Add new serach criterion to allow proper replication of ezfind searches

API/Repository/Values/Content/Query/Criterion/EzFindText.php

@@ -0,0 +1,19 @@
+<?php
+
+namespace Kaliop\EzFindSearchEngineBundle\API\Repository\Values\Content\Query\Criterion;
+
+use eZ\Publish\API\Repository\Values\Content\Query\Criterion as BaseCriterion;
+
+/**
+ * Full-text search criterion, based on eZFind configuration.
+ * When this criterion is used (NB: as part of a query's `query`, not `filter`), the given text will be matched against
+ * all content fields indexed in Solr as per eZFind configuration (including field boosts etc...)
+ * Note: when using the FullTextr criterion instead of this one, matching is done against the `ezf_df_text` Solr field
+ */
+class EzFindText extends BaseCriterion
+{
+    public function __construct($value)
+    {
+        $this->value = $value;
+    }
+}

Core/Persistence/eZFind/Content/Search/Common/Gateway/CriterionHandler/EzFindText.php

@@ -0,0 +1,60 @@
+<?php
+
+namespace Kaliop\EzFindSearchEngineBundle\Core\Persistence\eZFind\Content\Search\Common\Gateway\CriterionHandler;
+
+use eZ\Publish\API\Repository\Values\Content\Query\Criterion;
+use Kaliop\EzFindSearchEngineBundle\Core\Persistence\eZFind\Content\Search\Common\Gateway\CriteriaConverter;
+use Kaliop\EzFindSearchEngineBundle\Core\Persistence\eZFind\Content\Search\Common\Gateway\CriterionHandler;
+use Kaliop\EzFindSearchEngineBundle\API\Repository\Values\Content\Query\Criterion\EzFindText as EzFindTextCriterion;
+
+class EzFindText extends CriterionHandler
+{
+    /**
+     * @inheritdoc
+     */
+    public function accept(Criterion $criterion)
+    {
+        return $criterion instanceof EzFindTextCriterion;
+    }
+
+    /**
+     * If the full-text search contain wildcard search, build correct wildcard query
+     * with non-truncated words boosted.
+     *
+     * @inheritdoc
+     */
+    public function handle(CriteriaConverter $converter, Criterion $criterion)
+    {
+        $value = trim($criterion->value);
+
+        if (preg_match('/^".+"$/', $value)) {
+            // Quoted-string query: escape everything but the outher quotes
+            $value = '"' . $this->escapeValue(substr($value, 1, -1)) . '"';
+
+        } else if (preg_match('/(^\*|\*$)/', $value)) {
+            // Wildcard query: make the exact match stronger than the wildcard
+
+            // @bug we do not support wildcard chars in the middle of phrases
+
+            $value = $this->escapeValue($value);
+
+            // Escape spaces
+            $value = str_replace(' ', '\\ ', $value);
+
+            // wildcard match: un-escape wildcard char
+            $wildcard = str_replace('\\*', '*', $value);
+
+            // Non-wildcard match
+            $value = trim($value, '*');
+            $value = rtrim($value, '\\');
+
+            $value = $value . '^2 OR ' . $wildcard;
+
+        } else {
+            // plain query
+            $value = $this->escapeValue($value);
+        }
+
+        return $value;
+    }
+}

Core/Persistence/eZFind/Content/Search/Common/Gateway/CriterionHandler/FullText.php

@@ -35,7 +35,7 @@ public function handle(CriteriaConverter $converter, Criterion $criterion)
             $value = '"' . $this->escapeValue(substr($value, 1, -1)) . '"';
 
         } else if (preg_match('/(^\*|\*$)/', $value)) {
-            // Wildcard query: make the exacth match stronger than the wildcard
+            // Wildcard query: make the exact match stronger than the wildcard
 
             // @bug we do not support wildcard chars in the middle of phrases
 

Core/Repository/SearchService.php

@@ -241,7 +241,7 @@ protected function performSearch(
         $this->initializeQueryLimit($query);
 
