p43A

src/Config/NdjsonHybridRetrieverConfig.php

@@ -13,6 +13,7 @@ final class NdjsonHybridRetrieverConfig
      */
     public function __construct(
         private array $config = [],
+        private ?DomainVocabularyConfig $vocabulary = null,
     ) {
     }
 
@@ -216,55 +217,82 @@ final class NdjsonHybridRetrieverConfig
     /** @return string[] */
     public function genericProductTokens(): array
     {
-        return $this->requiredStringList('generic_product_tokens');
+        return $this->configuredStringListOrVocabularyView(
+            'generic_product_tokens',
+            'vocabulary_views.generic_product_tokens'
+        );
     }
 
     /** @return string[] */
     public function importantShortModelTokens(): array
     {
-        return $this->requiredStringList('important_short_model_tokens');
+        return $this->configuredStringListOrVocabularyView(
+            'important_short_model_tokens',
+            'vocabulary_views.important_short_model_tokens'
+        );
     }
 
     /** @return string[] */
     public function familyDescriptorTokens(): array
     {
-        return $this->requiredStringList('family_descriptor_tokens');
+        return $this->configuredStringListOrVocabularyView(
+            'family_descriptor_tokens',
+            'vocabulary_views.family_descriptor_tokens'
+        );
     }
 
     /** @return string[] */
     public function looksLikeReagentTokens(): array
     {
-        return $this->requiredStringList('looks_like_reagent_tokens');
+        return $this->configuredStringListOrVocabularyView(
+            'looks_like_reagent_tokens',
+            'vocabulary_views.looks_like_reagent_tokens'
+        );
     }
 
     /** @return string[] */
     public function looksLikeSafetyDocs(): array
     {
-        return $this->requiredStringList('looks_like_safety_docs');
+        return $this->configuredStringListOrVocabularyView(
+            'looks_like_safety_docs',
+            'vocabulary_views.looks_like_safety_docs'
+        );
     }
 
     /** @return string[] */
     public function looksLikeReagentWords(): array
     {
-        return $this->requiredStringList('looks_like_reagent_words');
+        return $this->configuredStringListOrVocabularyView(
+            'looks_like_reagent_words',
+            'vocabulary_views.looks_like_reagent_words'
+        );
     }
 
     /** @return string[] */
     public function looksLikeDocumentWords(): array
     {
-        return $this->requiredStringList('looks_like_document_words');
+        return $this->configuredStringListOrVocabularyView(
+            'looks_like_document_words',
+            'vocabulary_views.looks_like_document_words'
+        );
     }
 
     /** @return string[] */
     public function looksLikeSafetyWords(): array
     {
-        return $this->requiredStringList('looks_like_safety_words');
+        return $this->configuredStringListOrVocabularyView(
+            'looks_like_safety_words',
+            'vocabulary_views.looks_like_safety_words'
+        );
     }
 
     /** @return string[] */
     public function looksLikeDeviceWords(): array
     {
-        return $this->requiredStringList('looks_like_device_words');
+        return $this->configuredStringListOrVocabularyView(
+            'looks_like_device_words',
+            'vocabulary_views.looks_like_device_words'
+        );
     }
 
     /**
@@ -471,6 +499,74 @@ final class NdjsonHybridRetrieverConfig
         return $out;
     }
 
+
+    /** @return string[] */
+    private function configuredStringListOrVocabularyView(string $configPath, string $viewPathConfigPath): array
+    {
+        if ($this->hasKey($configPath)) {
+            return $this->requiredStringList($configPath);
+        }
+
+        if ($this->vocabulary === null) {
+            throw $this->missing($configPath);
+        }
+
+        $viewPath = $this->requiredPathString($viewPathConfigPath);
+        $terms = $this->vocabulary->view($viewPath, []);
+
+        if ($terms === []) {
+            throw $this->invalid($viewPathConfigPath, sprintf('references empty vocabulary view "%s"', $viewPath));
+        }
+
+        return $terms;
+    }
+
+    private function requiredPathString(string $key): string
+    {
+        $value = $this->requiredPathValue($key);
+
+        if (!is_scalar($value)) {
+            throw $this->invalid($key, 'must be a non-empty string');
+        }
+
+        $value = trim((string) $value);
+        if ($value === '') {
+            throw $this->invalid($key, 'must be a non-empty string');
+        }
+
+        return $value;
+    }
+
+    private function requiredPathValue(string $key): mixed
+    {
+        $current = $this->config;
+
+        foreach (explode('.', $key) as $segment) {
+            if (!is_array($current) || !array_key_exists($segment, $current)) {
+                throw $this->missing($key);
+            }
+
+            $current = $current[$segment];
+        }
+
+        return $current;
+    }
+
+    private function hasKey(string $key): bool
+    {
+        $current = $this->config;
+
+        foreach (explode('.', $key) as $segment) {
+            if (!is_array($current) || !array_key_exists($segment, $current)) {
+                return false;
+            }
+
+            $current = $current[$segment];
+        }
+
+        return true;
+    }
+
     private function requiredValue(string $key): mixed
     {
         if (!array_key_exists($key, $this->config)) {