PHPSTAN refactoring

Author: Michał Żarnecki

app/src/service/DocumentLoader.php

@@ -15,15 +15,15 @@ public function __construct(
     public function loadDocuments(): void
     {
         $path = __DIR__.'/../documents';
-        $files = array_diff(scandir($path), ['.', '..']);
+        $files = array_diff((array) scandir($path), ['.', '..']);
         $total = count($files);
 
         $skipFirstN = 1330; //replace this to skip N documents for debugging
         foreach ($files as $index => $file) {
             if ($index < $skipFirstN) {
                 continue;
             }
-            $document = file_get_contents($path.'/'.$file);
+            $document = (string) file_get_contents($path.'/'.$file);
             //cut document to tokens limit 8192
             if (str_word_count($document) >= 0.75 * 8000) {
                 $words = explode(' ', $document);
@@ -34,6 +34,8 @@ public function loadDocuments(): void
                 $responseDocumentsChunks = $this->textEncoder->getEmbeddings($document);
             } catch (\Throwable $e) {
                 error_log($e->getMessage());
+
+                return;
             }
 
             foreach ($responseDocumentsChunks as $chunk) {
@@ -46,13 +48,14 @@ public function loadDocuments(): void
 
     private function showProgress(int|string $index, int $total, int $skip): void
     {
+        if (! is_int($index)) {
+            return;
+        }
         $all = $total - $skip;
         $numLoaded = $index - $skip + 1;
         $progress = '';
         if ($numLoaded % 10 === 0) {
-            for ($i = 0; $i < $numLoaded / 10; $i++) {
-                $progress .= '-';
-            }
+            $progress = str_repeat('-', $numLoaded / 10);
             fwrite(STDOUT, "{$progress}\nLoaded {$numLoaded} of {$all} source documents to vector DB\n");
         }
     }

app/src/service/DocumentProvider.php

@@ -11,6 +11,7 @@ final class DocumentProvider extends AbstractDocumentRepository implements Stage
 {
     /**
      * @param  string  $distanceFunction  l2|cosine|innerProduct
+     * @return string[][]
      */
     public function getSimilarDocuments(
         string $prompt,
@@ -40,6 +41,8 @@ public function getSimilarDocuments(
 
     /**
      * Sometimes its worth to use efficient algorithm for querying vector DB and more precise to rerank chosen documents in
+     *
+     * @return string[][]
      */
     public function rerank(string $prompt, array $documents): array
     {

app/src/service/RAGPromptProvider.php

@@ -12,6 +12,9 @@ final class RAGPromptProvider implements StageInterface
 {
     public const CONTEXT_TOKEN_COUNT = 5000;
 
+    /**
+     * @param  string[]  $documents
+     */
     public function getRAGPrompt(array $documents, string $prompt): string
     {
         $contextTokenCount = self::CONTEXT_TOKEN_COUNT - TokenizerX::count($prompt) - 20;

app/src/service/TextSplitter.php

@@ -4,6 +4,9 @@
 
 class TextSplitter
 {
+    /**
+     * @return string[]
+     */
     public function splitDocumentIntoChunks(string $document, int $chunkSize, int $overlap): array
     {
         $chunks = [];

app/src/service/claude/AbstractClaudeAPIClient.php

@@ -29,6 +29,11 @@ public function __construct()
         ]);
     }
 
+    /**
+     * @return string[]
+     *
+     * @throws \JsonException
+     */
     protected function request(string $prompt, string $endpoint, string $model): array
     {
         $url = $endpoint;

app/src/service/claude/GeneratedTextFromClaudeProvider.php

@@ -4,6 +4,7 @@
 
 use League\Pipeline\StageInterface;
 use service\GeneratedTextProviderInterface;
+use service\pipeline\Payload;
 
 final class GeneratedTextFromClaudeProvider extends AbstractClaudeAPIClient implements GeneratedTextProviderInterface, StageInterface
 {
@@ -23,9 +24,8 @@ public function generateText(string $prompt, string $sourceDocuments): string
 
     /**
      * @param  Payload  $payload
-     * @return string
      */
-    public function __invoke($payload)
+    public function __invoke($payload): string
     {
         return $this->generateText($payload->getPrompt(), $payload->getRagPrompt());
     }

app/src/service/deepseek/AbstractDeepSeekAPIClient.php

@@ -26,6 +26,13 @@ public function __construct()
         ]);
     }
 
+    /**
+     * @param  string[]  $messages
+     * @param  string[]  $options
+     * @return string[]
+     *
+     * @throws \JsonException
+     */
     protected function request(array $messages, string $model, bool $stream = false, array $options = []): array
     {
         $data = array_merge([

app/src/service/evaluate/StringComparisonEvaluator.php

@@ -4,7 +4,7 @@
 
 class StringComparisonEvaluator
 {
-    public function calculateBLEU(string $reference, string $candidate, $n = 1): float
+    public function calculateBLEU(string $reference, string $candidate, int $n = 1): float
     {
         $candidateWords = explode(' ', $candidate);
         $referenceWords = explode(' ', $reference);
@@ -22,7 +22,7 @@ public function calculateBLEU(string $reference, string $candidate, $n = 1): flo
                     $matches++;
                 }
             }
-            $nGramMatches[$i] = $matches / max(is_countable($candidateNGrams) ? count($candidateNGrams) : 0, 1);
+            $nGramMatches[$i] = $matches / max((is_countable($candidateNGrams) ? count($candidateNGrams) : 0), 1);
         }
 
         $precision = array_product($nGramMatches);

app/src/test/TextSplitterTest.php

@@ -13,7 +13,7 @@ public function testTextSplitter_keepsCommonPartBetweenChunks(): void
     {
         $textSplitter = new TextSplitter();
         $chunks = $textSplitter->splitDocumentIntoChunks($this->getLoremIpsum(), 300, 100);
-        $this->assertEquals(substr((string) $chunks[0], strlen((string) $chunks[0]) - 100), substr((string) $chunks[1], 0, 100));
+        $this->assertEquals(substr($chunks[0], strlen($chunks[0]) - 100), substr($chunks[1], 0, 100));
     }
 
     public function testTextSplitter_splitsToExpectedNumberOfChunks(): void