RegularExpressionPatternRule: validate preg_quote'd patterns

conf/config.neon

@@ -1491,6 +1491,9 @@ services:
 -
 class: PHPStan\Type\Php\RegexArrayShapeMatcher
 
+-
+class: PHPStan\Type\Php\RegexExpressionHelper
+
 -
 class: PHPStan\Type\Php\ReflectionClassConstructorThrowTypeExtension
 tags:

src/Rules/Regexp/RegularExpressionPatternRule.php

@@ -9,6 +9,7 @@
 use PHPStan\Analyser\Scope;
 use PHPStan\Rules\Rule;
 use PHPStan\Rules\RuleErrorBuilder;
+use PHPStan\Type\Php\RegexExpressionHelper;
 use function in_array;
 use function sprintf;
 use function str_starts_with;
@@ -20,6 +21,12 @@
 class RegularExpressionPatternRule implements Rule
 {
 
+public function __construct(
+private RegexExpressionHelper $regexExpressionHelper,
+)
+{
+}
+
 public function getNodeType(): string
 {
 return FuncCall::class;
@@ -74,6 +81,9 @@ private function extractPatterns(FuncCall $functionCall, Scope $scope): array
 'preg_filter',
 ], true)
 ) {
+if ($patternNode instanceof Node\Expr\BinaryOp\Concat) {
+$patternType = $this->regexExpressionHelper->resolvePatternConcat($patternNode, $scope);
+}
 foreach ($patternType->getConstantStrings() as $constantStringType) {
 $patternStrings[] = $constantStringType->getValue();
 }

src/Type/Php/RegexArrayShapeMatcher.php

@@ -10,10 +10,8 @@
 use Nette\Utils\RegexpException;
 use Nette\Utils\Strings;
 use PhpParser\Node\Expr;
-use PhpParser\Node\Name;
 use PHPStan\Analyser\Scope;
 use PHPStan\Php\PhpVersion;
-use PHPStan\Reflection\InitializerExprTypeResolver;
 use PHPStan\ShouldNotHappenException;
 use PHPStan\TrinaryLogic;
 use PHPStan\Type\Accessory\AccessoryNonEmptyStringType;
@@ -56,7 +54,7 @@ final class RegexArrayShapeMatcher
 
 public function __construct(
 private PhpVersion $phpVersion,
-private InitializerExprTypeResolver $initializerExprTypeResolver,
+private RegexExpressionHelper $regexExpressionHelper,
 )
 {
 }
@@ -712,57 +710,10 @@ private function getLiteralValue(TreeNode $node): ?string
 private function getPatternType(Expr $patternExpr, Scope $scope): Type
 {
 if ($patternExpr instanceof Expr\BinaryOp\Concat) {
-return $this->resolvePatternConcat($patternExpr, $scope);
+return $this->regexExpressionHelper->resolvePatternConcat($patternExpr, $scope);
 }
 
 return $scope->getType($patternExpr);
 }
 
-/**
- * Ignores preg_quote() calls in the concatenation as these are not relevant for array-shape matching.
- *
- * This assumption only works for the ArrayShapeMatcher therefore it is not implemented for the common case in Scope.
- *
- * see https://github.com/phpstan/phpstan-src/pull/3233#discussion_r1676938085
- */
-private function resolvePatternConcat(Expr\BinaryOp\Concat $concat, Scope $scope): Type
-{
-$resolver = new class($scope) {
-
-public function __construct(private Scope $scope)
-{
-}
-
-public function resolve(Expr $expr): Type
-{
-if (
-$expr instanceof Expr\FuncCall
-&& $expr->name instanceof Name
-&& $expr->name->toLowerString() === 'preg_quote'
-) {
-return new ConstantStringType('');
-}
-
-if ($expr instanceof Expr\BinaryOp\Concat) {
-$left = $this->resolve($expr->left);
-$right = $this->resolve($expr->right);
-
-$strings = [];
-foreach ($left->toString()->getConstantStrings() as $leftString) {
-foreach ($right->toString()->getConstantStrings() as $rightString) {
-$strings[] = new ConstantStringType($leftString->getValue() . $rightString->getValue());
-}
-}
-
-return TypeCombinator::union(...$strings);
-}
-
-return $this->scope->getType($expr);
-}
-
-};
-
-return $this->initializerExprTypeResolver->getConcatType($concat->left, $concat->right, static fn (Expr $expr): Type => $resolver->resolve($expr));
-}
-
 }

