refactor: extract SolrIndexService

sitepark · Mar 12, 2024 · 299485a · 299485a
1 parent c1b2473
commit 299485a
Show file tree

Hide file tree

Showing 17 changed files with 364 additions and 205 deletions.
diff --git a/src/Console/Command/Indexer.php b/src/Console/Command/Indexer.php
@@ -33,7 +33,7 @@ class Indexer extends Command
      */
     public function __construct(
         private readonly iterable $documentEnricherList,
-        private readonly SolrIndexerBuilder $solrIndexerBuilder,
+        private readonly InternalResourceIndexerBuilder $solrIndexerBuilder,
         private readonly IndexerProgressBarFactory $progressBarFactory
     ) {
         parent::__construct();

diff --git a/src/Console/Command/SolrIndexerBuilder.php → ...ommand/InternalResourceIndexerBuilder.php b/src/Console/Command/SolrIndexerBuilder.php → ...ommand/InternalResourceIndexerBuilder.php
@@ -11,16 +11,17 @@
 use Atoolo\Search\Service\Indexer\DocumentEnricher;
 use Atoolo\Search\Service\Indexer\IndexDocument;
 use Atoolo\Search\Service\Indexer\IndexingAborter;
+use Atoolo\Search\Service\Indexer\InternalResourceIndexer;
 use Atoolo\Search\Service\Indexer\LocationFinder;
 use Atoolo\Search\Service\Indexer\SiteKit\ContentMatcher;
 use Atoolo\Search\Service\Indexer\SiteKit\DefaultSchema2xDocumentEnricher;
 use Atoolo\Search\Service\Indexer\SiteKit\HeadlineMatcher;
 use Atoolo\Search\Service\Indexer\SiteKit\RichtTextMatcher;
 use Atoolo\Search\Service\Indexer\SiteKit\SubDirTranslationSplitter;
-use Atoolo\Search\Service\Indexer\SolrIndexer;
+use Atoolo\Search\Service\Indexer\SolrIndexService;
 use Atoolo\Search\Service\SolrParameterClientFactory;
 
-class SolrIndexerBuilder
+class InternalResourceIndexerBuilder
 {
     private string $resourceDir;
     /**
@@ -35,8 +36,9 @@ public function __construct(
         private readonly ResourceBaseLocatorBuilder $resourceBaseLocatorBuilder
     ) {
     }
-    public function resourceDir(string $resourceDir): SolrIndexerBuilder
-    {
+    public function resourceDir(
+        string $resourceDir
+    ): InternalResourceIndexerBuilder {
         $this->resourceDir = $resourceDir;
         return $this;
     }
@@ -47,26 +49,26 @@ public function resourceDir(string $resourceDir): SolrIndexerBuilder
      */
     public function documentEnricherList(
         iterable $documentEnricherList
-    ): SolrIndexerBuilder {
+    ): InternalResourceIndexerBuilder {
         $this->documentEnricherList = $documentEnricherList;
         return $this;
     }
 
     public function progressBar(
         IndexerProgressBar $progressBar
-    ): SolrIndexerBuilder {
+    ): InternalResourceIndexerBuilder {
         $this->progressBar = $progressBar;
         return $this;
     }
 
     public function solrConnectionUrl(
         string $solrConnectionUrl
-    ): SolrIndexerBuilder {
+    ): InternalResourceIndexerBuilder {
         $this->solrConnectionUrl = $solrConnectionUrl;
         return $this;
     }
 
-    public function build(): SolrIndexer
+    public function build(): InternalResourceIndexer
     {
         $resourceBaseLocator = $this->resourceBaseLocatorBuilder->build(
             $this->resourceDir
@@ -106,17 +108,19 @@ public function build(): SolrIndexer
             0
         );
 
+        $solrIndexService = new SolrIndexService($clientFactory);
+
         $translationSplitter = new SubDirTranslationSplitter();
 
         $aborter = new IndexingAborter('.');
 
-        return new SolrIndexer(
+        return new InternalResourceIndexer(
             $documentEnricherList,
             $this->progressBar,
             $finder,
             $resourceLoader,
             $translationSplitter,
-            $clientFactory,
+            $solrIndexService,
             $aborter,
             'internal'
         );

diff --git a/src/Service/Indexer/BackgroundIndexer.php b/src/Service/Indexer/BackgroundIndexer.php
@@ -16,7 +16,7 @@
 class BackgroundIndexer implements Indexer
 {
     public function __construct(
-        private readonly SolrIndexerFactory $indexerFactory,
+        private readonly InternalResourceIndexerFactory $indexerFactory,
         private readonly IndexerStatusStore $statusStore,
         private readonly LoggerInterface $logger = new NullLogger(),
         private readonly LockFactory $lockFactory = new LockFactory(
@@ -59,7 +59,7 @@ public function getStatus(string $index): IndexerStatus
         return $this->statusStore->load($index);
     }
 
