Generated baseline can't be used if regex can't be compilated to UTF-8 chars

.github/workflows/e2e-tests.yml

@@ -322,6 +322,10 @@ jobs:
           - script: |
               cd e2e/bug-11819
               ../../bin/phpstan
+          - script: |
+              cd e2e/bug-12629
+              ../../bin/phpstan --generate-baseline # should generate without crash
+              ../../bin/phpstan # should re-analyze without new errors
 
     steps:
       - name: "Checkout"

e2e/bug-12629/phpstan-baseline.neon

@@ -0,0 +1,19 @@
+parameters:
+	ignoreErrors:
+		-
+			message: '#^Method Bug12629\\Bug12629\:\:is_macintosh_enc\(\) has no return type specified\.$#'
+			identifier: missingType.return
+			count: 1
+			path: src/bug-12629.php
+
+		-
+			message: '#^Method Bug12629\\Bug12629\:\:is_macintosh_enc\(\) has parameter \$s with no type specified\.$#'
+			identifier: missingType.parameter
+			count: 1
+			path: src/bug-12629.php
+
+		-
+			message: '#^Method Bug12629\\Bug12629\:\:is_macintosh_enc\(\) is unused\.$#'
+			identifier: method.unused
+			count: 1
+			path: src/bug-12629.php

e2e/bug-12629/phpstan.neon

@@ -0,0 +1,7 @@
+includes:
+	- phpstan-baseline.neon
+
+parameters:
+	level: 8
+	paths:
+		- src

e2e/bug-12629/src/bug-12629.php

@@ -0,0 +1,17 @@
+<?php
+
+namespace Bug12629;
+
+class Bug12629 {
+	private function is_macintosh_enc($s) {
+
+		if(!is_string($s)) {
+			return false;
+		}
+
+		preg_match_all("![\x80-\x9f]!u", $s, $matchesMacintosh);
+		preg_match_all("!\xc3[\x80-\x9f]!u", $s, $matchesUtf8);
+
+		return count($matchesMacintosh[0]) > 0 && 0 == count($matchesUtf8[0]);
+	}
+}

src/Rules/Regexp/RegularExpressionPatternRule.php

@@ -9,11 +9,15 @@
 use PHPStan\Analyser\Scope;
 use PHPStan\Rules\Rule;
 use PHPStan\Rules\RuleErrorBuilder;
+use PHPStan\ShouldNotHappenException;
 use PHPStan\Type\Regex\RegexExpressionHelper;
 use function in_array;
 use function sprintf;
 use function str_starts_with;
+use function strlen;
+use function strpos;
 use function strtolower;
+use function substr;
 
 /**
  * @implements Rule<Node\Expr\FuncCall>
@@ -123,7 +127,18 @@ private function validatePattern(string $pattern): ?string
 		try {
 			Strings::match('', $pattern);
 		} catch (RegexpException $e) {
-			return $e->getMessage();
+			$invalidPatternMessage = $e->getMessage();
+			try {
+				Strings::match($invalidPatternMessage, '//u');
+				return $invalidPatternMessage;
+			} catch (RegexpException) {
+				$patternPos = strpos($invalidPatternMessage, 'pattern:');
+				if ($patternPos === false) {
+					throw new ShouldNotHappenException();
+				}
+				// strip invalid utf-8 pattern contents to keep the error message NEON parsable.
+				return substr($invalidPatternMessage, 0, $patternPos + strlen('pattern:') - 1);
+			}
 		}
 
 		return null;

tests/PHPStan/Rules/Regexp/RegularExpressionPatternRuleTest.php

@@ -332,4 +332,21 @@ public function dataArrayShapePatterns(): iterable
 		];
 	}
 
+	public function testBug12629(): void
+	{
+		$this->analyse(
+			[__DIR__ . '/data/bug-12629.php'],
+			[
+				[
+					'Regex pattern is invalid: Compilation failed: UTF-8 error: isolated byte with 0x80 bit set at offset 1 in pattern',
+					12,
+				],
+				[
+					'Regex pattern is invalid: Compilation failed: UTF-8 error: byte 2 top bits not 0x80 at offset 0 in pattern',
+					13,
+				],
+			],
+		);
+	}
+
 }

tests/PHPStan/Rules/Regexp/data/bug-12629.php

@@ -0,0 +1,17 @@
+<?php
+
+namespace Bug12629b;
+
+class Bug12629 {
+	private function is_macintosh_enc($s) {
+
+		if(!is_string($s)) {
+			return false;
+		}
+
+		preg_match_all("![\x80-\x9f]!u", $s, $matchesMacintosh);
+		preg_match_all("!\xc3[\x80-\x9f]!u", $s, $matchesUtf8);
+
+		return count($matchesMacintosh[0]) > 0 && 0 == count($matchesUtf8[0]);
+	}
+}