IBX-6649: Added support for spell checking (#52)

.github/init_solr.sh

@@ -232,7 +232,11 @@ solr_cloud_configure_collection() {
     # modify solrconfig.xml to remove section that doesn't agree with our schema
     sed -i.bak '/<updateRequestProcessorChain name="add-unknown-fields-to-the-schema".*/,/<\/updateRequestProcessorChain>/d' ${TEMPLATE_DIR}/solrconfig.xml
     # Adapt autoSoftCommit to have a recommended value
-    sed -i.bak2 's/${solr.autoSoftCommit.maxTime:-1}/${solr.autoSoftCommit.maxTime:20}/' "${TEMPLATE_DIR}/solrconfig.xml" || exit_on_error "Can't modify file '${TEMPLATE_DIR}/solrconfig.xml'"
+    sed -i.bak 's/${solr.autoSoftCommit.maxTime:-1}/${solr.autoSoftCommit.maxTime:20}/' "${TEMPLATE_DIR}/solrconfig.xml" || exit_on_error "Can't modify file '${TEMPLATE_DIR}/solrconfig.xml'"
+    # Configure spellcheck component
+    sed -i.bar 's/<str name="field">_text_<\/str>/<str name="field">meta_content__text_t<\/str>/' "${TEMPLATE_DIR}/solrconfig.xml"
+    # Add spellcheck component to /select handler
+    sed -i.bak 's/<requestHandler name="\/select" class="solr.SearchHandler">/<requestHandler name="\/select" class="solr.SearchHandler">\n    <arr name="last-components">\n      <str>spellcheck<\/str>\n    <\/arr>/' "${TEMPLATE_DIR}/solrconfig.xml"
 }
 
 solr_cloud_upload_collection_configuration() {

bin/generate-solr-config.sh

@@ -121,6 +121,10 @@ fi
 # Adapt autoSoftCommit to have a recommended value, and remove add-unknown-fields-to-the-schema
 sed -i.bak '/<updateRequestProcessorChain name="add-unknown-fields-to-the-schema".*/,/<\/updateRequestProcessorChain>/d' $DESTINATION_DIR/solrconfig.xml
 sed -i.bak 's/${solr.autoSoftCommit.maxTime:-1}/${solr.autoSoftCommit.maxTime:20}/' $DESTINATION_DIR/solrconfig.xml
+# Configure spellcheck component
+sed -i.bar 's/<str name="field">_text_<\/str>/<str name="field">meta_content__text_t<\/str>/' $DESTINATION_DIR/solrconfig.xml
+# Add spellcheck component to /select handler
+sed -i.bak 's/<requestHandler name="\/select" class="solr.SearchHandler">/<requestHandler name="\/select" class="solr.SearchHandler">\n    <arr name="last-components">\n      <str>spellcheck<\/str>\n    <\/arr>/' $DESTINATION_DIR/solrconfig.xml
 
 rm $DESTINATION_DIR/solrconfig.xml.bak
 

src/lib/Handler.php

@@ -19,7 +19,7 @@
 use Ibexa\Core\Base\Exceptions\NotFoundException;
 
 /**
- * The Content Search handler retrieves sets of of Content objects, based on a
+ * The Content Search handler retrieves sets of Content objects, based on a
  * set of criteria.
  *
  * The basic idea of this class is to do the following:
@@ -33,7 +33,7 @@
  * sensible queries from all criterion definitions.
  *
  * 3) The query might be possible to optimize (remove empty statements),
- * reduce singular and and or constructs…
+ * reduce singular and or constructs…
  *
  * 4) Additionally we might need a post-query filtering step, which filters
  * content objects based on criteria, which could not be converted in to
@@ -153,7 +153,8 @@ public function findContent(Query $query, array $languageFilter = [])
             $this->gateway->findContent($query, $languageFilter),
             $query->facetBuilders,
             $query->aggregations,
-            $languageFilter
+            $languageFilter,
+            $query->spellcheck
         );
     }
 
