From c053dbc01983f6dd78a78b8154a00afb64088b33 Mon Sep 17 00:00:00 2001
From: Markus Staab <maggus.staab@googlemail.com>
Date: Tue, 17 Dec 2024 11:35:44 +0100
Subject: [PATCH] Support `#` comments in regex with `x` modifier

---
 src/Type/Regex/RegexGroupParser.php       | 20 +++++++++-----
 tests/PHPStan/Analyser/nsrt/bug-12242.php | 32 +++++++++++++++++++++++
 2 files changed, 45 insertions(+), 7 deletions(-)
 create mode 100644 tests/PHPStan/Analyser/nsrt/bug-12242.php

diff --git a/src/Type/Regex/RegexGroupParser.php b/src/Type/Regex/RegexGroupParser.php
index c818426111..a98fb20b42 100644
--- a/src/Type/Regex/RegexGroupParser.php
+++ b/src/Type/Regex/RegexGroupParser.php
@@ -23,6 +23,7 @@
 use function count;
 use function in_array;
 use function is_int;
+use function preg_replace;
 use function rtrim;
 use function sscanf;
 use function str_contains;
@@ -64,13 +65,6 @@ public function parseGroups(string $regex): ?array
 			return null;
 		}
 
-		$rawRegex = $this->regexExpressionHelper->removeDelimitersAndModifiers($regex);
-		try {
-			$ast = self::$parser->parse($rawRegex);
-		} catch (Exception) {
-			return null;
-		}
-
 		$modifiers = $this->regexExpressionHelper->getPatternModifiers($regex) ?? '';
 		foreach (self::NOT_SUPPORTED_MODIFIERS as $notSupportedModifier) {
 			if (str_contains($modifiers, $notSupportedModifier)) {
@@ -78,6 +72,18 @@ public function parseGroups(string $regex): ?array
 			}
 		}
 
+		if (str_contains($modifiers, 'x')) {
+			// in freespacing mode the # character starts a comment and runs until the end of the line
+			$regex = preg_replace('/[^?]#.*/', '', $regex) ?? '';
+		}
+
+		$rawRegex = $this->regexExpressionHelper->removeDelimitersAndModifiers($regex);
+		try {
+			$ast = self::$parser->parse($rawRegex);
+		} catch (Exception) {
+			return null;
+		}
+
 		$captureOnlyNamed = false;
 		if ($this->phpVersion->supportsPregCaptureOnlyNamedGroups()) {
 			$captureOnlyNamed = str_contains($modifiers, 'n');
diff --git a/tests/PHPStan/Analyser/nsrt/bug-12242.php b/tests/PHPStan/Analyser/nsrt/bug-12242.php
new file mode 100644
index 0000000000..cb6d424567
--- /dev/null
+++ b/tests/PHPStan/Analyser/nsrt/bug-12242.php
@@ -0,0 +1,32 @@
+<?php // lint >= 7.4
+
+namespace Bug12242;
+
+use function PHPStan\Testing\assertType;
+
+function foo(string $str): void
+{
+	$regexp = '/
+		# (
+		([\d,]*)
+		# )
+	/x';
+	if (preg_match($regexp, $str, $match)) {
+		assertType('array{string, string}', $match);
+	}
+}
+
+function bar(string $str): void
+{
+	$regexp = '/^
+            (\w+)        # column type [1]
+            [\(]         # (
+                ?([\d,]*)  # size or size, precision [2]
+            [\)]         # )
+            ?\s*         # whitespace
+            (\w*)        # extra description (UNSIGNED, CHARACTER SET, ...) [3]
+        $/x';
+	if (preg_match($regexp, $str, $matches)) {
+		assertType('array{string, non-empty-string, string, string}', $matches);
+	}
+}