Promote some warnings in MBString Regexes

Closes GH-5341

ext/mbstring/php_mbregex.c

@@ -591,8 +591,8 @@ static size_t _php_mb_regex_get_option_string(char *str, size_t len, OnigOptionT
 /* }}} */
 
 /* {{{ _php_mb_regex_init_options */
-static void
-_php_mb_regex_init_options(const char *parg, size_t narg, OnigOptionType *option, OnigSyntaxType **syntax, int *eval)
+static bool _php_mb_regex_init_options(const char *parg, size_t narg, OnigOptionType *option,
+	OnigSyntaxType **syntax)
 {
 	size_t n;
 	char c;
@@ -650,15 +650,14 @@ _php_mb_regex_init_options(const char *parg, size_t narg, OnigOptionType *option
 				case 'd':
 					*syntax = ONIG_SYNTAX_POSIX_EXTENDED;
 					break;
-				case 'e':
-					if (eval != NULL) *eval = 1;
-					break;
 				default:
-					break;
+					zend_value_error("Option \"%c\" is not supported", c);
+					return false;
 			}
 		}
 		if (option != NULL) *option|=optm;
 	}
+	return true;
 }
 /* }}} */
 
@@ -900,6 +899,11 @@ static void _php_mb_regex_ereg_exec(INTERNAL_FUNCTION_PARAMETERS, int icase)
 		RETURN_THROWS();
 	}
 
+	if (arg_pattern_len == 0) {
+		zend_argument_value_error(1, "must not be empty");
+		RETURN_THROWS();
+	}
+
 	if (array != NULL) {
 		array = zend_try_array_init(array);
 		if (!array) {
@@ -920,12 +924,6 @@ static void _php_mb_regex_ereg_exec(INTERNAL_FUNCTION_PARAMETERS, int icase)
 		options |= ONIG_OPTION_IGNORECASE;
 	}
 
-	if (arg_pattern_len == 0) {
-		php_error_docref(NULL, E_WARNING, "Empty pattern");
-		RETVAL_FALSE;
-		goto out;
-	}
-
 	re = php_mbregex_compile_pattern(arg_pattern, arg_pattern_len, options, MBREX(regex_default_syntax));
 	if (re == NULL) {
 		RETVAL_FALSE;
@@ -1007,7 +1005,7 @@ static void _php_mb_regex_ereg_replace_exec(INTERNAL_FUNCTION_PARAMETERS, OnigOp
 	smart_str out_buf = {0};
 	smart_str eval_buf = {0};
 	smart_str *pbuf;
-	int err, eval, n;
+	int err, n;
 	OnigUChar *pos;
 	OnigUChar *string_lim;
 	char *description = NULL;
@@ -1015,7 +1013,6 @@ static void _php_mb_regex_ereg_replace_exec(INTERNAL_FUNCTION_PARAMETERS, OnigOp
 	const mbfl_encoding *enc = php_mb_regex_get_mbctype_encoding();
 	ZEND_ASSERT(enc != NULL);
 
-	eval = 0;
 	{
 		char *option_str = NULL;
 		size_t option_str_len = 0;
@@ -1043,20 +1040,15 @@ static void _php_mb_regex_ereg_replace_exec(INTERNAL_FUNCTION_PARAMETERS, OnigOp
 		}
 
 		if (option_str != NULL) {
-			_php_mb_regex_init_options(option_str, option_str_len, &options, &syntax, &eval);
+			/* Initialize option and in case of failure it means there is a value error */
+			if (!_php_mb_regex_init_options(option_str, option_str_len, &options, &syntax)) {
+				RETURN_THROWS();
+			}
 		} else {
 			options |= MBREX(regex_default_options);
 			syntax = MBREX(regex_default_syntax);
 		}
 	}
-	if (eval) {
-		if (is_callable) {
-			php_error_docref(NULL, E_WARNING, "Option 'e' cannot be used with replacement callback");
-		} else {
-			php_error_docref(NULL, E_WARNING, "The 'e' option is no longer supported, use mb_ereg_replace_callback instead");
-		}
-		RETURN_FALSE;
-	}
 
