From c3b649b1acf0976a19734f818cb4e80c95fbbbaa Mon Sep 17 00:00:00 2001
From: Dorian Merz <merz@ub.uni-leipzig.de>
Date: Tue, 22 Jan 2019 10:50:23 +0100
Subject: [PATCH] refs #14353 - search for callnumber w/ colon * introduces
MungerInjectionFactory * registers several listeners on Search Pre event **
escapes colons in queries for configured handlers (e.g. Signatur) **
registers all configured shards for every query
---
local/config/vufind/searches.ini | 10 ++
module/finc/config/module.config.php | 10 +-
.../finc/Service/MungerInjectionFactory.php | 109 ++++++++++++++++++
3 files changed, 128 insertions(+), 1 deletion(-)
create mode 100644 module/finc/src/finc/Service/MungerInjectionFactory.php
diff --git a/local/config/vufind/searches.ini b/local/config/vufind/searches.ini
index ab20a8fc4da..f6d6e5cdecf 100644
--- a/local/config/vufind/searches.ini
+++ b/local/config/vufind/searches.ini
@@ -98,6 +98,11 @@ retain_filters_by_default = true
;default_filters[] = "institution:MyInstitution"
;default_filters[] = "(format:Book AND institution:MyInstitution)"
+; the escaped_colon_searches is used by a listener on the search-pre event
+; registered by the MungerInjectionFactory. This listener masks colons in the query string with a backslash
+; whenever the search handler is one of the following
+escaped_colon_searches[] = "Signatur"
+
[Cache]
; This controls whether the parsed searchspecs.yaml file will be stored to
; improve search performance; legal options are APC (use APC cache), File (store
@@ -554,6 +559,11 @@ topic = "Subjects"
; specified preferences using checkboxes (default if commented out = all shards):
;defaultChecked[] = "Library Catalog"
;defaultChecked[] = "Website"
+; The following line defines shards that shall NOT be queried by solr requests
+; all other shards present in the [IndexShards] section will always be added to any solr query
+; via a listener from the MungerInjectionFactory
+;on_user_search_only = "finc-live"
+
; Fields must be stripped if you have a field in your main index which is missing
; from any index includable by shards. This section can be ignored if you are
diff --git a/module/finc/config/module.config.php b/module/finc/config/module.config.php
index b6b91aa270e..2f5bcbbc70e 100644
--- a/module/finc/config/module.config.php
+++ b/module/finc/config/module.config.php
@@ -13,7 +13,15 @@ $config = [
'VuFind\Export' => 'finc\Service\Factory::getExport',
'VuFind\SessionManager' => 'finc\Session\ManagerFactory',
'VuFind\CookieManager' => 'finc\Service\Factory::getCookieManager'
- ]
+ ],
+ 'invokables' => [
+ 'mungerinjectionfactory' => 'finc\Service\MungerInjectionFactory'
+ ],
+ 'delegators' => [
+ 'VuFind\Search' => [
+ 'mungerinjectionfactory'
+ ],
+ ],
],
'controllers' => [
'factories' => [
diff --git a/module/finc/src/finc/Service/MungerInjectionFactory.php b/module/finc/src/finc/Service/MungerInjectionFactory.php
new file mode 100644
index 00000000000..b0a4460f826
--- /dev/null
+++ b/module/finc/src/finc/Service/MungerInjectionFactory.php
@@ -0,0 +1,109 @@
+<?php
+
+namespace finc\Service;
+
+use VuFindSearch\Query\Query,
+ VuFindSearch\Query\QueryGroup;
+use Zend\Config\Config;
+use Zend\ServiceManager\DelegatorFactoryInterface,
+ Zend\ServiceManager\ServiceLocatorInterface;
+use Zend\EventManager\EventInterface;
+use VuFindSearch\Service as SearchService;
+
+/**
+ * Class MungerInjectionFactory
+ * A Delegator Factory that registers several listeners at events triggered by the VuFind\Search service.
+ * @package finc\Service
+ */
+class MungerInjectionFactory implements DelegatorFactoryInterface {
+
+ /**
+ * @var SearchService
+ */
+ protected $instance;
+
+ /**
+ * @var array names of search handlers for which colons should be escaped
+ */
+ protected $searches_to_escape;
+
+ /**
+ * @var array shard configuration to register in all queries
+ */
+ protected $shards_to_register;
+
+ /**
+ * Creates a delegator of VuFind/Search to register several listeners.
+ * @param ServiceLocatorInterface $serviceLocator
+ * @param string $name
+ * @param string $requestedName
+ * @param callable $callback
+ * @return mixed
+ */
+ public function createDelegatorWithName(ServiceLocatorInterface $serviceLocator, $name, $requestedName, $callback)
+ {
+ $instance = call_user_func($callback);
+ $searchConfig = $serviceLocator->get('VuFind\Config')->get('searches');
+ $e = $instance->getEventManager()->getSharedManager();
+ $handlers = $searchConfig->General->escaped_colon_searches;
+ if (!empty($handlers)) {
+ $this->searches_to_escape = $handlers->toArray();
+ $e->attach('VuFind\Search', 'pre',
+ function (EventInterface $event) {
+ $params = $event->getParams();
+ if (isset($params['query'])) {
+ $params['query'] = $this->escapeColons($params['query']);
+ }
+ }
+ );
+ }
+ $shards = $searchConfig->IndexShards->toArray();
+ if ($excludedShards = $searchConfig->ShardPreferences->on_user_search_only) {
+ $shards = array_diff_key($shards, array_flip(explode(',', $excludedShards)));
+ }
+ if (!empty($shards)) {
+ $this->shards_to_register = $shards;
+ $e->attach('VuFind\Search', 'pre', [$this, 'registerShards']);
+ }
+ return $instance;
+ }
+
+ /**
+ * Escapes colons in Queries or recursively in QueryGroups.
+ * This prevents queries from being interpreted as advanced queries in Lucene syntax.
+ * cf. \VuFindSearch\Backend\Solr\LuceneSyntaxHelper::containsAdvancedLuceneSyntax
+ * @param Query|QueryGroup $queryOrGroup
+ * @return mixed
+ */
+ private function escapeColons($queryOrGroup) {
+
+ if ($queryOrGroup instanceof QueryGroup) {
+ $handler = $queryOrGroup->getReducedHandler();
+ if (is_null($handler) || in_array($handler,$this->searches_to_escape)) {
+ foreach ($queryOrGroup->getQueries() as $query) {
+ $this->escapeColons($query);
+ }
+ }
+ } elseif (in_array($queryOrGroup->getHandler(),$this->searches_to_escape)) {
+ $queryOrGroup->setString(
+ // mask whitespaces that follow a colon
+ // that avoids the removal of that very colon via
+ // \VuFindSearch\Backend\Solr\LuceneSyntaxHelper::normalizeColons
+ preg_replace('/(?<=\:)\s/', '\ ', $queryOrGroup->getString())
+ );
+ }
+ return $queryOrGroup;
+ }
+
+ /**
+ * Event Listener on Search/Pre that registers all configured shards for every search request
+ * @param EventInterface $event
+ */
+ public function registerShards(EventInterface $event) {
+
+ $params = $event->getParam('params');
+ if (empty($params->get('shards'))) {
+ $params->set('shards',implode(',',$this->shards_to_register));
+ }
+ }
+}
\ No newline at end of file
--
GitLab