From 042f634fd5322bfdd717ea1f55e435b8e6066f67 Mon Sep 17 00:00:00 2001
From: Demian Katz <demian.katz@villanova.edu>
Date: Fri, 16 Nov 2012 13:07:52 -0500
Subject: [PATCH] Resolving VUFIND-629 (improved Wikipedia results using
authority data and VIAF API).
---
config/vufind/searches.ini | 7 +-
module/VuFind/config/module.config.php | 6 +-
.../src/VuFind/Recommend/AuthorInfo.php | 124 +++++++++++++++---
.../src/VuFind/RecordDriver/SolrAuth.php | 20 +++
4 files changed, 140 insertions(+), 17 deletions(-)
diff --git a/config/vufind/searches.ini b/config/vufind/searches.ini
index 9ace9c2cfba..d6f5f4d77ef 100644
--- a/config/vufind/searches.ini
+++ b/config/vufind/searches.ini
@@ -207,9 +207,14 @@ CallNumber = callnumber
;
; AuthorFacets
; Display author names based on the search term.
-; AuthorInfo
+; AuthorInfo:[use_viaf]
; Load author information from external providers based on the current search
; and the "authors" setting in the [Content] section of config.ini.
+; The [use_viaf] setting may be set to true to use the VIAF web service in
+; combination with your Solr authority index to pick more appropriate authors;
+; this defaults to false if unset. If you do not have local authority records
+; to draw on, the OCLC FAST data works well with the [use_viaf] option; see:
+; http://vufind.org/wiki/open_data_sources#oclc_fast
; AuthorityRecommend:[field1]:[query1]:[field2]:[query2]:...:[field-n]:[query-n]
; Displays see and see also results to user based on search of Authority Index.
; Any number of filter queries may be specified as parameters to limit which
diff --git a/module/VuFind/config/module.config.php b/module/VuFind/config/module.config.php
index d6614d5e856..2475b5eac28 100644
--- a/module/VuFind/config/module.config.php
+++ b/module/VuFind/config/module.config.php
@@ -231,6 +231,11 @@ $config = array(
'recommend_plugin_manager' => array(
'abstract_factories' => array('VuFind\Recommend\PluginFactory'),
'factories' => array(
+ 'authorinfo' => function ($sm) {
+ return new \VuFind\Recommend\AuthorInfo(
+ $sm->getServiceLocator()->get('SearchManager')
+ );
+ },
'worldcatidentities' => function ($sm) {
return new \VuFind\Recommend\WorldCatIdentities(
$sm->getServiceLocator()->get('VuFind\WorldCatUtils')
@@ -244,7 +249,6 @@ $config = array(
),
'invokables' => array(
'authorfacets' => 'VuFind\Recommend\AuthorFacets',
- 'authorinfo' => 'VuFind\Recommend\AuthorInfo',
'authorityrecommend' => 'VuFind\Recommend\AuthorityRecommend',
'catalogresults' => 'VuFind\Recommend\CatalogResults',
'europeanaresults' => 'VuFind\Recommend\EuropeanaResults',
diff --git a/module/VuFind/src/VuFind/Recommend/AuthorInfo.php b/module/VuFind/src/VuFind/Recommend/AuthorInfo.php
index 43516a560b9..b4271d3926e 100644
--- a/module/VuFind/src/VuFind/Recommend/AuthorInfo.php
+++ b/module/VuFind/src/VuFind/Recommend/AuthorInfo.php
@@ -65,6 +65,30 @@ class AuthorInfo implements RecommendInterface, TranslatorAwareInterface
*/
protected $lang;
+ /**
+ * Search manager
+ *
+ * @var \VuFind\Search\Manager
+ */
+ protected $searchManager;
+
+ /**
+ * Should we use VIAF for authorized names?
+ *
+ * @var bool
+ */
+ protected $useViaf = false;
+
+ /**
+ * Constructor
+ *
+ * @param \VuFind\Search\Manager $searchManager Search manager
+ */
+ public function __construct(\VuFind\Search\Manager $searchManager)
+ {
+ $this->searchManager = $searchManager;
+ }
+
/**
* setConfig
*
@@ -78,6 +102,13 @@ class AuthorInfo implements RecommendInterface, TranslatorAwareInterface
{
$translator = $this->getTranslator();
$this->lang = is_object($translator) ? $translator->getLocale() : 'en';
+
+ $parts = explode(':', $settings);
+ if (isset($parts[0]) && !empty($parts[0])
+ && strtolower(trim($parts[0])) !== 'false'
+ ) {
+ $this->useViaf = true;
+ }
}
/**
@@ -403,6 +434,81 @@ class AuthorInfo implements RecommendInterface, TranslatorAwareInterface
return $info;
}
+ /**
+ * Normalize an author name using internal logic.