-    private function getIndexer(string $index): SolrIndexer
+    private function getIndexer(string $index): InternalResourceIndexer
     {
         $progressHandler = new BackgroundIndexerProgressState(
             $index,

diff --git a/src/Service/Indexer/IndexDocumentDumper.php b/src/Service/Indexer/IndexDocumentDumper.php
@@ -19,7 +19,9 @@ public function __construct(
 
     /**
      * @param string[] $paths
-     * @return array<int,array<string,mixed>>.
+     * @return array<int,array<string,mixed>>
+     *    Returns the raw array data of the documents to be able to
+     *    output them as JSON, for example.
      */
     public function dump(array $paths): array
     {

diff --git a/src/Service/Indexer/SolrIndexer.php → ...rvice/Indexer/InternalResourceIndexer.php b/src/Service/Indexer/SolrIndexer.php → ...rvice/Indexer/InternalResourceIndexer.php
@@ -9,7 +9,6 @@
 use Atoolo\Search\Dto\Indexer\IndexerParameter;
 use Atoolo\Search\Dto\Indexer\IndexerStatus;
 use Atoolo\Search\Indexer;
-use Atoolo\Search\Service\SolrClientFactory;
 use Exception;
 use Solarium\QueryType\Update\Result as UpdateResult;
 use Throwable;
@@ -35,7 +34,7 @@
  *    but always a list of documents. The entire list is divided into chunks
  *    and indexed chunk-wise.
  */
-class SolrIndexer implements Indexer
+class InternalResourceIndexer implements Indexer
 {
     /**
      * @param iterable<DocumentEnricher<IndexDocument>> $documentEnricherList
@@ -46,7 +45,7 @@ public function __construct(
         private readonly LocationFinder $finder,
         private readonly ResourceLoader $resourceLoader,
         private readonly TranslationSplitter $translationSplitter,
-        private readonly SolrClientFactory $clientFactory,
+        private readonly SolrIndexService $indexService,
         private readonly IndexingAborter $aborter,
         private readonly string $source
     ) {
@@ -61,8 +60,8 @@ public function remove(string $index, array $idList): void
             return;
         }
 
-        $this->deleteByIdList($index, $idList);
-        $this->commit($index);
+        $this->indexService->deleteByIdList($index, $this->source, $idList);
+        $this->indexService->commit($index);
     }
 
     public function abort(string $index): void
@@ -137,7 +136,7 @@ private function indexResources(
             $this->indexerProgressHandler->startUpdate($total);
         }
 
-        $availableIndexes = $this->getAvailableIndexes();
+        $availableIndexes = $this->indexService->getAvailableIndexes();
         $splitterResult = $this->translationSplitter->split($pathList);
 
         $this->indexTranslationSplittedResources(
@@ -232,9 +231,13 @@ private function indexResourcesPerLanguageIndex(
             $parameter->cleanupThreshold > 0 &&
             $successCount >= $parameter->cleanupThreshold
         ) {
-            $this->deleteByProcessId($index, $processId);
+            $this->indexService->deleteExcludingProcessId(
+                $index,
+                $this->source,
+                $processId
+            );
         }
-        $this->commit($index);
+        $this->indexService->commit($index);
     }
 
     /**
@@ -322,12 +325,9 @@ private function add(
         string $processId,
         array $resources
     ): UpdateResult {
-        $client = $this->clientFactory->create($solrCore);
 
-        $update = $client->createUpdate();
-        $update->setDocumentClass(IndexSchema2xDocument::class);
+        $updater = $this->indexService->updater($solrCore);
 
-        $documents = [];
         foreach ($resources as $resource) {
             foreach ($this->documentEnricherList as $enricher) {
                 if (!$enricher->isIndexable($resource)) {
@@ -337,89 +337,29 @@ private function add(
             }
             try {
                 /** @var IndexSchema2xDocument $doc */
-                $doc = $update->createDocument();
+                $doc = $updater->createDocument();
                 foreach ($this->documentEnricherList as $enricher) {
                     $doc = $enricher->enrichDocument(
                         $resource,
                         $doc,
                         $processId
                     );
                 }
-                $documents[] = $doc;
+                $updater->addDocument($doc);
             } catch (Throwable $e) {
                 $this->indexerProgressHandler->error($e);
             }
         }
-        // add the documents and a commit command to the update query
-        $update->addDocuments($documents);
 