 	/* create regex pattern buffer */
 	re = php_mbregex_compile_pattern(arg_pattern, arg_pattern_len, options, syntax);
@@ -1122,7 +1114,9 @@ static void _php_mb_regex_ereg_replace_exec(INTERNAL_FUNCTION_PARAMETERS, OnigOp
 					zval_ptr_dtor(&retval);
 				} else {
 					if (!EG(exception)) {
-						php_error_docref(NULL, E_WARNING, "Unable to call custom replacement function");
+						zend_throw_error(NULL, "Unable to call custom replacement function");
+						zval_ptr_dtor(&subpats);
+						RETURN_THROWS();
 					}
 				}
 				zval_ptr_dtor(&subpats);
@@ -1251,6 +1245,7 @@ PHP_FUNCTION(mb_split)
 	onig_region_free(regs, 1);
 
 	/* see if we encountered an error */
+	// ToDo investigate if this can actually/should happen ...
 	if (err <= -2) {
 		OnigUChar err_str[ONIG_MAX_ERROR_MESSAGE_LEN];
 		onig_error_code_to_str(err_str, err);
@@ -1295,7 +1290,9 @@ PHP_FUNCTION(mb_ereg_match)
 		}
 
 		if (option_str != NULL) {
-			_php_mb_regex_init_options(option_str, option_str_len, &option, &syntax, NULL);
+			if(!_php_mb_regex_init_options(option_str, option_str_len, &option, &syntax)) {
+				RETURN_THROWS();
+			}
 		} else {
 			option |= MBREX(regex_default_options);
 			syntax = MBREX(regex_default_syntax);
@@ -1348,7 +1345,7 @@ static void _php_mb_regex_ereg_search_exec(INTERNAL_FUNCTION_PARAMETERS, int mod
 	}
 
 	if (arg_options) {
-		_php_mb_regex_init_options(arg_options, arg_options_len, &option, &syntax, NULL);
+		_php_mb_regex_init_options(arg_options, arg_options_len, &option, &syntax);
 	} else {
 		option |= MBREX(regex_default_options);
 		syntax = MBREX(regex_default_syntax);
@@ -1375,13 +1372,13 @@ static void _php_mb_regex_ereg_search_exec(INTERNAL_FUNCTION_PARAMETERS, int mod
 	}
 
 	if (MBREX(search_re) == NULL) {
-		php_error_docref(NULL, E_WARNING, "No regex given");
-		RETURN_FALSE;
+		zend_throw_error(NULL, "No pattern was provided");
+		RETURN_THROWS();
 	}
 
 	if (str == NULL) {
-		php_error_docref(NULL, E_WARNING, "No string given");
-		RETURN_FALSE;
+		zend_throw_error(NULL, "No string was provided");
+		RETURN_THROWS();
 	}
 
 	MBREX(search_regs) = onig_region_new();
@@ -1480,13 +1477,13 @@ PHP_FUNCTION(mb_ereg_search_init)
 	}
 
 	if (arg_pattern && arg_pattern_len == 0) {
-		php_error_docref(NULL, E_WARNING, "Empty pattern");
-		RETURN_FALSE;
+		zend_argument_value_error(2, "must not be empty");
+		RETURN_THROWS();
 	}
 
 	if (arg_options) {
 		option = 0;
-		_php_mb_regex_init_options(arg_options, arg_options_len, &option, &syntax, NULL);
+		_php_mb_regex_init_options(arg_options, arg_options_len, &option, &syntax);
 	} else {
 		option = MBREX(regex_default_options);
 		syntax = MBREX(regex_default_syntax);
@@ -1557,6 +1554,7 @@ PHP_FUNCTION(mb_ereg_search_getregs)
 			onig_foreach_name(MBREX(search_re), mb_regex_groups_iter, &args);
 		}
 	} else {
+		// TODO This seems to be some logical error, promote to Error
 		RETVAL_FALSE;
 	}
 }
@@ -1588,12 +1586,12 @@ PHP_FUNCTION(mb_ereg_search_setpos)
 	}
 
 	if (position < 0 || (!Z_ISUNDEF(MBREX(search_str)) && Z_TYPE(MBREX(search_str)) == IS_STRING && (size_t)position > Z_STRLEN(MBREX(search_str)))) {
-		php_error_docref(NULL, E_WARNING, "Position is out of range");
-		MBREX(search_pos) = 0;
-		RETURN_FALSE;
+		zend_argument_value_error(1, "is out of range");
+		RETURN_THROWS();
 	}
 