         $searchParameters = $this->getLegacySearchParameters($query, $fieldFilters, $filterOnUserPermissions, $returnType);
-
+//var_dump($searchParameters);
         /** @var array $searchResult */
         $searchResult = $this->getLegacyKernel()->runCallback(
             function () use ($searchParameters) {
@@ -315,7 +315,7 @@ protected function getLegacySearchParameters(Query $query, array $fieldFilters,
                 $scoreSort = true;
             }
         }
-
+//var_dump($query->criterion);die();
         $criterionFilter = array();
         if ($query->criterion) {
             $criterionFilter = $this->extractFilter($query->criterion);
@@ -330,7 +330,10 @@ protected function getLegacySearchParameters(Query $query, array $fieldFilters,
             $searchParameters['facet'] = $this->extractFacet($query->facetBuilders);
         }
 
-        if ($scoreSort) {
+        if ($this->isEzFindCriterion($query->criterion)) {
+            $searchParameters['query'] = reset($criterionFilter);
+            $searchParameters['filter'] = $filterFilter;
+        } elseif ($scoreSort) {
             // since we are sorting by score, we need to generate the solr query, as that is what is used to calculate score
             $searchParameters['query'] = $this->filterCriteriaConverter->generateQueryString($criterionFilter);
             $searchParameters['filter'] = $filterFilter;
@@ -389,6 +392,20 @@ protected function getReturnType(Query $query, $forceReturnType = null)
         return ($query instanceof KaliopQuery && $query->returnType !== null) ? $query->returnType : $this->defaultReturnType;
     }
 
+    /**
+     * Returns true if there is a single search criterion of type EzFindText
+     * @param Query\Criterion|Query\Criterion[] $criteria $criteria
+     * @return bool
+     */
+    protected function isEzFindCriterion($criteria)
+    {
+        if (!is_array($criteria)) {
+            $criteria = array($criteria);
+        }
+
+        return (count($criteria) == 1 && $criteria[0] instanceof KaliopQuery\Criterion\EzFindText);
+    }
+
     /**
      * @param Query\Criterion|Query\Criterion[] $criteria
      * @return array
@@ -447,7 +464,7 @@ protected function throwIfSearchError($searchResult)
         if (!is_array($searchResult)) {
             throw new eZFindException('The legacy search result is not an array');
         }
-
+//var_dump($searchResult);die();
         if (isset($searchResult['SearchExtras']) && $searchResult['SearchExtras'] instanceof ezfSearchResultInfo) {
             $errors = $searchResult['SearchExtras']->attribute('error');
             /// @todo what if $errors it is an empty string, an array with unexepcted members or not even an array ?

Resources/config/handlers/criteria.yml

@@ -37,6 +37,11 @@ services:
         tags:
             - {name: ezfind_search_engine.content.criterion_handler.filter}
 
+    ezfind_search_engine.content.criterion_handler.ezfind_text:
+        class: Kaliop\EzFindSearchEngineBundle\Core\Persistence\eZFind\Content\Search\Common\Gateway\CriterionHandler\EzFindText
+        tags:
+            - {name: ezfind_search_engine.content.criterion_handler.filter}
+
     ezfind_search_engine.content.criterion_handler.location_id:
         class: Kaliop\EzFindSearchEngineBundle\Core\Persistence\eZFind\Content\Search\Common\Gateway\CriterionHandler\LocationId
         tags:

TODO.md

@@ -4,6 +4,9 @@
   - fulltext: might currently differ
   - objectstateid: what happens when many states are passed in?
 
+* check what happens when an ezfindtext criterion is used in the 'filter' member of a query, or when it is used together
+  with other filters 
+
 * check list of all missing criteria converters compared to std search service (for Content searches)
   - contenttypegroup: can not do
   - datemetadata: fails with operator IN

WHATSNEW.md

@@ -1,3 +1,21 @@
+Version 1.4
+===========
+
+* New: a new Query Criterion, `EzFindText`, can now be used to fully emulated legacy 'ezfind searches', i.e. do a
+    full-text search across all indexed content fields, instead of matching the `ezf_df_text` field as is done when using
+    a `FullText` criterion. This can greatly improve relevancy sorting.
+
+    Ex:
+
+        use Kaliop\EzFindSearchEngineBundle\API\Repository\Values\Content\Query\Criterion\EzFindText;
+        use Kaliop\EzFindSearchEngineBundle\API\Repository\Values\Content\Query\SortClause\Score; 
+        ...
+        $query->query = new EzFindText('hello world');
+        $query->sortClauses = [new Score('descending')];
+
+    *NB* the `EzFindText` criterion should be used as unique member of your query's `query` member.
+    When using it, use the query's `filter` member to add any other criteria.
+
 Version 1.3
 ===========