EZP-28179: [5.4] Support for selective indexing since date & no-purge (…

…#114)

* EZP-28179: [5.4] Improve index command with support for selective indexing since date

Backporting some of the features added to re-index command in EZP-28176.
First and formost no-purge and since options to be able to re index only part of the index

* CS

* Fix issues with usage of undefined logger service on symfony 2

bundle/Command/SolrCreateIndexCommand.php

@@ -15,9 +15,11 @@
 use Symfony\Component\Console\Input\InputInterface;
 use Symfony\Component\Console\Output\OutputInterface;
 use Symfony\Component\Console\Input\InputArgument;
+use Symfony\Component\Console\Input\InputOption;
 use eZ\Publish\SPI\Persistence\Content\ContentInfo;
 use eZ\Publish\Core\Base\Exceptions\NotFoundException;
 use EzSystems\EzPlatformSolrSearchEngine\Handler as SolrSearchEngineHandler;
+use DateTime;
 use RuntimeException;
 use PDO;
 
@@ -29,6 +31,9 @@ protected function configure()
             ->setName('ezplatform:solr_create_index')
             ->setDescription('Indexes the configured database in configured Solr index')
             ->addArgument('bulk_count', InputArgument::OPTIONAL, 'Number of Content objects indexed at once', 5)
+            ->addOption('no-commit', null, InputOption::VALUE_NONE, 'Do not commit after each bulk iteration')
+            ->addOption('no-purge', null, InputOption::VALUE_NONE, 'Do not purge before indexing, hence rather refresh index')
+            ->addOption('since', null, InputOption::VALUE_OPTIONAL, 'Index changes since a given time, any format understood by DateTime. Implies "no-purge".')
             ->setHelp(
                 <<<EOT
 The command <info>%command.name%</info> indexes current configured database in configured Solr storage.
@@ -40,6 +45,8 @@ protected function execute(InputInterface $input, OutputInterface $output)
     {
         $this->logger = $this->getContainer()->get('logger');
 
+        $commit = !$input->getOption('no-commit');
+        $purge = !$input->getOption('no-purge');
         $bulkCount = $input->getArgument('bulk_count');
         if (!is_numeric($bulkCount) || (int)$bulkCount < 1) {
             throw new RuntimeException("'bulk_count' argument should be > 0, got '{$bulkCount}'");
@@ -61,23 +68,34 @@ protected function execute(InputInterface $input, OutputInterface $output)
 
         // Indexing Content
         $query = $databaseHandler->createSelectQuery();
+        $where = $query->expr->eq('status', ContentInfo::STATUS_PUBLISHED);
+        if ($since = $input->getOption('since')) {
+            $date = new DateTime($since);
+            $where = [
+                $where,
+                $query->expr->gte('modified', $date->getTimestamp()),
+            ];
+            $purge = false;
+        }
+
         $query->select('count(id)')
             ->from('ezcontentobject')
-            ->where($query->expr->eq('status', ContentInfo::STATUS_PUBLISHED));
+            ->where($where);
         $stmt = $query->prepare();
         $stmt->execute();
         $totalCount = $stmt->fetchColumn();
 
         $query = $databaseHandler->createSelectQuery();
         $query->select('id', 'current_version')
             ->from('ezcontentobject')
-            ->where($query->expr->eq('status', ContentInfo::STATUS_PUBLISHED));
-
+            ->where($where);
         $stmt = $query->prepare();
         $stmt->execute();
 
-        /** @var \EzSystems\EzPlatformSolrSearchEngine\Handler $searchHandler */
-        $searchHandler->purgeIndex();
+        if ($purge) {
+            $output->writeln('Purging index before starting re-indexing (use no-purge to skip this)..');
+            $searchHandler->purgeIndex();
+        }
 
         $output->writeln('Indexing Content...');
 
@@ -115,14 +133,16 @@ protected function execute(InputInterface $input, OutputInterface $output)
 
             if (!empty($documents)) {
                 $searchHandler->bulkIndexDocuments($documents);
+
+                if ($commit) {
+                    // Make the bulk changes available for search
+                    $searchHandler->commit();
+                }
             }
 
             $progress->advance($k);
         } while (($i += $bulkCount) < $totalCount);
 
-        // Make changes available for search
-        $searchHandler->commit();
-
         $progress->finish();
         $output->writeln('');
     }

lib/Gateway/HttpClient/Stream.php

@@ -21,7 +21,7 @@
 class Stream implements HttpClient
 {
     /**
-     * @var \Psr\Log\LoggerInterface
+     * @var \Psr\Log\LoggerInterface|null
      */
     private $logger;
 
@@ -43,12 +43,12 @@ class Stream implements HttpClient
     /**
      * Stream constructor.
      *
-     * @param \Psr\Log\LoggerInterface $logger
+     * @param \Psr\Log\LoggerInterface|null $logger
      * @param int $timeout Timeout for connection in seconds.
      * @param int $retry Number of times to re-try connection.
      * @param int $retryWaitMs Time in milli seconds.
      */
-    public function __construct(LoggerInterface $logger, $timeout = 10, $retry = 5, $retryWaitMs = 100)
+    public function __construct(LoggerInterface $logger = null, $timeout = 10, $retry = 5, $retryWaitMs = 100)
     {
         $this->logger = $logger;
         $this->connectionTimeout = $timeout;
@@ -83,9 +83,12 @@ public function request($method, Endpoint $endpoint, $path, Message $message = n
             usleep($this->retryWaitMs * 1000);
         } while ($i < $this->connectionRetry);
 
-        $this->logger->error(
-            sprintf('Connection to %s failed, attempted %d times', $endpoint->getURL(), $this->connectionRetry)
-        );
+        if ($this->logger instanceof LoggerInterface) {
+            $this->logger->error(
+                sprintf('Connection to %s failed, attempted %d times', $endpoint->getURL(), $this->connectionRetry)
+            );
+        }
+
         throw new ConnectionException($endpoint->getURL(), $path, $method);
     }
 

lib/Resources/config/container/solr/services.yml

@@ -8,7 +8,7 @@ parameters:
 services:
     ezpublish.search.solr.gateway.client.http.stream:
         class: "%ezpublish.search.solr.gateway.client.http.stream.class%"
-        arguments: ["@logger"]
+        arguments: ["@?logger"]
 
     # Note: services tagged with 'ezpublish.search.solr.query.content.criterion_visitor'
     # are registered to this one using compilation pass