Fix matches[0] type for regexes containing \K

Author: mvorisek

Diff for: src/Type/Regex/RegexGroupParser.php

@@ -110,7 +110,7 @@ public function parseGroups(string $regex): ?RegexAstWalkResult
 			RegexGroupWalkResult::createEmpty(),
 		);
 
-		if (!$subjectAsGroupResult->mightContainEmptyStringLiteral()) {
+		if (!$subjectAsGroupResult->mightContainEmptyStringLiteral() && !$this->containsEscapeK($ast)) {
 			// we could handle numeric-string, in case we know the regex is delimited by ^ and $
 			if ($subjectAsGroupResult->isNonFalsy()->yes()) {
 				$astWalkResult = $astWalkResult->withSubjectBaseType(
@@ -171,6 +171,21 @@ private function updateCapturingAstAddEmptyToken(TreeNode $ast): void
 		$ast->setChildren([$emptyAlternationAst]);
 	}
 
+	private function containsEscapeK(TreeNode $ast): bool
+	{
+		if ($ast->getId() === 'token' && $ast->getValueToken() === 'match_point_reset') {
+			return true;
+		}
+
+		foreach ($ast->getChildren() as $child) {
+			if ($this->containsEscapeK($child)) {
+				return true;
+			}
+		}
+
+		return false;
+	}
+
 	private function walkRegexAst(
 		TreeNode $ast,
 		?RegexAlternation $alternation,

Diff for: tests/PHPStan/Analyser/nsrt/preg_match_shapes.php

@@ -1011,7 +1011,65 @@ function bug12749f(string $str): void
 	}
 }
 
-function bug12397(string $string) : array {
+function bug12397(string $string): void {
 	$m = preg_match('#\b([A-Z]{2,})-(\d+)#', $string, $match);
 	assertType('list{0?: string, 1?: non-falsy-string, 2?: numeric-string}', $match);
 }
+
+function bug12792(string $string): void {
+	if (preg_match('~a\Kb~', $string, $match) === 1) {
+		assertType('array{string}', $match); // could be array{'b'}
+	}
+
+	if (preg_match('~a\K~', $string, $match) === 1) {
+		assertType('array{string}', $match); // could be array{''}
+	}
+
+	if (preg_match('~a\K.+~', $string, $match) === 1) {
+		assertType('array{string}', $match); // could be array{non-empty-string}
+	}
+
+	if (preg_match('~a\K.*~', $string, $match) === 1) {
+		assertType('array{string}', $match);
+	}
+
+	if (preg_match('~a\K(.+)~', $string, $match) === 1) {
+		assertType('array{string, non-empty-string}', $match); // could be array{non-empty-string, non-empty-string}
+	}
+
+	if (preg_match('~a\K(.*)~', $string, $match) === 1) {
+		assertType('array{string, string}', $match);
+	}
+
+	if (preg_match('~a\K(.+?)~', $string, $match) === 1) {
+		assertType('array{string, non-empty-string}', $match); // could be array{non-empty-string, non-empty-string}
+	}
+
+	if (preg_match('~a\K(.*?)~', $string, $match) === 1) {
+		assertType('array{string, string}', $match);
+	}
+
+	if (preg_match('~a\K(?=.+)~', $string, $match) === 1) {
+		assertType('array{string}', $match); // could be array{''}
+	}
+
+	if (preg_match('~a\K(?=.*)~', $string, $match) === 1) {
+		assertType('array{string}', $match); // could be array{''}
+	}
+
+	if (preg_match('~a(?:x\Kb|c)~', $string, $match) === 1) {
+		assertType('array{string}', $match); // could be array{'ac'|'b'}
+	}
+
+	if (preg_match('~a(?:c|x\Kb)~', $string, $match) === 1) {
+		assertType('array{string}', $match); // could be array{'ac'|'b'}
+	}
+
+	if (preg_match('~a(y|(?:x\Kb|c))d~', $string, $match) === 1) {
+		assertType('array{string, non-empty-string}', $match); // could be array{'acd'|'ayd'|'bd', 'c'|'xb'|'y'}
+	}
+
+	if (preg_match('~a((?:c|x\Kb)|y)d~', $string, $match) === 1) {
+		assertType('array{string, non-empty-string}', $match); // could be array{'acd'|'ayd'|'bd', 'c'|'xb'|'y'}
+	}
+}

Diff for: tests/PHPStan/Analyser/nsrt/preg_replace_callback_shapes.php

@@ -45,3 +45,14 @@ function ($matches) {
 		PREG_OFFSET_CAPTURE|PREG_UNMATCHED_AS_NULL
 	);
 };
+
+function bug12792(string $string) : void {
+	preg_replace_callback(
+		'~\'(?:[^\']+|\'\')*+\'\K|\[(\w*)\]~',
+		function ($matches) {
+			assertType("array{0: string, 1?: string}", $matches);
+			return '';
+		},
+		$string
+	);
+}