Use a weighted search

conf/schema.xml

@@ -122,59 +122,11 @@
         <field name="fulltext_id_success" type="string" indexed="true" stored="true" multiValued="true"/>
         <field name="fulltext_id_failure" type="string" indexed="true" stored="true" multiValued="true"/>
 
-        <!--
-            Catchall field, containing all other searchable text fields (implemented
-            via copyField further on in this schema)
-            indexes tokens both normally and in reverse for efficient
-            leading wildcard queries.
-        -->
-        <field name="text" type="text" indexed="true" stored="false" multiValued="true" omitNorms="false"/>
-
-        <!-- Catchall field for simple search without fulltext. -->
-        <field name="simple" type="text" indexed="true" stored="false" multiValued="true" omitNorms="false"/>
-
         <dynamicField name="enrichment_*" multiValued="true" stored="true" indexed="true" type="string" />
     </fields>
 
     <uniqueKey>id</uniqueKey>
 
-    <!--
-        Copy all searchable text fields into catchall field for keyword searches across everything.
-
-        TODO Werden alle Titel und Zusammenfassungen indiziert oder nur die in der Sprache des Dokuments?
-    -->
-    <copyField source="abstract" dest="text"/>
-    <copyField source="title" dest="text"/>
-    <copyField source="author" dest="text"/>
-    <copyField source="subject" dest="text"/>
-    <copyField source="title_parent" dest="text"/>
-    <copyField source="title_additional" dest="text"/>
-    <copyField source="title_sub" dest="text"/>
-    <copyField source="creating_corporation" dest="text"/>
-    <copyField source="contributing_corporation" dest="text"/>
-    <copyField source="publisher_name" dest="text"/>
-    <copyField source="publisher_place" dest="text"/>
-    <copyField source="identifier" dest="text"/>
-    <copyField source="persons" dest="text"/>
-    <copyField source="enrichment_*" dest="text"/>
-    <!-- Add fulltext to text field for simple search with fulltext. -->
-    <copyField source="fulltext" dest="text"/>
-
-    <!-- Duplicate content of text in simple field for simple search without fulltext. -->
-    <copyField source="abstract" dest="simple"/>
-    <copyField source="title" dest="simple"/>
-    <copyField source="author" dest="simple"/>
-    <copyField source="subject" dest="simple"/>
-    <copyField source="title_parent" dest="simple"/>
-    <copyField source="title_additional" dest="simple"/>
-    <copyField source="title_sub" dest="simple"/>
-    <copyField source="creating_corporation" dest="simple"/>
-    <copyField source="contributing_corporation" dest="simple"/>
-    <copyField source="publisher_name" dest="simple"/>
-    <copyField source="publisher_place" dest="simple"/>
-    <copyField source="identifier" dest="simple"/>
-    <copyField source="persons" dest="simple"/>
-
     <!-- TODO why -->
     <copyField source="author" dest="author_facet"/>
 

conf/solrconfig.xml

@@ -61,7 +61,7 @@
     <requestHandler name="standard" class="solr.SearchHandler" default="true">
         <lst name="defaults">
             <str name="echoParams">explicit</str>
-            <str name="df">text</str>
+            <str name="df">title</str>
             <str name="q.op">AND</str>
         </lst>
     </requestHandler>

src/Query.php

@@ -33,6 +33,8 @@
 namespace Opus\Search;
 
 use InvalidArgumentException;
+use Opus\Common\Config;
+use Opus\Search\Config as SearchConfig;
 use Opus\Search\Facet\Set;
 use Opus\Search\Filter\AbstractFilterBase;
 use RuntimeException;
@@ -42,6 +44,7 @@
 use function array_merge;
 use function array_shift;
 use function array_unique;
+use function boolval;
 use function count;
 use function ctype_digit;
 use function intval;
@@ -74,7 +77,7 @@
  * @method int getRows( int $default = null )
  * @method string[] getFields( array $default = null )
  * @method array getSort( array $default = null )
- * @method bool getUnion( bool $default = null )
+ * @method bool getUnion( bool $default = false )
  * @method AbstractFilterBase getFilter(AbstractFilterBase $default = null ) retrieves condition to be met by resulting documents
  * @method Set getFacet( Set $default = null )
  * @method $this setStart( int $offset )
@@ -86,6 +89,8 @@
  * @method $this setFacet( Set $facet )
  * @method $this addFields( string $fields )
  * @method $this addSort( $sorting )
+ * @method $this setWeightedFields( int[] $weightedFields ) assigns boost factors to fields (e.g. [ 'title' => 10, 'abstract' => 0.5 ])
+ * @method $this setWeightMultiplier( int $multiplier ) multiplier to further increase boost factors when matching phrases
  */
 class Query
 {
@@ -95,14 +100,16 @@ class Query
     public function reset()
     {
         $this->data = [
-            'start'      => null,
-            'rows'       => null,
-            'fields'     => null,
-            'sort'       => null,
-            'union'      => null,
-            'filter'     => null,
-            'facet'      => null,
-            'subfilters' => null,
+            'start'            => null,
+            'rows'             => null,
+            'fields'           => null,
+            'sort'             => null,
+            'union'            => false,
+            'filter'           => null,
+            'facet'            => null,
+            'subfilters'       => null,
+            'weightedfields'   => null,
+            'weightmultiplier' => null,
         ];
     }
 