@@ -224,7 +225,9 @@ public function findLocations(LocationQuery $query, array $languageFilter = [])
         return $this->locationResultExtractor->extract(
             $this->gateway->findLocations($query, $languageFilter),
             $query->facetBuilders,
-            $query->aggregations
+            $query->aggregations,
+            $languageFilter,
+            $query->spellcheck
         );
     }
 
@@ -478,6 +481,7 @@ public function supports(int $capabilityFlag): bool
             case SearchService::CAPABILITY_SCORING:
             case SearchService::CAPABILITY_FACETS:
             case SearchService::CAPABILITY_CUSTOM_FIELDS:
+            case SearchService::CAPABILITY_SPELLCHECK:
             case SearchService::CAPABILITY_ADVANCED_FULLTEXT:
             case SearchService::CAPABILITY_AGGREGATIONS:
                 return true;

src/lib/Query/Common/QueryConverter/NativeQueryConverter.php

@@ -100,6 +100,13 @@ public function convert(Query $query, array $languageSettings = [])
             }
         }
 
+        if ($query->spellcheck !== null) {
+            $params['spellcheck'] = 'true';
+            $params['spellcheck.q'] = $query->spellcheck->getQuery();
+            $params['spellcheck.count'] = 1;
+            $params['spellcheck.collate'] = 'true';
+        }
+
         return $params;
     }
 

src/lib/ResultExtractor.php

@@ -6,9 +6,11 @@
  */
 namespace Ibexa\Solr;
 
+use Ibexa\Contracts\Core\Repository\Values\Content\Query\Spellcheck;
 use Ibexa\Contracts\Core\Repository\Values\Content\Search\AggregationResultCollection;
 use Ibexa\Contracts\Core\Repository\Values\Content\Search\SearchHit;
 use Ibexa\Contracts\Core\Repository\Values\Content\Search\SearchResult;
+use Ibexa\Contracts\Core\Repository\Values\Content\Search\SpellcheckResult;
 use Ibexa\Contracts\Solr\ResultExtractor\AggregationResultExtractor;
 use Ibexa\Solr\Gateway\EndpointRegistry;
 use Ibexa\Solr\Query\FacetFieldVisitor;
@@ -49,8 +51,13 @@ public function __construct(
      *
      * @return \Ibexa\Contracts\Core\Repository\Values\Content\Search\SearchResult
      */
-    public function extract($data, array $facetBuilders = [], array $aggregations = [], array $languageFilter = [])
-    {
+    public function extract(
+        $data,
+        array $facetBuilders = [],
+        array $aggregations = [],
+        array $languageFilter = [],
+        ?Spellcheck $spellcheck = null
+    ) {
         $result = new SearchResult(
             [
                 'time' => $data->responseHeader->QTime / 1000,
@@ -61,6 +68,7 @@ public function extract($data, array $facetBuilders = [], array $aggregations =
 
         $result->facets = $this->extractFacets($data, $facetBuilders, $languageFilter);
         $result->aggregations = $this->extractAggregations($data, $aggregations, $languageFilter);
+        $result->spellcheck = $this->extractSpellcheck($data, $spellcheck);
 
         foreach ($data->response->docs as $doc) {
             $result->searchHits[] = $this->extractSearchHit($doc, $languageFilter);
@@ -186,6 +194,22 @@ protected function extractSearchHit(stdClass $doc, array $languageFilter): Searc
             ]
         );
     }
+
+    protected function extractSpellcheck(stdClass $data, ?Spellcheck $spellcheck): ?SpellcheckResult
+    {
+        if ($spellcheck === null) {
+            return null;
+        }
+
+        if (isset($data->spellcheck)) {
+            $incorrect = !empty($data->spellcheck->collations);
+            $query = $data->spellcheck->collations[1] ?? $spellcheck->getQuery();
+
+            return new SpellcheckResult($query, $incorrect);
+        }
+
+        return new SpellcheckResult($spellcheck->getQuery(), false);
+    }
 }
 
 class_alias(ResultExtractor::class, 'EzSystems\EzPlatformSolrSearchEngine\ResultExtractor');