Fix a few regex class parsing usecases

resources/RegexGrammar.pp

@@ -42,14 +42,16 @@
 //
 
 // Character classes.
+// tokens suffixed with "fc_" are the same as without such suffix but followed by "class:_class"
+%token  negative_class_fc_       \[\^(?=\])         -> class_fc
+%token  class_fc_                \[(?=\])           -> class_fc
+%token class_fc:_class           \]                 -> class
 %token  negative_class_          \[\^               -> class
 %token  class_                   \[                 -> class
 %token class:posix_class         \[:\^?[a-z]+:\]
 %token class:class_              \[
-%token class:_class_literal      (?<=[^\\]\[|[^\\]\[\^)\]
 %token class:_class              \]                 -> default
 %token class:range               \-
-%token class:escaped_end_class   \\\]
 // taken over from literals but class:character has \b support on top (backspace in character classes)
 %token class:character           \\([aefnrtb]|c[\x00-\x7f])
 %token class:dynamic_character   \\([0-7]{3}|x[0-9a-zA-Z]{2}|x{[0-9a-zA-Z]+})
@@ -58,7 +60,8 @@
 
 // Internal options.
 // See https://www.regular-expressions.info/refmodifiers.html
-%token  internal_option          \(\?([imsxnJUX^]|xx)?-?([imsxnJUX^]|xx)\)
+// and https://www.php.net/manual/en/regexp.reference.internal-options.php
+%token  internal_option          \(\?[imsxnJUX^]*-?[imsxnJUX^]+\)
 
 // Lookahead and lookbehind assertions.
 %token  lookahead_               \(\?=
@@ -88,7 +91,7 @@
 %token  nc:_named_capturing      >                  -> default
 %token  nc:capturing_name        .+?(?=(?<!\\)>)
 %token  non_capturing_           \(\?:
-%token  non_capturing_internal_option \(\?([imsxnJUX^]|xx)?-?([imsxnJUX^]|xx):
+%token  non_capturing_internal_option \(\?[imsxnJUX^]*-?[imsxnJUX^]+:
 %token  non_capturing_reset_     \(\?\|
 %token  atomic_group_            \(\?>
 %token  capturing_               \(
@@ -177,10 +180,14 @@
 
 #class:
     (
-        ::negative_class_:: #negativeclass
+        ::negative_class_fc_:: #negativeclass
+        <_class>
+      | ::class_fc_::
+        <_class>
+      | ::negative_class_:: #negativeclass
       | ::class_::
     )
-    ( <range> | <_class_literal> )? ( <posix_class> | <class_> | range() | literal() | <escaped_end_class> )* <range>?
+    <range>? ( <posix_class> | <class_> | range() <range>? | literal() )* <range>?
     ::_class::
 
 #range:

src/Type/Regex/RegexGroupParser.php

@@ -525,11 +525,11 @@ private function getLiteralValue(TreeNode $node, ?array &$onlyLiterals, bool $ap
 
 		if (
 			in_array($token, [
-				'literal', 'escaped_end_class',
+				'literal',
 				// literal "-" in front/back of a character class like '[-a-z]' or '[abc-]', not forming a range
 				'range',
 				// literal "[" or "]" inside character classes '[[]' or '[]]'
-				'class_', '_class_literal',
+				'class_', '_class',
 			], true)
 		) {
 			if (str_contains($patternModifiers, 'x') && trim($value) === '') {
@@ -544,7 +544,6 @@ private function getLiteralValue(TreeNode $node, ?array &$onlyLiterals, bool $ap
 
 			if (
 				$appendLiterals
-				&& in_array($token, ['literal', 'range', 'class_', '_class_literal'], true)
 				&& $onlyLiterals !== null
 				&& (!in_array($value, ['.'], true) || $isEscaped || $inCharacterClass)
 			) {

tests/PHPStan/Analyser/nsrt/preg_match_shapes.php

@@ -467,6 +467,9 @@ function bug11323(string $s): void {
 	if (preg_match('{([-\p{L}[\]*|\x03\a\b+?{}(?:)-]+[^[:digit:]?{}a-z0-9#-k]+)(a-z)}', $s, $matches)) {
 		assertType("array{string, non-falsy-string, 'a-z'}", $matches);
 	}
+	if (preg_match('{(\d+)(?i)insensitive((?xs-i)case SENSITIVE here.+and dot matches new lines)}', $s, $matches)) {
+		assertType('array{string, numeric-string, non-falsy-string}', $matches);
+	}
 	if (preg_match('{(\d+)(?i)insensitive((?x-i)case SENSITIVE here(?i:insensitive non-capturing group))}', $s, $matches)) {
 		assertType('array{string, numeric-string, non-falsy-string}', $matches);
 	}
@@ -778,3 +781,121 @@ function testLtrimDelimiter (string $string): void {
 		assertType("array{string, 'x'}", $matches);
 	}
 }
+
+function testUnescapeBackslash (string $string): void {
+	if (preg_match(<<<'EOD'
+		~(\[)~
+		EOD, $string, $matches)) {
+		assertType("array{string, '['}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		~(\d)~
+		EOD, $string, $matches)) {
+		assertType("array{string, numeric-string}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		~(\\d)~
+		EOD, $string, $matches)) {
+		assertType("array{string, '\\\d'}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		~(\\\d)~
+		EOD, $string, $matches)) {
+		assertType("array{string, non-falsy-string}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		~(\\\\d)~
+		EOD, $string, $matches)) {
+		assertType("array{string, '\\\\\\\d'}", $matches);
+	}
+}
+
+function testEscapedDelimiter (string $string): void {
+	if (preg_match(<<<'EOD'
+		/(\/)/
+		EOD, $string, $matches)) {
+		assertType("array{string, '/'}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		~(\~)~
+		EOD, $string, $matches)) {
+		assertType("array{string, '~'}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		~(\[2])~
+		EOD, $string, $matches)) {
+		assertType("array{string, '[2]'}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		[(\[2\])]
+		EOD, $string, $matches)) {
+		assertType("array{string, '[2]'}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		~(\{2})~
+		EOD, $string, $matches)) {
+		assertType("array{string, '{2}'}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		{(\{2\})}
+		EOD, $string, $matches)) {
+		assertType("array{string, '{2}'}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		~([a\]])~
+		EOD, $string, $matches)) {
+		assertType("array{string, ']'|'a'}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		~([a[])~
+		EOD, $string, $matches)) {
+		assertType("array{string, '['|'a'}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		~([a\]b])~
+		EOD, $string, $matches)) {
+		assertType("array{string, ']'|'a'|'b'}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		~([a[b])~
+		EOD, $string, $matches)) {
+		assertType("array{string, '['|'a'|'b'}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		~([a\[b])~
+		EOD, $string, $matches)) {
+		assertType("array{string, '['|'a'|'b'}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		[([a\[b])]
+		EOD, $string, $matches)) {
+		assertType("array{string, '['|'a'|'b'}", $matches);
+	}
+
+	if (preg_match(<<<'EOD'
+		{(x\\\{)|(y\\\\\})}
+		EOD, $string, $matches)) {
+		assertType("array{string, '', 'y\\\\\\\}'}|array{string, 'x\\\{'}", $matches);
+	}
+}
+
+function bugUnescapedDashAfterRange (string $string): void {
+	if (preg_match('/([0-1-y])/', $string, $matches)) {
+		assertType("array{string, non-empty-string}", $matches);
+	}
+}