From b1a03ae5666628921a9a27d29000f6a4865ebb5b Mon Sep 17 00:00:00 2001
From: Mathieu Darse <mdarse@jolicode.com>
Date: Thu, 19 Mar 2015 13:38:20 +0100
Subject: [PATCH] Smart facet value escaping

---
 .../Core/Provider/SearchEngineServiceProvider.php   | 11 ++++++++++-
 .../SearchEngine/Elastic/ElasticSearchEngine.php    | 10 ++++++----
 .../SearchEngine/Elastic/Search/FacetsResponse.php  | 13 +++++++------
 3 files changed, 23 insertions(+), 11 deletions(-)

diff --git a/lib/Alchemy/Phrasea/Core/Provider/SearchEngineServiceProvider.php b/lib/Alchemy/Phrasea/Core/Provider/SearchEngineServiceProvider.php
index 58ff3eff48..2cf76ae22a 100644
--- a/lib/Alchemy/Phrasea/Core/Provider/SearchEngineServiceProvider.php
+++ b/lib/Alchemy/Phrasea/Core/Provider/SearchEngineServiceProvider.php
@@ -20,6 +20,8 @@ use Alchemy\Phrasea\SearchEngine\Elastic\IndexerSubscriber;
 use Alchemy\Phrasea\SearchEngine\Elastic\Indexer\RecordIndexer;
 use Alchemy\Phrasea\SearchEngine\Elastic\Indexer\TermIndexer;
 use Alchemy\Phrasea\SearchEngine\Elastic\RecordHelper;
+use Alchemy\Phrasea\SearchEngine\Elastic\Search\Escaper;
+use Alchemy\Phrasea\SearchEngine\Elastic\Search\FacetsResponse;
 use Alchemy\Phrasea\SearchEngine\Elastic\Search\QueryParser;
 use Alchemy\Phrasea\SearchEngine\Elastic\Thesaurus;
 use Alchemy\Phrasea\SearchEngine\Phrasea\PhraseaEngine;
@@ -67,10 +69,17 @@ class SearchEngineServiceProvider implements ServiceProviderInterface
             return new ElasticSearchEngine(
                 $app,
                 $app['elasticsearch.client'],
-                $app['elasticsearch.options']['index']
+                $app['elasticsearch.options']['index'],
+                $app['locales.available'],
+                $app['elasticsearch.record_helper'],
+                $app['elasticsearch.facets_response.factory']
             );
         });
 
+        $app['elasticsearch.facets_response.factory'] = $app->protect(function (array $response) use ($app) {
+            return new FacetsResponse(new Excaper(), $response);
+        });
+
 
         /* Indexer related services */
 
diff --git a/lib/Alchemy/Phrasea/SearchEngine/Elastic/ElasticSearchEngine.php b/lib/Alchemy/Phrasea/SearchEngine/Elastic/ElasticSearchEngine.php
index f8bbc59620..d39cae1975 100644
--- a/lib/Alchemy/Phrasea/SearchEngine/Elastic/ElasticSearchEngine.php
+++ b/lib/Alchemy/Phrasea/SearchEngine/Elastic/ElasticSearchEngine.php
@@ -13,6 +13,7 @@ namespace Alchemy\Phrasea\SearchEngine\Elastic;
 
 use Alchemy\Phrasea\SearchEngine\Elastic\Indexer\RecordIndexer;
 use Alchemy\Phrasea\SearchEngine\Elastic\Indexer\TermIndexer;
+use Alchemy\Phrasea\SearchEngine\Elastic\RecordHelper;
 use Alchemy\Phrasea\SearchEngine\Elastic\Search\FacetsResponse;
 use Alchemy\Phrasea\SearchEngine\Elastic\Search\QueryContext;
 use Alchemy\Phrasea\SearchEngine\SearchEngineInterface;
@@ -39,12 +40,13 @@ class ElasticSearchEngine implements SearchEngineInterface
     private $locales;
     private $recordHelper;
 
-    public function __construct(Application $app, Client $client, $indexName)
+    public function __construct(Application $app, Client $client, $indexName, array $locales, RecordHelper $recordHelper, Closure $facetsResponseFactory)
     {
         $this->app = $app;
         $this->client = $client;
-        $this->locales = array_keys($app['locales.available']);
-        $this->recordHelper = $this->app['elasticsearch.record_helper'];
+        $this->locales = array_keys($locales);
+        $this->recordHelper = $recordHelper;
+        $this->facetsResponseFactory = $facetsResponseFactory;
 
         if ('' === trim($indexName)) {
             throw new \InvalidArgumentException('The provided index name is invalid.');
@@ -292,7 +294,7 @@ class ElasticSearchEngine implements SearchEngineInterface
             $results[] = ElasticsearchRecordHydrator::hydrate($hit['_source'], $n++);
         }
 
-        $facets = new FacetsResponse($res);
+        $facets = $this->facetsResponseFactory->__invoke($res);
 
         $query['ast'] = $this->app['query_parser']->parse($string)->dump();
         $query['query_main'] = $recordQuery;
diff --git a/lib/Alchemy/Phrasea/SearchEngine/Elastic/Search/FacetsResponse.php b/lib/Alchemy/Phrasea/SearchEngine/Elastic/Search/FacetsResponse.php
index 52d0ccdcd6..c651aebff1 100644
--- a/lib/Alchemy/Phrasea/SearchEngine/Elastic/Search/FacetsResponse.php
+++ b/lib/Alchemy/Phrasea/SearchEngine/Elastic/Search/FacetsResponse.php
@@ -7,10 +7,13 @@ use JsonSerializable;
 
 class FacetsResponse implements JsonSerializable
 {
+    private $escaper;
     private $facets = array();
 
-    public function __construct(array $response)
+    public function __construct(Escaper $escaper, array $response)
     {
+        $this->escaper = $escaper;
+
         if (!isset($response['aggregations'])) {
             return;
         }
@@ -47,12 +50,10 @@ class FacetsResponse implements JsonSerializable
 
     private function buildQuery($name, $value)
     {
-        // Strip double quotes from values to prevent broken queries
-        $value = str_replace('/"/u', ' ', $value);
-        // TODO escape value when escaping is supported in query parser
+        $value = $this->escaper->escapeWord($value);
         return ($name === 'Collection') ?
-            sprintf('collection:"%s"', $value) :
-            sprintf('"%s" IN %s', $value, $name);
+            sprintf('collection:%s', $value) :
+            sprintf('%s IN %s', $value, $name);
     }
 
     private function throwAggregationResponseError()