src/Type/Php/RegexExpressionHelper.php

@@ -0,0 +1,69 @@
+<?php declare(strict_types = 1);
+
+namespace PHPStan\Type\Php;
+
+use PhpParser\Node\Expr;
+use PhpParser\Node\Name;
+use PHPStan\Analyser\Scope;
+use PHPStan\Reflection\InitializerExprTypeResolver;
+use PHPStan\Type\Constant\ConstantStringType;
+use PHPStan\Type\Type;
+use PHPStan\Type\TypeCombinator;
+
+final class RegexExpressionHelper
+{
+
+public function __construct(
+private InitializerExprTypeResolver $initializerExprTypeResolver,
+)
+{
+}
+
+/**
+ * Ignores preg_quote() calls in the concatenation as these are not relevant for array-shape matching.
+ *
+ * This assumption only works for the ArrayShapeMatcher therefore it is not implemented for the common case in Scope.
+ *
+ * see https://github.com/phpstan/phpstan-src/pull/3233#discussion_r1676938085
+ */
+public function resolvePatternConcat(Expr\BinaryOp\Concat $concat, Scope $scope): Type
+{
+$resolver = new class($scope) {
+
+public function __construct(private Scope $scope)
+{
+}
+
+public function resolve(Expr $expr): Type
+{
+if (
+$expr instanceof Expr\FuncCall
+&& $expr->name instanceof Name
+&& $expr->name->toLowerString() === 'preg_quote'
+) {
+return new ConstantStringType('');
+}
+
+if ($expr instanceof Expr\BinaryOp\Concat) {
+$left = $this->resolve($expr->left);
+$right = $this->resolve($expr->right);
+
+$strings = [];
+foreach ($left->toString()->getConstantStrings() as $leftString) {
+foreach ($right->toString()->getConstantStrings() as $rightString) {
+$strings[] = new ConstantStringType($leftString->getValue() . $rightString->getValue());
+}
+}
+
+return TypeCombinator::union(...$strings);
+}
+
+return $this->scope->getType($expr);
+}
+
+};
+
+return $this->initializerExprTypeResolver->getConcatType($concat->left, $concat->right, static fn (Expr $expr): Type => $resolver->resolve($expr));
+}
+
+}

tests/PHPStan/Rules/Regexp/RegularExpressionPatternRuleTest.php

@@ -4,6 +4,7 @@
 
 use PHPStan\Rules\Rule;
 use PHPStan\Testing\RuleTestCase;
+use PHPStan\Type\Php\RegexExpressionHelper;
 use function sprintf;
 use const PHP_VERSION_ID;
 
@@ -15,7 +16,9 @@ class RegularExpressionPatternRuleTest extends RuleTestCase
 
 protected function getRule(): Rule
 {
-return new RegularExpressionPatternRule();
+return new RegularExpressionPatternRule(
+self::getContainer()->getByType(RegexExpressionHelper::class),
+);
 }
 
 public function testValidRegexPatternBefore73(): void
@@ -115,6 +118,30 @@ public function testValidRegexPatternBefore73(): void
 'Regex pattern is invalid: Compilation failed: missing ) at offset 1 in pattern: ~(~',
 43,
 ],
+[
+'Regex pattern is invalid: Delimiter must not be alphanumeric, backslash, or NUL in pattern: nok',
+57,
+],
+[
+'Regex pattern is invalid: Delimiter must not be alphanumeric, backslash, or NUL in pattern: nok',
+58,
+],
+[
+'Regex pattern is invalid: Compilation failed: missing closing parenthesis at offset 1 in pattern: ~(~',
+59,
+],
+[
+'Regex pattern is invalid: Delimiter must not be alphanumeric, backslash, or NUL in pattern: noknono',
+61,
+],
+[
+'Regex pattern is invalid: Delimiter must not be alphanumeric, backslash, or NUL in pattern: noknope',
+62,
+],
+[
+'Regex pattern is invalid: Compilation failed: missing closing parenthesis at offset 1 in pattern: ~(~',
+63,
+],
 ],
 );
 }
@@ -221,6 +248,30 @@ public function testValidRegexPatternAfter73(): void
 'Regex pattern is invalid: Compilation failed: missing closing parenthesis at offset 1 in pattern: ~(~',
 43,
 ],
+[
+sprintf('Regex pattern is invalid: Delimiter must not be %s in pattern: nok', $messagePart),
+57,
+],
+[
+sprintf('Regex pattern is invalid: Delimiter must not be %s in pattern: nok', $messagePart),
+58,
+],
+[
+'Regex pattern is invalid: Compilation failed: missing closing parenthesis at offset 1 in pattern: ~(~',
+59,
+],
+[
+sprintf('Regex pattern is invalid: Delimiter must not be %s in pattern: noknono', $messagePart),
+61,
+],
+[
+sprintf('Regex pattern is invalid: Delimiter must not be %s in pattern: noknope', $messagePart),
+62,
+],
+[
+'Regex pattern is invalid: Compilation failed: missing closing parenthesis at offset 1 in pattern: ~(~',
+63,
+],
 ],
 );
 }

tests/PHPStan/Rules/Regexp/data/valid-regex-pattern.php

@@ -1,4 +1,4 @@
-<?php
+<?php namespace RegexExpressionPatterns;
 
 $string = (function (): string {})();
 
@@ -48,3 +48,17 @@
 ],
 ''
 );
+
+function doFoo(string $s) {
+preg_match('~ok'. preg_quote($s, '~') .'~', '');
+preg_match('~ok'. preg_quote($s) .'~', '');
+
+// invalid preg_quote delimiters will be reported by RegularExpressionQuotingRule
+preg_match('nok'. preg_quote($s), '');
+preg_match('nok'. preg_quote($s), '');
+preg_match('~('. preg_quote($s, '~') .'~', '');
+
+preg_replace('nok'. preg_quote($s).'nono', '');
+preg_replace('nok'. preg_quote($s).'nope', '');
+preg_replace('~('. preg_quote($s, '~') .'~', '');
+}