 	MBREX(search_pos) = position;
+	// TODO Return void
 	RETURN_TRUE;
 }
 /* }}} */
@@ -1628,7 +1626,9 @@ PHP_FUNCTION(mb_regex_set_options)
 	if (string != NULL) {
 		opt = 0;
 		syntax = NULL;
-		_php_mb_regex_init_options(string, string_len, &opt, &syntax, NULL);
+		if(!_php_mb_regex_init_options(string, string_len, &opt, &syntax)) {
+			RETURN_THROWS();
+		}
 		_php_mb_regex_set_options(opt, syntax, &prev_opt, &prev_syntax);
 		opt = prev_opt;
 		syntax = prev_syntax;

ext/mbstring/tests/bug43994.phpt

@@ -25,106 +25,76 @@ foreach($inputs as $input) {
     }
     echo "\n-- Iteration $iterator --\n";
     echo "Without \$regs arg:\n";
-    var_dump( mb_ereg($input, 'hello, world') );
+    try {
+        var_dump( mb_ereg($input, 'hello, world') );
+    } catch (\ValueError $e) {
+        echo $e->getMessage() . \PHP_EOL;
+    }
+
     echo "With \$regs arg:\n";
-    var_dump(mb_ereg($input, 'hello, world', $mb_regs));
+    try {
+        var_dump(mb_ereg($input, 'hello, world', $mb_regs));
+    } catch (\ValueError $e) {
+        echo $e->getMessage() . \PHP_EOL;
+    }
+
     var_dump($mb_regs);
     $iterator++;
 };
 ?>
---EXPECTF--
+--EXPECT--
 -- Iteration 1 --
 Without $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
+mb_ereg(): Argument #1 ($pattern) must not be empty
 With $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
-array(0) {
-}
+mb_ereg(): Argument #1 ($pattern) must not be empty
+NULL
 
 -- Iteration 2 --
 Without $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
+mb_ereg(): Argument #1 ($pattern) must not be empty
 With $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
-array(0) {
-}
+mb_ereg(): Argument #1 ($pattern) must not be empty
+NULL
 
 -- Iteration 3 --
 Without $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
+mb_ereg(): Argument #1 ($pattern) must not be empty
 With $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
-array(0) {
-}
+mb_ereg(): Argument #1 ($pattern) must not be empty
+NULL
 
 -- Iteration 4 --
 Without $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
+mb_ereg(): Argument #1 ($pattern) must not be empty
 With $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
-array(0) {
-}
+mb_ereg(): Argument #1 ($pattern) must not be empty
+NULL
 
 -- Iteration 5 --
 Without $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
+mb_ereg(): Argument #1 ($pattern) must not be empty
 With $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
-array(0) {
-}
+mb_ereg(): Argument #1 ($pattern) must not be empty
+NULL
 
 -- Iteration 6 --
 Without $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
+mb_ereg(): Argument #1 ($pattern) must not be empty
 With $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
-array(0) {
-}
+mb_ereg(): Argument #1 ($pattern) must not be empty
+NULL
 
 -- Iteration 7 --
 Without $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
+mb_ereg(): Argument #1 ($pattern) must not be empty
 With $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
-array(0) {
-}
+mb_ereg(): Argument #1 ($pattern) must not be empty
+NULL
 
