Rewrite OptimizedDirectorySourceLocator to use PhpFileCleaner from Composer

See composer/composer#10107

Author: ondrejmirtes

phpstan-baseline.neon

@@ -130,14 +130,19 @@ parameters:
 			path: src/Reflection/BetterReflection/SourceLocator/OptimizedDirectorySourceLocator.php
 
 		-
-			message: "#^Only booleans are allowed in a negated boolean, int\\|false given\\.$#"
+			message: "#^Only booleans are allowed in a negated boolean, int\\|false\\|null given\\.$#"
 			count: 1
 			path: src/Reflection/BetterReflection/SourceLocator/OptimizedDirectorySourceLocator.php
 
+		-
+			message: "#^Only booleans are allowed in &&, int\\|false given on the right side\\.$#"
+			count: 1
+			path: src/Reflection/BetterReflection/SourceLocator/PhpFileCleaner.php
+
 		-
 			message: "#^Only booleans are allowed in an if condition, int\\|false given\\.$#"
 			count: 1
-			path: src/Reflection/BetterReflection/SourceLocator/OptimizedDirectorySourceLocator.php
+			path: src/Reflection/BetterReflection/SourceLocator/PhpFileCleaner.php
 
 		-
 			message: "#^Method PHPStan\\\\Reflection\\\\ClassReflection\\:\\:__construct\\(\\) has parameter \\$reflection with generic class ReflectionClass but does not specify its types\\: T$#"

src/Php/PhpVersion.php

@@ -142,4 +142,9 @@ public function supportsReadOnlyProperties(): bool
 		return $this->versionId >= 80100;
 	}
 
+	public function supportsEnums(): bool
+	{
+		return $this->versionId >= 80100;
+	}
+
 }

src/Reflection/BetterReflection/SourceLocator/OptimizedDirectorySourceLocator.php

@@ -8,16 +8,23 @@
 use PHPStan\BetterReflection\Reflector\Reflector;
 use PHPStan\BetterReflection\SourceLocator\Ast\Strategy\NodeToReflection;
 use PHPStan\BetterReflection\SourceLocator\Type\SourceLocator;
+use PHPStan\Php\PhpVersion;
 use function array_key_exists;
 
 class OptimizedDirectorySourceLocator implements SourceLocator
 {
 
 	private \PHPStan\Reflection\BetterReflection\SourceLocator\FileNodesFetcher $fileNodesFetcher;
 
+	private PhpVersion $phpVersion;
+
+	private PhpFileCleaner $cleaner;
+
 	/** @var string[] */
 	private array $files;
 
+	private string $extraTypes;
+
 	/** @var array<string, string>|null */
 	private ?array $classToFile = null;
 
@@ -39,11 +46,24 @@ class OptimizedDirectorySourceLocator implements SourceLocator
 	 */
 	public function __construct(
 		FileNodesFetcher $fileNodesFetcher,
+		PhpVersion $phpVersion,
 		array $files
 	)
 	{
 		$this->fileNodesFetcher = $fileNodesFetcher;
+		$this->phpVersion = $phpVersion;
 		$this->files = $files;
+
+		$extraTypes = '';
+		$extraTypesArray = [];
+		if ($this->phpVersion->supportsEnums()) {
+			$extraTypes = '|enum';
+			$extraTypesArray[] = 'enum';
+		}
+
+		$this->extraTypes = $extraTypes;
+
+		$this->cleaner = new PhpFileCleaner(array_merge(['class', 'interface', 'trait', 'function'], $extraTypesArray));
 	}
 
 	public function locateIdentifier(Reflector $reflector, Identifier $identifier): ?Reflection
@@ -197,79 +217,18 @@ private function findSymbols(string $file): array
 			return ['classes' => [], 'functions' => []];
 		}
 
-		if (!preg_match('{\b(?:class|interface|trait|function)\s}i', $contents)) {
+		if (!preg_match_all(sprintf('{\b(?:class|interface|trait|function%s)\s}i', $this->extraTypes), $contents, $matches)) {
 			return ['classes' => [], 'functions' => []];
 		}
 
