Restrict search on visible private fields

Search on private fields is only supported on bare text search right now (TextNode).
This also fix and issue on databox::get_collection_unique_ids() returning the
value from the first call over and over.

lib/Alchemy/Phrasea/SearchEngine/Elastic/AST/TextNode.php

@@ -3,6 +3,7 @@
 namespace Alchemy\Phrasea\SearchEngine\Elastic\AST;
 
 use Alchemy\Phrasea\SearchEngine\Elastic\Search\QueryContext;
+use Alchemy\Phrasea\SearchEngine\Elastic\Search\QueryHelper;
 use Alchemy\Phrasea\SearchEngine\Elastic\Thesaurus\Term;
 
 class TextNode extends AbstractTermNode implements ContextAbleInterface
@@ -44,16 +45,58 @@ class TextNode extends AbstractTermNode implements ContextAbleInterface
             )
         );
 
-        if ($conceptQueries = $this->buildConceptQueries($context)) {
-            $textQuery = $query;
-            $query = array();
-            $query['bool']['should'] = $conceptQueries;
-            $query['bool']['should'][] = $textQuery;
+        foreach ($this->buildPrivateFieldQueries($context) as $private_field_query) {
+            $query = QueryHelper::applyBooleanClause($query, 'should', $private_field_query);
+        }
+
+        foreach ($this->buildConceptQueries($context) as $concept_query) {
+            $query = QueryHelper::applyBooleanClause($query, 'should', $concept_query);
         }
 
         return $query;
     }
 
+    private function buildPrivateFieldQueries(QueryContext $context)
+    {
+        // We make a boolean clause for each collection set to shrink query size
+        // (instead of a clause for each field, with his collection set)
+        $fields_map = [];
+        $collections_map = [];
+        foreach ($context->getAllowedPrivateFields() as $field) {
+            $collections = $context->getAllowedCollectionsOnPrivateField($field);
+            $hash = self::hashCollections($collections);
+            $collections_map[$hash] = $collections;
+            if (!isset($fields_map[$hash])) {
+                $fields_map[$hash] = [];
+            }
+            // Merge fields with others having the same collections
+            $fields = $context->localizeField($field->getIndexFieldName());
+            foreach ($fields as $fields_map[$hash][]);
+        }
+
+        $queries = [];
+        foreach ($fields_map as $hash => $fields) {
+            // Right to query on a private field is dependant of document collection
+            // Here we make sure we can only match on allowed collections
+            $match = [];
+            $match['multi_match']['fields'] = $fields;
+            $match['multi_match']['query'] = $this->text;
+            $match['multi_match']['operator'] = 'and';
+            $query = [];
+            $query['bool']['must'][0]['terms']['base_id'] = $collections_map[$hash];
+            $query['bool']['must'][1] = $match;
+            $queries[] = $query;
+        }
+
+        return $queries;
+    }
+
+    private static function hashCollections(array $collections)
+    {
+        sort($collections, SORT_REGULAR);
+        return implode('|', $collections);
+    }
+
     public function __toString()
     {
         return sprintf('<text:%s>', Term::dump($this));

lib/Alchemy/Phrasea/SearchEngine/Elastic/ElasticSearchEngine.php

@@ -300,6 +300,9 @@ class ElasticSearchEngine implements SearchEngineInterface
         );
     }
 
+    /**
+     * @todo Move in search engine service provider
+     */
     private function createQueryContext(SearchEngineOptions $options)
     {
         // TODO handle $user when null
@@ -321,6 +324,7 @@ class ElasticSearchEngine implements SearchEngineInterface
      *     "OtherFieldName" => [4],
      * ]
      *
+     * @todo Move in query context
      * @param SearchEngineOptions $options
      * @return array
      */
@@ -334,8 +338,11 @@ class ElasticSearchEngine implements SearchEngineInterface
 