@@ -184,6 +191,83 @@ protected function normalizeDirection($ascending)
         return $ascending;
     }
 
+    /**
+     * Returns true if a weighted search shall be used, otherwise returns false.
+     *
+     * @return bool
+     */
+    public function getWeightedSearch()
+    {
+        if (! isset($this->data['weightedsearch'])) {
+            $config = Config::get();
+
+            if (isset($config->search->weightedSearch)) {
+                $this->data['weightedsearch'] = boolval($config->search->weightedSearch);
+            } else {
+                $this->data['weightedsearch'] = false;
+            }
+        }
+
+        return $this->data['weightedsearch'];
+    }
+
+    /**
+     * Set to true if a weighted search shall be used, otherwise set to false.
+     *
+     * @param  bool $value
+     * @return $this fluent interface
+     */
+    public function setWeightedSearch($value)
+    {
+        $this->data['weightedsearch'] = ! ! $value;
+
+        return $this;
+    }
+
+    /**
+     * Returns boost factors keyed by field (e.g. [ 'title' => 10, 'abstract' => 0.5 ]).
+     *
+     * @return int[]
+     */
+    public function getWeightedFields()
+    {
+        if ($this->data['weightedfields'] === null) {
+            $config = Config::get();
+
+            if (isset($config->search->simple)) {
+                $this->data['weightedfields'] = $config->search->simple->toArray();
+            } else {
+                $this->data['weightedfields'] = [];
+            }
+        }
+
+        return $this->data['weightedfields'];
+    }
+
+    /**
+     * Returns a positive integer used as a multiplier to further increase field-specific boost factors when
+     * matching phrases (i.e., in cases where all query terms appear in close proximity).
+     *
+     * For example, with a weight multiplier of 5, the weightedfields array [ 'title' => 10, 'abstract' => 0.5 ]
+     * would be translated to [ 'title' => 50, 'abstract' => 2.5 ] when matching phrases.
+     *
+     * @return int
+     */
+    public function getWeightMultiplier()
+    {
+        if ($this->data['weightmultiplier'] === null) {
+            $config = Config::get();
+
+            if (isset($config->search->weightMultiplier)) {
+                $this->data['weightmultiplier'] = $config->search->weightMultiplier;
+            } else {
+                $this->data['weightmultiplier'] = 1;
+            }
+        }
+
+        return $this->data['weightmultiplier'];
+    }
+
     /**
      * Retrieves value of selected query parameter.
      *
@@ -214,6 +298,7 @@ public function set($name, $value, $adding = false)
         switch ($name) {
             case 'start':
             case 'rows':
+            case 'weightmultiplier':
                 if ($adding) {
                     throw new InvalidArgumentException('invalid parameter access on ' . $name);
                 }
@@ -300,6 +385,18 @@ public function set($name, $value, $adding = false)
 
             case 'subfilters':
                 throw new RuntimeException('invalid access on sub filters');
+
+            case 'weightedfields':
+                if ($adding) {
+                    throw new InvalidArgumentException('invalid parameter access on ' . $name);
+                }
+
+                if (! is_array($value)) {
+                    throw new InvalidArgumentException('invalid query fields option');
+                }
+
+                $this->data[$name] = $value;
+                break;
         }
 
         return $this;
@@ -469,7 +566,7 @@ public function getSubFilters()
      */
     public static function getParameterDefault($name, $fallbackIfMissing, $oldName = null)
     {
-        $config   = Config::getDomainConfiguration();
+        $config   = SearchConfig::getDomainConfiguration();
         $defaults = $config->parameterDefaults;
 
         if ($defaults instanceof Zend_Config) {

src/Result/Base.php

@@ -38,8 +38,10 @@
 use Opus\Search\Log;
 use RuntimeException;
 