-		// strip heredocs/nowdocs
-		$heredocRegex = '{
-            # opening heredoc/nowdoc delimiter (word-chars)
-            <<<[ \t]*+([\'"]?)(\w++)\\1
-            # needs to be followed by a newline
-            (?:\r\n|\n|\r)
-            # the meat of it, matching line by line until end delimiter
-            (?:
-                # a valid line is optional white-space (possessive match) not followed by the end delimiter, then anything goes for the rest of the line
-                [\t ]*+(?!\\2 \b)[^\r\n]*+
-                # end of line(s)
-                [\r\n]++
-            )*
-            # end delimiter
-            [\t ]*+ \\2 (?=\b)
-        }x';
-
-		// run first assuming the file is valid unicode
-		$contentWithoutHeredoc = preg_replace($heredocRegex . 'u', 'null', $contents);
-		if ($contentWithoutHeredoc === null) {
-			// run again without unicode support if the file failed to be parsed
-			$contents = preg_replace($heredocRegex, 'null', $contents);
-		} else {
-			$contents = $contentWithoutHeredoc;
-		}
-		unset($contentWithoutHeredoc);
-
-		if ($contents === null) {
-			return ['classes' => [], 'functions' => []];
-		}
-		// strip strings
-		$contents = preg_replace('{"[^"\\\\]*+(\\\\.[^"\\\\]*+)*+"|\'[^\'\\\\]*+(\\\\.[^\'\\\\]*+)*+\'}s', 'null', $contents);
-		if ($contents === null) {
-			return ['classes' => [], 'functions' => []];
-		}
-		// strip leading non-php code if needed
-		if (strpos($contents, '<?') !== 0) {
-			$contents = preg_replace('{^.+?<\?}s', '<?', $contents, 1, $replacements);
-			if ($contents === null) {
-				return ['classes' => [], 'functions' => []];
-			}
-			if ($replacements === 0) {
-				return ['classes' => [], 'functions' => []];
-			}
-		}
-		// strip non-php blocks in the file
-		$contents = preg_replace('{\?>(?:[^<]++|<(?!\?))*+<\?}s', '?><?', $contents);
-		if ($contents === null) {
-			return ['classes' => [], 'functions' => []];
-		}
-		// strip trailing non-php code if needed
-		$pos = strrpos($contents, '?>');
-		if ($pos !== false && strpos(substr($contents, $pos), '<?') === false) {
-			$contents = substr($contents, 0, $pos);
-		}
-		// strip comments if short open tags are in the file
-		if (preg_match('{(<\?)(?!(php|hh))}i', $contents)) {
-			$contents = preg_replace('{//.* | /\*(?:[^*]++|\*(?!/))*\*/}x', '', $contents);
-			if ($contents === null) {
-				return ['classes' => [], 'functions' => []];
-			}
-		}
+		$contents = $this->cleaner->clean($contents, count($matches[0]));
 
-		preg_match_all('{
+		preg_match_all(sprintf('{
             (?:
-                 \b(?<![\$:>])(?P<type>class|interface|trait|function) \s++ (?P<byref>&\s*)? (?P<name>[a-zA-Z_\x7f-\xff:][a-zA-Z0-9_\x7f-\xff:\-]*+)
+                 \b(?<![\$:>])(?P<type>class|interface|trait|function%s) \s++ (?P<byref>&\s*)? (?P<name>[a-zA-Z_\x7f-\xff:][a-zA-Z0-9_\x7f-\xff:\-]*+)
                | \b(?<![\$:>])(?P<ns>namespace) (?P<nsname>\s++[a-zA-Z_\x7f-\xff][a-zA-Z0-9_\x7f-\xff]*+(?:\s*+\\\\\s*+[a-zA-Z_\x7f-\xff][a-zA-Z0-9_\x7f-\xff]*+)*+)? \s*+ [\{;]
             )
-        }ix', $contents, $matches);
+        }ix', $this->extraTypes), $contents, $matches);
 
 		$classes = [];
 		$functions = [];

src/Reflection/BetterReflection/SourceLocator/OptimizedDirectorySourceLocatorFactory.php

@@ -3,6 +3,7 @@
 namespace PHPStan\Reflection\BetterReflection\SourceLocator;
 
 use PHPStan\File\FileFinder;
+use PHPStan\Php\PhpVersion;
 
 class OptimizedDirectorySourceLocatorFactory
 {
@@ -11,16 +12,20 @@ class OptimizedDirectorySourceLocatorFactory
 
 	private FileFinder $fileFinder;
 
-	public function __construct(FileNodesFetcher $fileNodesFetcher, FileFinder $fileFinder)
+	private PhpVersion $phpVersion;
+
+	public function __construct(FileNodesFetcher $fileNodesFetcher, FileFinder $fileFinder, PhpVersion $phpVersion)
 	{
 		$this->fileNodesFetcher = $fileNodesFetcher;
 		$this->fileFinder = $fileFinder;
+		$this->phpVersion = $phpVersion;
 	}
 
 	public function createByDirectory(string $directory): OptimizedDirectorySourceLocator
 	{
 		return new OptimizedDirectorySourceLocator(
 			$this->fileNodesFetcher,
+			$this->phpVersion,
 			$this->fileFinder->findFiles([$directory])->getFiles()
 		);
 	}
@@ -33,6 +38,7 @@ public function createByFiles(array $files): OptimizedDirectorySourceLocator
 	{
 		return new OptimizedDirectorySourceLocator(
 			$this->fileNodesFetcher,
+			$this->phpVersion,
 			$files
 		);
 	}