 0)
 	ZEND_ARG_TYPE_INFO(0, seconds, IS_LONG, 0)
@@ -807,7 +807,7 @@ ZEND_BEGIN_ARG_WITH_RETURN_TYPE_MASK_EX(arginfo_hex2bin, 0, 1, MAY_BE_STRING|MAY
 	ZEND_ARG_TYPE_INFO(0, data, IS_STRING, 0)
 ZEND_END_ARG_INFO()
 
-ZEND_BEGIN_ARG_WITH_RETURN_TYPE_MASK_EX(arginfo_strspn, 0, 2, MAY_BE_LONG|MAY_BE_FALSE)
+ZEND_BEGIN_ARG_WITH_RETURN_TYPE_INFO_EX(arginfo_strspn, 0, 2, IS_LONG, 0)
 	ZEND_ARG_TYPE_INFO(0, str, IS_STRING, 0)
 	ZEND_ARG_TYPE_INFO(0, mask, IS_STRING, 0)
 	ZEND_ARG_TYPE_INFO_WITH_DEFAULT_VALUE(0, start, IS_LONG, 0, "0")
@@ -930,7 +930,7 @@ ZEND_BEGIN_ARG_WITH_RETURN_TYPE_MASK_EX(arginfo_substr, 0, 2, MAY_BE_STRING|MAY_
 	ZEND_ARG_TYPE_INFO_WITH_DEFAULT_VALUE(0, length, IS_LONG, 1, "null")
 ZEND_END_ARG_INFO()
 
-ZEND_BEGIN_ARG_WITH_RETURN_TYPE_MASK_EX(arginfo_substr_replace, 0, 3, MAY_BE_STRING|MAY_BE_ARRAY|MAY_BE_FALSE)
+ZEND_BEGIN_ARG_WITH_RETURN_TYPE_MASK_EX(arginfo_substr_replace, 0, 3, MAY_BE_STRING|MAY_BE_ARRAY)
 	ZEND_ARG_TYPE_MASK(0, str, MAY_BE_ARRAY|MAY_BE_STRING, NULL)
 	ZEND_ARG_TYPE_MASK(0, replace, MAY_BE_ARRAY|MAY_BE_STRING, NULL)
 	ZEND_ARG_TYPE_MASK(0, start, MAY_BE_ARRAY|MAY_BE_LONG, NULL)
@@ -1082,7 +1082,7 @@ ZEND_BEGIN_ARG_WITH_RETURN_TYPE_MASK_EX(arginfo_strpbrk, 0, 2, MAY_BE_STRING|MAY
 	ZEND_ARG_TYPE_INFO(0, char_list, IS_STRING, 0)
 ZEND_END_ARG_INFO()
 
-ZEND_BEGIN_ARG_WITH_RETURN_TYPE_MASK_EX(arginfo_substr_compare, 0, 3, MAY_BE_LONG|MAY_BE_FALSE)
+ZEND_BEGIN_ARG_WITH_RETURN_TYPE_INFO_EX(arginfo_substr_compare, 0, 3, IS_LONG, 0)
 	ZEND_ARG_TYPE_INFO(0, main_str, IS_STRING, 0)
 	ZEND_ARG_TYPE_INFO(0, str, IS_STRING, 0)
 	ZEND_ARG_TYPE_INFO(0, offset, IS_LONG, 0)