         // this executes the query and returns the result
-        return $client->update($update);
-    }
-
-    private function deleteByProcessId(string $core, string $processId): void
-    {
-        $this->deleteByQuery(
-            $core,
-            '-crawl_process_id:' . $processId . ' AND ' .
-            ' sp_source:' . $this->source
-        );
-    }
-
-    /**
-     * @param string[] $idList
-     */
-    private function deleteByIdList(string $core, array $idList): void
-    {
-        $this->deleteByQuery(
-            $core,
-            'sp_id:(' . implode(' ', $idList) . ') AND ' .
-            'sp_source:' . $this->source
-        );
+        return $updater->update();
     }
 
     private function deleteErrorProtocol(string $core): void
     {
-        $this->deleteByQuery(
+        $this->indexService->deleteByQuery(
             $core,
             'crawl_status:error OR crawl_status:warning'
         );
     }
-
-    private function deleteByQuery(string $core, string $query): void
-    {
-        $client = $this->clientFactory->create($core);
-        $update = $client->createUpdate();
-        $update->addDeleteQuery($query);
-        $client->update($update);
-    }
-
-    private function commit(string $core): void
-    {
-        $client = $this->clientFactory->create($core);
-        $update = $client->createUpdate();
-        $update->addCommit();
-        $update->addOptimize();
-        $client->update($update);
-    }
-
-    /**
-     * @return string[]
-     */
-    private function getAvailableIndexes(): array
-    {
-        $client = $this->clientFactory->create('');
-        $coreAdminQuery = $client->createCoreAdmin();
-        $statusAction = $coreAdminQuery->createStatus();
-        $coreAdminQuery->setAction($statusAction);
-
-        $availableIndexes = [];
-        $response = $client->coreAdmin($coreAdminQuery);
-        $statusResults = $response->getStatusResults() ?? [];
-        foreach ($statusResults as $statusResult) {
-            $availableIndexes[] = $statusResult->getCoreName();
-        }
-
-        return $availableIndexes;
-    }
 }
diff --git a/src/Service/Indexer/SolrIndexerFactory.php → ...ndexer/InternalResourceIndexerFactory.php b/src/Service/Indexer/SolrIndexerFactory.php → ...ndexer/InternalResourceIndexerFactory.php
@@ -5,9 +5,8 @@
 namespace Atoolo\Search\Service\Indexer;
 
 use Atoolo\Resource\ResourceLoader;
-use Atoolo\Search\Service\SolrClientFactory;
 
-class SolrIndexerFactory
+class InternalResourceIndexerFactory
 {
     /**
      * @param iterable<DocumentEnricher<IndexDocument>> $documentEnricherList
@@ -17,22 +16,22 @@ public function __construct(
         private readonly LocationFinder $finder,
         private readonly ResourceLoader $resourceLoader,
         private readonly TranslationSplitter $translationSplitter,
-        private readonly SolrClientFactory $clientFactory,
+        private readonly SolrIndexService $solrService,
         private readonly IndexingAborter $aborter,
         private readonly string $source
     ) {
     }
 
     public function create(
         IndexerProgressHandler $progressHandler
-    ): SolrIndexer {
-        return new SolrIndexer(
+    ): InternalResourceIndexer {
+        return new InternalResourceIndexer(
             $this->documentEnricherList,
             $progressHandler,
             $this->finder,
             $this->resourceLoader,
             $this->translationSplitter,
-            $this->clientFactory,
+            $this->solrService,
             $this->aborter,
             $this->source
         );

diff --git a/src/Service/Indexer/SiteKit/DefaultSchema2xDocumentEnricher.php b/src/Service/Indexer/SiteKit/DefaultSchema2xDocumentEnricher.php
@@ -378,7 +378,6 @@ private function idWithoutSignature(string $id): int
      * - https://gitlab.sitepark.com/customer-projects/stadtundland/blob/develop/stadtundland-module/src/publish/php/SP/Stadtundland/Component/PurchaseExpose.php#L38
      * - https://gitlab.sitepark.com/ies-modules/citycall/blob/develop/citycall-module/src/main/php/src/SP/CityCall/Component/Intro.php#L51
      * - https://gitlab.sitepark.com/ies-modules/sitekit-real-estate/blob/develop/src/publish/php/SP/RealEstate/Component/Expose.php#L47
-     * - https://gitlab.sitepark.com/sitekit/xzufi-php/-/blob/develop/php/SP/Xzufi/Renderer/SolrData.php?ref_type=heads#L78
      */
 
     private function getLocaleFromResource(Resource $resource): string