+ *
+ * @param string $author Author name
+ *
+ * @return string
+ */
+ protected function normalizeName($author)
+ {
+ // remove dates
+ $author = preg_replace('/[0-9]+-[0-9]*/', '', $author);
+ // if name is rearranged by commas
+ $author = trim($author, ', .');
+ $nameParts = explode(', ', $author);
+ $last = $nameParts[0];
+ // - move all names up an index, move last name to last
+ // - Last, First M. -> First M. Last
+ for ($i=1;$i<count($nameParts);$i++) {
+ $nameParts[$i-1] = $nameParts[$i];
+ }
+ $nameParts[count($nameParts)-1] = $last;
+ $author = implode($nameParts, ' ');
+ return $author;
+ }
+
+ /**
+ * Translate an LCCN to a Wikipedia name through the VIAF web service. Returns
+ * false if no value can be found.
+ *
+ * @param string $lccn LCCN
+ *
+ * @return string|bool
+ */
+ protected function getWikipediaNameFromViaf($lccn)
+ {
+ $param = urlencode("LC|$lccn");
+ $url = "http://viaf.org/viaf/sourceID/{$param}/justlinks.json";
+ $client = new \VuFind\Http\Client();
+ $result = $client->setUri($url)->setMethod('GET')->send();
+ if (!$result->isSuccess()) {
+ return false;
+ }
+ $details = json_decode($result->getBody());
+ return isset($details->WKP[0]) ? $details->WKP[0] : false;
+ }
+
+ /**
+ * Normalize an author name using VIAF.
+ *
+ * @param string $author Author name
+ *
+ * @return string
+ */
+ protected function normalizeNameWithViaf($author)
+ {
+ // Do authority search:
+ $auth = $this->searchManager->setSearchClassId('SolrAuth')->getResults();
+ $auth->getParams()->setBasicSearch('"' . $author . '"', 'MainHeading');
+ $results = $auth->getResults();
+
+ // Find first useful LCCN:
+ foreach ($results as $i => $current) {
+ $lccn = $current->tryMethod('getRawLCCN');
+ if (!empty($lccn)) {
+ $name = $this->getWikipediaNameFromViaf($lccn);
+ if (!empty($name)) {
+ return $name;
+ }
+ }
+ }
+
+ // No LCCN found? Use the default normalization routine:
+ return $this->normalizeName($author);
+ }
+
/**
* Takes the search term and extracts a normal name from it
*
@@ -415,21 +521,9 @@ class AuthorInfo implements RecommendInterface, TranslatorAwareInterface
$author = $search[0]['lookfor'];
// remove quotes
$author = str_replace('"', '', $author);
- // remove dates
- $author = preg_replace('/[0-9]+-[0-9]*/', '', $author);
- // if name is rearranged by commas
- $author = trim($author, ', .');
- $nameParts = explode(', ', $author);
- $last = $nameParts[0];
- // - move all names up an index, move last name to last
- // - Last, First M. -> First M. Last
- for ($i=1;$i<count($nameParts);$i++) {
- $nameParts[$i-1] = $nameParts[$i];
- }
- $nameParts[count($nameParts)-1] = $last;
- $author = implode($nameParts, ' ');
- // remove punctuation
- return $author;
+ return $this->useViaf
+ ? $this->normalizeNameWithViaf($author)
+ : $this->normalizeName($author);
}
return '';
}
diff --git a/module/VuFind/src/VuFind/RecordDriver/SolrAuth.php b/module/VuFind/src/VuFind/RecordDriver/SolrAuth.php
index b24cca0055d..da979b6a631 100644
--- a/module/VuFind/src/VuFind/RecordDriver/SolrAuth.php
+++ b/module/VuFind/src/VuFind/RecordDriver/SolrAuth.php
@@ -89,4 +89,24 @@ class SolrAuth extends SolrMarc
&& is_array($this->fields['use_for'])
? $this->fields['use_for'] : array();
}
+
+ /**
+ * Get a raw LCCN (not normalized). Returns false if none available.
+ *
+ * @return string|bool
+ */
+ public function getRawLCCN()
+ {
+ $lccn = $this->getFirstFieldValue('010');
+ if (!empty($lccn)) {
+ return $lccn;
+ }
+ $lccns = $this->getFieldArray('700', array('0'));
+ foreach ($lccns as $lccn) {
+ if (substr($lccn, 0, '5') == '(DLC)') {
+ return substr($lccn, 5);
+ }
+ }
+ return false;
+ }
}
--
GitLab