From e2c94bb28c00bc0d52a7c905852971121a307e7c Mon Sep 17 00:00:00 2001
From: Markus Staab <maggus.staab@googlemail.com>
Date: Sat, 27 Jul 2024 22:34:22 +0200
Subject: [PATCH] remove delimiters and modifiers before parsing

---
 src/Type/Php/RegexArrayShapeMatcher.php       |  3 +-
 src/Type/Php/RegexExpressionHelper.php        | 46 +++++++++++++------
 .../Analyser/nsrt/preg_match_shapes.php       |  4 +-
 .../Analyser/nsrt/preg_match_shapes_php82.php |  5 ++
 4 files changed, 41 insertions(+), 17 deletions(-)

diff --git a/src/Type/Php/RegexArrayShapeMatcher.php b/src/Type/Php/RegexArrayShapeMatcher.php
index 3e65133284..626bc4c35d 100644
--- a/src/Type/Php/RegexArrayShapeMatcher.php
+++ b/src/Type/Php/RegexArrayShapeMatcher.php
@@ -406,8 +406,9 @@ private function parseGroups(string $regex): ?array
 			return null;
 		}
 
+		$rawRegex = $this->regexExpressionHelper->removeDelimitersAndModifiers($regex);
 		try {
-			$ast = self::$parser->parse($regex);
+			$ast = self::$parser->parse($rawRegex);
 		} catch (Exception) {
 			return null;
 		}
diff --git a/src/Type/Php/RegexExpressionHelper.php b/src/Type/Php/RegexExpressionHelper.php
index b06d977e94..532bc4c914 100644
--- a/src/Type/Php/RegexExpressionHelper.php
+++ b/src/Type/Php/RegexExpressionHelper.php
@@ -72,11 +72,33 @@ public function resolve(Expr $expr): Type
 
 	public function getPatternModifiers(string $pattern): ?string
 	{
-		$delimiter = $this->getDelimiterFromString(new ConstantStringType($pattern));
-		if ($delimiter === null) {
+		$endDelimiterPos = $this->getEndDelimiterPos($pattern);
+
+		if ($endDelimiterPos === false) {
 			return null;
 		}
 
+		return substr($pattern, $endDelimiterPos + 1);
+	}
+
+	public function removeDelimitersAndModifiers(string $pattern): string
+	{
+		$endDelimiterPos = $this->getEndDelimiterPos($pattern);
+
+		if ($endDelimiterPos === false) {
+			return $pattern;
+		}
+
+		return substr($pattern, 1, $endDelimiterPos - 1);
+	}
+
+	private function getEndDelimiterPos(string $pattern): false|int
+	{
+		$startDelimiter = $this->getPatternDelimiter($pattern);
+		if ($startDelimiter === null) {
+			return false;
+		}
+
 		// delimiter variants, see https://www.php.net/manual/en/regexp.reference.delimiters.php
 		$bracketStyleDelimiters = [
 			'{' => '}',
@@ -84,18 +106,14 @@ public function getPatternModifiers(string $pattern): ?string
 			'[' => ']',
 			'<' => '>',
 		];
-		if (array_key_exists($delimiter, $bracketStyleDelimiters)) {
-			$endDelimiterPos = strrpos($pattern, $bracketStyleDelimiters[$delimiter]);
+		if (array_key_exists($startDelimiter, $bracketStyleDelimiters)) {
+			$endDelimiterPos = strrpos($pattern, $bracketStyleDelimiters[$startDelimiter]);
 		} else {
 			// same start and end delimiter
-			$endDelimiterPos = strrpos($pattern, $delimiter);
+			$endDelimiterPos = strrpos($pattern, $startDelimiter);
 		}
 
-		if ($endDelimiterPos === false) {
-			return null;
-		}
-
-		return substr($pattern, $endDelimiterPos + 1);
+		return $endDelimiterPos;
 	}
 
 	/**
@@ -113,7 +131,7 @@ public function getPatternDelimiters(Concat $concat, Scope $scope): array
 
 		$delimiters = [];
 		foreach ($left->getConstantStrings() as $leftString) {
-			$delimiter = $this->getDelimiterFromString($leftString);
+			$delimiter = $this->getPatternDelimiter($leftString->getValue());
 			if ($delimiter === null) {
 				continue;
 			}
@@ -123,13 +141,13 @@ public function getPatternDelimiters(Concat $concat, Scope $scope): array
 		return $delimiters;
 	}
 
-	private function getDelimiterFromString(ConstantStringType $string): ?string
+	private function getPatternDelimiter(string $regex): ?string
 	{
-		if ($string->getValue() === '') {
+		if ($regex === '') {
 			return null;
 		}
 
-		return substr($string->getValue(), 0, 1);
+		return substr($regex, 0, 1);
 	}
 
 }
diff --git a/tests/PHPStan/Analyser/nsrt/preg_match_shapes.php b/tests/PHPStan/Analyser/nsrt/preg_match_shapes.php
index c273a83a47..af57075078 100644
--- a/tests/PHPStan/Analyser/nsrt/preg_match_shapes.php
+++ b/tests/PHPStan/Analyser/nsrt/preg_match_shapes.php
@@ -29,9 +29,9 @@ function doMatch(string $s): void {
 	assertType('array{}|array{string, non-empty-string}', $matches);
 
 	if (preg_match('(Price: (£|€))i', $s, $matches)) {
-		assertType('array{string, non-empty-string, non-empty-string}', $matches);
+		assertType('array{string, non-empty-string}', $matches);
 	}
-	assertType('array{}|array{string, non-empty-string, non-empty-string}', $matches);
+	assertType('array{}|array{string, non-empty-string}', $matches);
 
 	if (preg_match('_foo(.)\_i_i', $s, $matches)) {
 		assertType('array{string, non-empty-string}', $matches);
diff --git a/tests/PHPStan/Analyser/nsrt/preg_match_shapes_php82.php b/tests/PHPStan/Analyser/nsrt/preg_match_shapes_php82.php
index e337e4707a..dfbcab477e 100644
--- a/tests/PHPStan/Analyser/nsrt/preg_match_shapes_php82.php
+++ b/tests/PHPStan/Analyser/nsrt/preg_match_shapes_php82.php
@@ -39,3 +39,8 @@ function (string $s): void {
 		assertType('array{0: string, num: numeric-string, 1: numeric-string}', $matches);
 	}
 };
+function (string $s): void {
+	if (preg_match('[(\d+)(?P<num>\d+)]n', $s, $matches)) {
+		assertType('array{0: string, num: numeric-string, 1: numeric-string}', $matches);
+	}
+};