 -- Iteration 8 --
 Without $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
+mb_ereg(): Argument #1 ($pattern) must not be empty
 With $regs arg:
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
-array(0) {
-}
+mb_ereg(): Argument #1 ($pattern) must not be empty
+NULL

ext/mbstring/tests/bug69151.phpt

@@ -8,13 +8,14 @@ if (!function_exists('mb_ereg')) die('skip mbregex support not available');
 --FILE--
 <?php
 $str = "\x80";
-var_dump(
-    false === mb_eregi('.', $str, $matches),
-    [] === $matches,
-    NULL === mb_ereg_replace('.', "\\0", $str),
-    false === mb_ereg_search_init("\x80", '.'),
-    false === mb_ereg_search()
-);
+
+var_dump(false === mb_eregi('.', $str, $matches));
+var_dump([] === $matches);
+
+var_dump(NULL === mb_ereg_replace('.', "\\0", $str));
+
+var_dump(false === mb_ereg_search_init("\x80", '.'));
+var_dump(false === mb_ereg_search());
 ?>
 --EXPECT--
 bool(true)

ext/mbstring/tests/bug72164.phpt

@@ -10,9 +10,13 @@ if (!function_exists('mb_ereg')) die('skip mbregex support not available');
 $var0 = "e";
 $var2 = "";
 $var3 = NULL;
-$var8 = mb_ereg_replace($var2,$var3,$var3,$var0);
-var_dump($var8);
+try {
+    $var8 = mb_ereg_replace($var2,$var3,$var3,$var0);
+    var_dump($var8);
+} catch (\ValueError $e) {
+    echo $e->getMessage() . \PHP_EOL;
+}
+
 ?>
---EXPECTF--
-Warning: mb_ereg_replace(): The 'e' option is no longer supported, use mb_ereg_replace_callback instead in %s on line %d
-bool(false)
+--EXPECT--
+Option "e" is not supported

ext/mbstring/tests/bug72399.phpt

@@ -8,8 +8,17 @@ if (!function_exists('mb_ereg')) die('skip mbregex support not available');
 --FILE--
 <?php
 $var5 = mb_ereg_search_init("","2");
+var_dump($var5);
 $var6 = mb_eregi_replace("2","","");
-$var13 = mb_ereg_search_pos();
+var_dump($var6);
+
+try {
+    $var13 = mb_ereg_search_pos();
+} catch (\Error $e) {
+    echo $e->getMessage() . \PHP_EOL;
+}
 ?>
---EXPECTF--
-Warning: mb_ereg_search_pos(): No regex given in %sbug72399.php on line %d
+--EXPECT--
+bool(true)
+string(0) ""
+No pattern was provided

ext/mbstring/tests/bug76999.phpt

@@ -11,12 +11,16 @@ mb_regex_set_options("pr");
 var_dump(mb_regex_set_options("m"));
 var_dump(mb_regex_set_options("mdi"));
 var_dump(mb_regex_set_options("m"));
-var_dump(mb_regex_set_options("a"));
+try {
+    var_dump(mb_regex_set_options("a"));
+} catch (\ValueError $e) {
+    echo $e->getMessage() . \PHP_EOL;
+}
 var_dump(mb_regex_set_options());
 ?>
 --EXPECT--
 string(2) "pr"
 string(2) "mr"
 string(3) "imd"
+Option "a" is not supported
 string(2) "mr"
-string(1) "r"

ext/mbstring/tests/bug77418.phpt

@@ -8,6 +8,7 @@ if (!function_exists('mb_split')) die('skip mb_split() not available');
 --FILE--
 <?php
 mb_regex_encoding("UTF-32");
+
 var_dump(mb_split("\x00\x00\x00\x5c\x00\x00\x00B","000000000000000000000000000000"));
 ?>
 --EXPECT--

ext/mbstring/tests/empty_pattern.phpt

@@ -8,12 +8,21 @@ if (!function_exists('mb_ereg')) die('skip mbregex support not available');
 --FILE--
 <?php
 
-mb_ereg_search_init("","","");
+try {
+    mb_ereg_search_init("","","");
+} catch (\ValueError $e) {
+    echo $e->getMessage() . \PHP_EOL;
+}
+
 mb_split("","");
-mb_ereg_search_regs();
+
+try {
+    mb_ereg_search_regs();
+} catch (\Error $e) {
+    echo $e->getMessage() . \PHP_EOL;
+}
 