         $map = $this->structure->getCollectionsUsedByPrivateFields();
         // Remove collections base_id which access is restricted.
-        foreach ($map as $_ => &$collections) {
+        foreach ($map as $key => &$collections) {
             $collections = array_intersect($collections, $allowed_collections);
+            if (!$collections) {
+                unset($map[$key]);
+            }
         }
 
         return $map;

lib/Alchemy/Phrasea/SearchEngine/Elastic/Search/QueryContext.php

@@ -3,6 +3,7 @@
 namespace Alchemy\Phrasea\SearchEngine\Elastic\Search;
 
 use Alchemy\Phrasea\SearchEngine\Elastic\Exception\QueryException;
+use Alchemy\Phrasea\SearchEngine\Elastic\Structure\Field;
 use Alchemy\Phrasea\SearchEngine\Elastic\Structure\Structure;
 
 /**
@@ -46,10 +47,7 @@ class QueryContext
     public function getRawFields()
     {
         if ($this->fields === null) {
-            return array(
-                'caption_all.raw',
-                'private_caption_all.raw'
-            );
+            return array('caption_all.raw');
         }
 
         $fields = array();
@@ -66,10 +64,7 @@ class QueryContext
     public function getLocalizedFields()
     {
         if ($this->fields === null) {
-            return array_merge(
-                $this->localizeField('caption_all'),
-                $this->localizeField('private_caption_all')
-            );
+            return $this->localizeField('caption_all');
         }
 
         $fields = array();
@@ -81,7 +76,27 @@ class QueryContext
         return $fields;
     }
 
-    private function localizeField($field)
+    public function getAllowedPrivateFields()
+    {
+        $allowed_field_names = array_keys($this->privateCollectionMap);
+
+        return array_map(array($this->structure, 'get'), $allowed_field_names);
+    }
+
+    public function getAllowedCollectionsOnPrivateField(Field $field)
+    {
+        $name = $field->getName();
+        if (!isset($this->privateCollectionMap[$name])) {
+            throw new \OutOfRangeException('Given field is not an allowed private field.');
+        }
+
+        return $this->privateCollectionMap[$name];
+    }
+
+    /**
+     * @todo Maybe we should put this logic in Field class?
+     */
+    public function localizeField($field)
     {
         $fields = array();
         foreach ($this->locales as $locale) {

lib/Alchemy/Phrasea/SearchEngine/Elastic/Search/QueryHelper.php

@@ -0,0 +1,47 @@
+<?php
+
+namespace Alchemy\Phrasea\SearchEngine\Elastic\Search;
+
+class QueryHelper
+{
+    private function __construct() {}
+
+    /**
+     * Apply conjunction or disjunction between a query and a sub query clause
+     *
+     * @param  array  $query     Query
+     * @param  string $type      "must" for conjunction, "should" for disjunction
+     * @param  array  $sub_query Clause query
+     * @return array             Resulting query
+     */
+    public static function applyBooleanClause(array $query, $type, array $clause)
+    {
+        if (!in_array($type, ['must', 'should'])) {
+            throw new \InvalidArgumentException(sprintf('Type must be either "must" or "should", "%s" given', $type));
+        }
+
+        if (!isset($query['bool'])) {
+            // Wrap in a boolean query
+            $bool = [];
+            $bool['bool'][$type][] = $query;
+            $bool['bool'][$type][] = $clause;
+
+            return $bool;
+        } elseif (isset($query['bool'][$type])) {
+            // Reuse the existing boolean clause group
+            if (!is_array($query['bool'][$type])) {
+                // Wrap the previous clause in an array
+                $previous_clause = $query['bool'][$type];
+                $query['bool'][$type] = [];
+                $query['bool'][$type][] = $previous_clause;
+            }
+            $query['bool'][$type][] = $clause;
+
+            return $query;
+        } else {
+            $query['bool'][$type] = $clause;
+
+            return $query;
+        }
+    }
+}

lib/classes/databox.php

@@ -232,10 +232,10 @@ class databox extends base
 
     public function get_collection_unique_ids()
     {
-        static $base_ids;
+        static $base_ids_cache = [];
 
-        if (isset($base_ids)) {
-            return $base_ids;
+        if (isset($base_ids_cache[$this->id])) {
+            return $base_ids_cache[$this->id];
         }
 
         $conn = $this->get_appbox()->get_connection();
@@ -250,7 +250,7 @@ class databox extends base
             $base_ids[] = (int) $row['base_id'];
         }
 
-        return $base_ids;
+        return $base_ids_cache[$this->id] = $base_ids;
     }
 
     protected function get_available_collections()