RegexArrayShapeMatcher - Fix matching literal dot character

phpstan · Sep 16, 2024 · 9437056 · 9437056
1 parent b3c25b8
commit 9437056
Show file tree

Hide file tree

Showing 2 changed files with 49 additions and 5 deletions.
diff --git a/src/Type/Regex/RegexGroupParser.php b/src/Type/Regex/RegexGroupParser.php
@@ -445,14 +445,14 @@ private function walkGroupAst(
 			foreach ($children as $child) {
 				$oldLiterals = $onlyLiterals;
 
-				$this->getLiteralValue($child, $oldLiterals, true, $patternModifiers);
+				$this->getLiteralValue($child, $oldLiterals, true, $patternModifiers, true);
 				foreach ($oldLiterals ?? [] as $oldLiteral) {
 					$newLiterals[] = $oldLiteral;
 				}
 			}
 			$onlyLiterals = $newLiterals;
 		} elseif ($ast->getId() === 'token') {
-			$literalValue = $this->getLiteralValue($ast, $onlyLiterals, !$inClass, $patternModifiers);
+			$literalValue = $this->getLiteralValue($ast, $onlyLiterals, !$inClass, $patternModifiers, false);
 			if ($literalValue !== null) {
 				if (Strings::match($literalValue, '/^\d+$/') === null) {
 					$isNumeric = TrinaryLogic::createNo();
@@ -508,7 +508,7 @@ private function isMaybeEmptyNode(TreeNode $node, string $patternModifiers, bool
 			}
 		}
 
-		$literal = $this->getLiteralValue($node, $onlyLiterals, false, $patternModifiers);
+		$literal = $this->getLiteralValue($node, $onlyLiterals, false, $patternModifiers, false);
 		if ($literal !== null) {
 			if ($literal !== '' && $literal !== '0') {
 				$isNonFalsy = true;
@@ -528,7 +528,7 @@ private function isMaybeEmptyNode(TreeNode $node, string $patternModifiers, bool
 	/**
 	 * @param array<string>|null $onlyLiterals
 	 */
-	private function getLiteralValue(TreeNode $node, ?array &$onlyLiterals, bool $appendLiterals, string $patternModifiers): ?string
+	private function getLiteralValue(TreeNode $node, ?array &$onlyLiterals, bool $appendLiterals, string $patternModifiers, bool $inCharacterClass): ?string
 	{
 		if ($node->getId() !== 'token') {
 			return null;
@@ -551,15 +551,17 @@ private function getLiteralValue(TreeNode $node, ?array &$onlyLiterals, bool $ap
 				return null;
 			}
 
+			$isEscaped = false;
 			if (strlen($value) > 1 && $value[0] === '\\') {
 				$value = substr($value, 1) ?: '';
+				$isEscaped = true;
 			}
 
 			if (
 				$appendLiterals
 				&& in_array($token, ['literal', 'range', 'class_', '_class_literal'], true)
 				&& $onlyLiterals !== null
-				&& !in_array($value, ['.'], true)
+				&& (!in_array($value, ['.'], true) || $isEscaped || $inCharacterClass)
 			) {
 				if ($onlyLiterals === []) {
 					$onlyLiterals = [$value];

diff --git a/tests/PHPStan/Analyser/nsrt/bug-11699.php b/tests/PHPStan/Analyser/nsrt/bug-11699.php
@@ -0,0 +1,42 @@
+<?php declare(strict_types = 1);
+
+namespace Bug11699;
+
+use function PHPStan\Testing\assertType;
+
+function doFoo():void {
+	$string = 'Foo.bar';
+	$match = [];
+	$result = preg_match('~(?<AB>[\~,\?\.])~', $string, $match);
+	if ($result === 1) {
+		assertType("','|'.'|'?'|'~'", $match['AB']);
+	}
+}
+
+function doFoo1():void {
+	$string = 'Foo.bar';
+	$match = [];
+	$result = preg_match('~(?<AB>[\~,\?.])~', $string, $match); // dot in character class does not need to be escaped
+	if ($result === 1) {
+		assertType("','|'.'|'?'|'~'", $match['AB']);
+	}
+}
+
+function doFoo2():void {
+	$string = 'Foo.bar';
+	$match = [];
+	$result = preg_match('~(?<AB>.)~', $string, $match);
+	if ($result === 1) {
+		assertType("non-empty-string", $match['AB']);
+	}
+}
+
+
+function doFoo3():void {
+	$string = 'Foo.bar';
+	$match = [];
+	$result = preg_match('~(?<AB>\.)~', $string, $match);
+	if ($result === 1) {
+		assertType("'.'", $match['AB']);
+	}
+}