+use function array_filter;
 use function array_key_exists;
 use function array_map;
+use function array_values;
 use function count;
 use function ctype_digit;
 use function intval;
@@ -192,9 +194,11 @@ public function getFacet($fieldName)
      * Retrieves set of matching and locally existing documents returned in
      * response to some search query.
      *
+     * @param bool $ignoreZeroScoreMatches ignore any matches with score 0.0
+     *        (true) or not (false); defaults to true
      * @return ResultMatch[]
      */
-    public function getReturnedMatches()
+    public function getReturnedMatches($ignoreZeroScoreMatches = true)
     {
         if ($this->data['matches'] === null) {
             return [];
@@ -208,7 +212,10 @@ public function getReturnedMatches()
         foreach ($this->data['matches'] as $match) {
             try {
                 $match->getDocument();
-                $matches[] = $match;
+                $ignoreMatch = $ignoreZeroScoreMatches === true && $match->getScore() === 0.0;
+                if ($ignoreMatch !== true) {
+                    $matches[] = $match;
+                }
             } catch (DocumentException $e) {
                 Log::get()->warn('skipping matching but locally missing document #' . $match->getId());
             }
@@ -223,18 +230,22 @@ public function getReturnedMatches()
      *
      * @note If query was requesting to retrieve non-qualified matches this set
      *       might include IDs of documents that doesn't exist locally anymore.
+     * @param bool $ignoreZeroScoreMatches ignore any matches with score 0.0
+     *        (true) or not (false); defaults to true
      * @return int[]
      */
-    public function getReturnedMatchingIds()
+    public function getReturnedMatchingIds($ignoreZeroScoreMatches = true)
     {
         if ($this->data['matches'] === null) {
             return [];
         }
 
-        return array_map(function ($match) {
-            /** @var ResultMatch $match */
-            return $match->getId();
+        $matchingIds = array_map(function (ResultMatch $match) use ($ignoreZeroScoreMatches) {
+            $ignoreMatch = $ignoreZeroScoreMatches === true && $match->getScore() === 0.0;
+            return $ignoreMatch !== true ? $match->getId() : null;
         }, $this->data['matches']);
+
+        return array_values(array_filter($matchingIds));
     }
 
     /**
@@ -247,7 +258,7 @@ public function getReturnedMatchingIds()
      *       has changed in that it's returning set of Opus_Document instances
      *       rather than set of Opus_Search_Util_Result instances.
      * @note The wording is less specific in that all information in response to
-     *       search query may considered results of search. Thus this new API
+     *       search query may be considered results of search. Thus this new API
      *       prefers "matches" over "results".
      */
     public function getResults()

src/Solr/Solarium/Adapter.php

@@ -77,6 +77,7 @@
 use function file_exists;
 use function filesize;
 use function filter_var;
+use function implode;
 use function in_array;
 use function intval;
 use function is_array;
@@ -615,6 +616,25 @@ protected function applyParametersOnQuery(
                 $query->setSorts($sortings);
             }
 
+            $isWeightedSearch = $parameters->getWeightedSearch();
+            if ($isWeightedSearch === true) {
+                // get the edismax component
+                $edismax = $query->getEDisMax();
+
+                // NOTE: query is now an edismax query
+                $weightedFields = $parameters->getWeightedFields();
+                if (! empty($weightedFields)) {
+                    $queryFields = $this->getQueryFieldsString($weightedFields);
+                    $edismax->setQueryFields($queryFields);
+
+                    $weightMultiplier = $parameters->getWeightMultiplier();
+                    if ($weightMultiplier !== null) {
+                        $phraseFields = $this->getPhraseFieldsString($weightedFields, $weightMultiplier);
+                        $edismax->setPhraseFields($phraseFields);
+                    }
+                }
+            }
+
             $facet = $parameters->getFacet();
             if ($facet !== null) {
                 $facetSet = $query->getFacetSet();
@@ -880,4 +900,38 @@ public function setTimeout($timeout)
             $this->client->setOptions($options, true);
         }
     }
+
+    /**
+     * Converts an array containing boost factors keyed by field into a query fields string that can be used
+     * as input for the Solr `qf` request parameter.
+     *
+     * @param int[] $weightedFields assigns boost factors to fields, e.g.: [ 'title' => 10, 'abstract' => 0.5 ]
+     * @return string query fields string, e.g.: "title^10 abstract^0.5"
+     */
+    protected function getQueryFieldsString($weightedFields)
+    {
+        $queryFields = [];
+        foreach ($weightedFields as $field => $boostFactor) {
+            $queryFields[] = "$field^$boostFactor";
+        }
+
+        return implode(' ', $queryFields);
+    }
+
+    /**
+     * Generates a phrase fields string that can be used as input for the Solr `pf` request parameter.
+     *
+     * @param int[] $weightedFields assigns boost factors to fields, e.g.: [ 'title' => 10, 'abstract' => 0.5 ]
+     * @param int   $weightMultiplier factor by which each boost factor will be multiplied when matching phrases, e.g.: 5
+     * @return string phrase fields string, e.g.: "title^50 abstract^2.5"
+     */
+    protected function getPhraseFieldsString($weightedFields, $weightMultiplier)
+    {
+        $phraseFields = [];
+        foreach ($weightedFields as $field => $boostFactor) {
+            $phraseFields[] = "$field^" . $boostFactor * $weightMultiplier;
+        }
+
+        return implode(' ', $phraseFields);
+    }
 }