tests & refactoring

Author: Michał Żarnecki

README.md

@@ -282,6 +282,25 @@ Pick one from l2|cosine|innerProduct or support other one (see https://github.co
     ) {
 ```    
 
+## Test
+This project support static code analysis with PHP Linter, Rector PHP, PHPStan and testing with phpunit.  
+Find below useful commands.
+
+**Run all tests:** \
+`cd ./app/src && composer test`
+
+Run PHP Linter \
+`cd ./app/src && composer lint`
+
+Run Rector PHP \
+`cd ./app/src && composer refactor process`
+
+Run PHPStan: \
+`cd ./app/src && composer test:types`
+
+Run phpunit: \
+`cd ./app/src && composer test:unit`
+
 ## 📚 Resources
 
 - Dataset: "Website Classification" by Hetul Mehta on [Kaggle](https://www.kaggle.com/datasets/hetulmehta/website-classification)

app/src/composer.json

@@ -46,7 +46,7 @@
         "test:lint": "pint --test -v",
         "test:refactor": "rector --dry-run ",
         "test:types": "phpstan analyse --ansi --memory-limit 4G",
-        "test:unit": "cd ./test && php ../vendor/bin/phpunit",
+        "test:unit": "cd ./test && php ../vendor/bin/phpunit .",
         "test": [
             "@test:lint",
             "@test:refactor",

app/src/service/DocumentLoader.php

@@ -39,7 +39,7 @@ public function loadDocuments(): void
             }
 
             foreach ($responseDocumentsChunks as $chunk) {
-                $this->insertDocument($file, $document, $chunk);
+                $this->insertDocument((string) $file, $document, $chunk);
             }
             $this->showProgress($index, $total, $skipFirstN);
         }

app/src/service/DocumentProvider.php

@@ -42,13 +42,14 @@ public function getSimilarDocuments(
     /**
      * Sometimes its worth to use efficient algorithm for querying vector DB and more precise to rerank chosen documents in
      *
-     * @return string[][]
+     * @param  string[][]  $documents
+     * @return string[]
      */
     public function rerank(string $prompt, array $documents): array
     {
         $documentsReranked = [];
         foreach ($documents as $document) {
-            $intersection = array_intersect(explode(' ', (string) $document['text']), explode(' ', $prompt));
+            $intersection = array_intersect(explode(' ', $document['text']), explode(' ', $prompt));
             $index = count($intersection);
             while (isset($documentsReranked[$index])) {
                 $index++;

app/src/service/PromptResolver.php

@@ -11,9 +11,9 @@ final class PromptResolver implements StageInterface
 {
     public function getPromptFromInput(): string
     {
-        $argv = [];
         $prompt = $_POST['prompt'] ?? null;
         if (! $prompt) {
+            global $argv;
             $prompt = $argv[0] ?? null;
         }
         if (! $prompt) {

app/src/service/claude/AbstractClaudeAPIClient.php

@@ -30,7 +30,7 @@ public function __construct()
     }
 
     /**
-     * @return string[]
+     * @return string[][][]
      *
      * @throws \JsonException
      */

app/src/service/deepseek/AbstractDeepSeekAPIClient.php

@@ -27,9 +27,9 @@ public function __construct()
     }
 
     /**
-     * @param  string[]  $messages
+     * @param  string[][]  $messages
      * @param  string[]  $options
-     * @return string[]
+     * @return string[][][][]
      *
      * @throws \JsonException
      */

app/src/service/evaluate/StringComparisonEvaluator.php

@@ -22,7 +22,7 @@ public function calculateBLEU(string $reference, string $candidate, int $n = 1):
                     $matches++;
                 }
             }
-            $nGramMatches[$i] = $matches / max((is_countable($candidateNGrams) ? count($candidateNGrams) : 0), 1);
+            $nGramMatches[$i] = $matches / max(count($candidateNGrams), 1);
         }
 
         $precision = array_product($nGramMatches);
@@ -51,8 +51,8 @@ public function calculateROUGE(string $reference, string $candidate, int $n = 1)
             }
         }
 
-        $recall = $matches / max(is_countable($referenceNGrams) ? count($referenceNGrams) : 0, 1);
-        $precision = $matches / max(is_countable($candidateNGrams) ? count($candidateNGrams) : 0, 1);
+        $recall = $matches / max(count($referenceNGrams), 1);
+        $precision = $matches / max(count($candidateNGrams), 1);
         $f1Score = ($recall + $precision > 0)
             ? 2 * ($recall * $precision) / ($recall + $precision)
             : 0;
@@ -65,6 +65,7 @@ public function calculateROUGE(string $reference, string $candidate, int $n = 1)
     }
 
     /**
+     * @param  string[]  $words
      * @return string[]
      */
     private function getNGrams(array $words, int $n): array

app/src/service/gemini/AbstractGeminiAPIClient.php

@@ -29,6 +29,12 @@ public function __construct()
         ]);
     }
 
+    /**
+     * @param  string[][][][]  $data
+     * @return string[][][][][][]
+     *
+     * @throws \JsonException
+     */
     protected function request(string $method, string $model, array $data): array
     {
         $url = "models/$model:$method?key=".$this->apiKey;

app/src/service/gemini/GeneratedTextFromGeminiProvider.php

@@ -20,9 +20,11 @@ public function generateText(string $prompt, string $sourceDocuments): string
             'generateContent',
             $this->model,
             [
-                'contents' => ['parts' => [[
-                    'text' => $content,
-                ]]],
+                'contents' => [
+                    'parts' => [
+                        ['text' => $content],
+                    ],
+                ],
             ]
         );