 ?>
---EXPECTF--
-Warning: mb_ereg_search_init(): Empty pattern in %s on line %d
-
-Warning: mb_ereg_search_regs(): No regex given in %s on line %d
+--EXPECT--
+mb_ereg_search_init(): Argument #2 ($pattern) must not be empty
+No pattern was provided

ext/mbstring/tests/mb_ereg1.phpt

@@ -16,13 +16,13 @@ $a = array(
 foreach ($a as $args) {
     try {
         var_dump(mb_ereg($args[0], $args[1], $args[2]));
-    } catch (TypeError $e) {
-        echo $e->getMessage(), "\n";
+    } catch (\TypeError|\ValueError $e) {
+        echo get_class($e) . ': ' . $e->getMessage() . \PHP_EOL;
     }
     var_dump($args);
 }
 ?>
---EXPECTF--
+--EXPECT--
 bool(false)
 array(3) {
   [0]=>
@@ -33,19 +33,16 @@ array(3) {
   array(0) {
   }
 }
-
-Warning: mb_ereg(): Empty pattern in %s on line %d
-bool(false)
+ValueError: mb_ereg(): Argument #1 ($pattern) must not be empty
 array(3) {
   [0]=>
   string(0) ""
   [1]=>
   string(0) ""
   [2]=>
-  array(0) {
-  }
+  string(0) ""
 }
-mb_ereg(): Argument #1 ($pattern) must be of type string, array given
+TypeError: mb_ereg(): Argument #1 ($pattern) must be of type string, array given
 array(3) {
   [0]=>
   array(0) {
@@ -55,7 +52,7 @@ array(3) {
   [2]=>
   string(0) ""
 }
-mb_ereg(): Argument #2 ($string) must be of type string, array given
+TypeError: mb_ereg(): Argument #2 ($string) must be of type string, array given
 array(3) {
   [0]=>
   int(1)

ext/mbstring/tests/mb_ereg_search_setpos.phpt

@@ -11,22 +11,32 @@ mb_regex_encoding('iso-8859-1');
 $test_str = 'Iñtërnâtiônàlizætiøn'; // Length = 20
 
 var_dump(mb_ereg_search_setpos(50)); // OK
-var_dump(mb_ereg_search_setpos(-1)); // Error
+try {
+    var_dump(mb_ereg_search_setpos(-1)); // Error
+} catch (\ValueError $e) {
+    echo $e->getMessage() . \PHP_EOL;
+}
 
 mb_ereg_search_init($test_str);
 
 $positions = array( 5, 20, 21, 25, 0, -5, -20, -30);
 foreach($positions as $pos) {
     echo("\n* Position: $pos :\n");
-    var_dump(mb_ereg_search_setpos($pos));
-    var_dump(mb_ereg_search_getpos());
+    try {
+        var_dump(mb_ereg_search_setpos($pos));
+    } catch (\ValueError $e) {
+        echo $e->getMessage() . \PHP_EOL;
+    }
+    try {
+        var_dump(mb_ereg_search_getpos());
+    } catch (\ValueError $e) {
+        echo $e->getMessage() . \PHP_EOL;
+    }
 }
 ?>
---EXPECTF--
+--EXPECT--
 bool(true)
-
-Warning: mb_ereg_search_setpos(): Position is out of range in %s on line %d
-bool(false)
+mb_ereg_search_setpos(): Argument #1 ($position) is out of range
 
 * Position: 5 :
 bool(true)
@@ -37,16 +47,12 @@ bool(true)
 int(20)
 
 * Position: 21 :
-
-Warning: mb_ereg_search_setpos(): Position is out of range in %s on line %d
-bool(false)
-int(0)
+mb_ereg_search_setpos(): Argument #1 ($position) is out of range
+int(20)
 
 * Position: 25 :
-
-Warning: mb_ereg_search_setpos(): Position is out of range in %s on line %d
-bool(false)
-int(0)
+mb_ereg_search_setpos(): Argument #1 ($position) is out of range
+int(20)
 
 * Position: 0 :
 bool(true)
@@ -61,7 +67,5 @@ bool(true)
 int(0)
 
 * Position: -30 :
-
-Warning: mb_ereg_search_setpos(): Position is out of range in %s on line %d
-bool(false)
+mb_ereg_search_setpos(): Argument #1 ($position) is out of range
 int(0)