gh-2611: Introduce decorator to expand PHAR files for indexer (#2634)

CHANGELOG.md

@@ -5,7 +5,7 @@ Changelog
 
 Features:
 
-  - PHAR Indexing #2412 @dantleech
+  - PHAR Indexing #2412 #2611 @dantleech
 
 Improvements:
 

lib/Indexer/Adapter/Filesystem/FilesystemFileListProvider.php

@@ -4,6 +4,7 @@
 
 use Phpactor\Filesystem\Domain\FilePath;
 use Phpactor\Filesystem\Domain\Filesystem;
+use Phpactor\Indexer\Adapter\Php\FileInfoPharExpandingIterator;
 use Phpactor\Indexer\Model\FileList;
 use Phpactor\Indexer\Model\FileListProvider;
 use SplFileInfo;
@@ -45,6 +46,11 @@ public function provideFileList(Index $index, ?string $subPath = null): FileList
             });
         }
 
-        return FileList::fromInfoIterator($files->getSplFileInfoIterator());
+        return FileList::fromInfoIterator(
+            new FileInfoPharExpandingIterator(
+                $files->getSplFileInfoIterator(),
+                $this->supportedExtensions,
+            )
+        );
     }
 }

lib/Indexer/Adapter/Php/FileInfoPharExpandingIterator.php

@@ -0,0 +1,47 @@
+<?php
+
+namespace Phpactor\Indexer\Adapter\Php;
+
+use Iterator;
+use IteratorAggregate;
+use Phar;
+use RecursiveIteratorIterator;
+use SplFileInfo;
+use Traversable;
+use UnexpectedValueException;
+
+/**
+ * @implements IteratorAggregate<SplFileInfo>
+ */
+class FileInfoPharExpandingIterator implements IteratorAggregate
+{
+    /**
+     * @param Iterator<SplFileInfo> $innerIterator
+     * @param list<string> $supportedExtensions
+     */
+    public function __construct(private Iterator $innerIterator, private array $supportedExtensions = ['php'])
+    {
+    }
+
+    public function getIterator(): Traversable
+    {
+        foreach ($this->innerIterator as $fileInfo) {
+            if ($fileInfo->getExtension() !== 'phar') {
+                yield $fileInfo;
+            }
+            try {
+                $phar = new Phar($fileInfo->getPathname());
+            } catch (UnexpectedValueException) {
+                continue;
+            }
+            $iterator = new RecursiveIteratorIterator($phar);
+            foreach ($iterator as $fileInfo) {
+                assert($fileInfo instanceof SplFileInfo);
+                if (!in_array($fileInfo->getExtension(), $this->supportedExtensions)) {
+                    continue;
+                }
+                yield $fileInfo;
+            }
+        }
+    }
+}

lib/Indexer/Model/FileList.php

@@ -7,6 +7,7 @@
 use Iterator;
 use IteratorAggregate;
 use SplFileInfo;
+use Traversable;
 
 /**
  * @implements \IteratorAggregate<SplFileInfo>
@@ -38,9 +39,9 @@ public static function empty(): self
     }
 
     /**
-     * @param Iterator<SplFileInfo> $splFileInfos
+     * @param Traversable<SplFileInfo> $splFileInfos
      */
-    public static function fromInfoIterator(Iterator $splFileInfos): self
+    public static function fromInfoIterator(Traversable $splFileInfos): self
     {
         return new self($splFileInfos);
     }

lib/Indexer/Model/IndexJob.php

@@ -21,25 +21,12 @@ public function __construct(private IndexBuilder $indexBuilder, private FileList
      */
     public function generator(): Generator
     {
-
         foreach ($this->fileList as $fileInfo) {
             assert($fileInfo instanceof SplFileInfo);
             if ($fileInfo->isLink()) {
                 continue;
             }
 
-            // TODO: could refactor this to iterate the PHAR in the file list provider.
-            if ($fileInfo->getExtension() === 'phar') {
-                try {
-                    $phar = new Phar($fileInfo->getPathname());
-                } catch (UnexpectedValueException $e) {
-                    continue;
-                }
-                iterator_to_array($this->indexPharFile($phar));
-                yield $fileInfo->getPathname();
-                continue;
-            }
-
             $contents = @file_get_contents($fileInfo->getPathname());
 
             if (false === $contents) {
@@ -63,25 +50,4 @@ public function size(): int
     {
         return $this->fileList->count();
     }
-    /**
-     * @return Generator<string>
-     */
-    private function indexPharFile(Phar $phar): Generator
-    {
-        $iterator = new RecursiveIteratorIterator($phar);
-        /** @var PharFileInfo $file */
-        foreach ($iterator as $file) {
-            if (!$file->isFile()) {
-                continue;
-            }
-            if ($file->getExtension() !== 'php') {
-                continue;
-            }
-
-            $this->indexBuilder->index(
-                TextDocumentBuilder::fromUri($file->getPathname())->build()
-            );
-            yield $file->getPathname();
-        }
-    }
 }