--- a/Utils/WikiTagUtils.php Thu Nov 24 13:05:33 2011 +0100
+++ b/Utils/WikiTagUtils.php Sun Nov 27 23:35:54 2011 +0100
@@ -13,106 +13,6 @@
/**
- * Get or create tag. Returns an array(tag:WikiTagTag, revision_id=int, created:Boolean)
- */
-
- /**
- *
- * Enter description here ...
- * @param unknown_type $tag_label
- * @param unknown_type $doctrine
- * @return multitype:boolean Ambigous <NULL, \IRI\Bundle\WikiTagBundle\Entity\Tag> Ambigous <NULL, unknown, mixed, string> (array(\IRI\Bundle\WikiTagBundle\Model\TagInterface, revision_id=int, created:Boolean))
- */
- public static function getOrCreateTag($tag_label, $doctrine)
- {
- $tag_label_normalized = WikiTagUtils::normalizeTag($tag_label);
- // We get the wikipedia references for the tag_label
- // We get or create the tag object
- $tags = $doctrine->getRepository('WikiTagBundle:Tag')->findBy(array('normalizedLabel' => $tag_label_normalized));
- $tag = null;
- foreach ($tags as $t){
- if($tag==null || $t->getUrlStatus()!=Tag::$TAG_URL_STATUS_DICT['null_result']){
- $tag = $t;
- if($t->getUrlStatus()!=Tag::$TAG_URL_STATUS_DICT['null_result']){
- break;
- }
- }
- }
- $wp_request_done = false;
- if($tag==null){
- $tag = new Tag();
- $tag->setLabel($tag_label_normalized);
- $tag->setOriginalLabel($tag_label);
- $tag->setNormalizedLabel($tag_label_normalized);
- $created = true;
- }
- else{
- $created = false;
- $match_exists = false;
- // Even if a tag with the normalised label exists, IF this tag is not wikipedia semantised,
- // we search if a wikipedia semantised version exists in the base
- foreach ($tags as $t){
- if($t->getUrlStatus()==Tag::$TAG_URL_STATUS_DICT['match']){
- $tag = $t;
- $match_exists = true;
- break;
- }
- }
- if($match_exists==false){
- $wp_response = WikiTagUtils::getWikipediaInfo($tag_label_normalized);
- $status = $wp_response['status'];
- if($status==Tag::$TAG_URL_STATUS_DICT['match']){
- $tag = new Tag();
- $tag->setLabel($tag_label_normalized);
- $tag->setOriginalLabel($tag_label);
- $tag->setNormalizedLabel($tag_label_normalized);
- $created = true;
- $wp_request_done = true;
- }
- }
- }
-
- // We request Wikipedia if the tag is created
- if($created==true){
- if($wp_request_done==false){
- $wp_response = WikiTagUtils::getWikipediaInfo($tag_label_normalized);
- }
- $new_label = $wp_response['new_label'];
- $status = $wp_response['status'];
- $url = $wp_response['wikipedia_url'];
- $pageid = $wp_response['pageid'];
- $dbpedia_uri = $wp_response["dbpedia_uri"];
- $wikipedia_revision_id = $wp_response['revision_id'];
-
- # We save the datas
- if($new_label!=null){
- $tag->setLabel($new_label);
- }
- if($status!=null){
- $tag->setUrlStatus($status);
- }
- $tag->setWikipediaUrl($url);
- $tag->setWikipediaPageId($pageid);
- $tag->setDbpediaUri($dbpedia_uri);
-
- // Save datas.
- $em = $doctrine->getEntityManager();
- $em->persist($tag);
- $em->flush();
-
- }
- else if($tag!=null && $tag->getWikipediaPageId()!=null){
- $wp_response = WikiTagUtils::getWikipediaInfo(null, $tag->getWikipediaPageId());
- $wikipedia_revision_id = $wp_response['revision_id'];
- }
- else{
- $wikipedia_revision_id = null;
- }
-
- return array($tag, $wikipedia_revision_id, $created);//, $wpReponse);
- }
-
- /**
* Cleans the tag label
*/
public static function normalizeTag($tag_label)
@@ -122,19 +22,30 @@
}
$tag_label = trim($tag_label);//tag.strip()
$tag_label = str_replace("_", " ", $tag_label);//tag.replace("_", " ")
- $tag_label = str_replace("Œ", "oe", $tag_label);
- $tag_label = str_replace("œ", "oe", $tag_label);
- $tag_label = preg_replace('/\s+/', ' ', $tag_label);//" ".join(tag.split())
+ $tag_label = preg_replace('/\s+/u', ' ', $tag_label);//" ".join(tag.split())
$tag_label = ucfirst($tag_label);//tag[0].upper() + tag[1:]
return $tag_label;
}
/**
+ * Query wikipedia with a normalized label or a pageid
+ * return an array with the form
+ * array(
+ * 'new_label'=>$new_label,
+ * 'alternative_label'=>$alternative_label,
+ * 'status'=>$status,
+ * 'wikipedia_url'=>$url,
+ * 'wikipedia_alternative_url'=>$alternative_url,
+ * 'pageid'=>$pageid,
+ * 'alternative_pageid'=>$alternative_pageid,
+ * 'dbpedia_uri'=>$dbpedia_uri,
+ * 'revision_id'=> ,
+ * 'response'=> the original wikipedia json response
+ * )
*
- * TODO: Enter description here ...
- * @param unknown_type $tag_label_normalized
- * @param unknown_type $page_id
- * @return multitype:NULL unknown |multitype:Ambigous <NULL, unknown> multitype:number mixed Ambigous <NULL, string> Ambigous <unknown, mixed>
+ * @param string $tag_label_normalized
+ * @param bigint $page_id
+ * @return array
*/
public static function getWikipediaInfo($tag_label_normalized, $page_id=null)
{
@@ -181,7 +92,11 @@
$status = Tag::$TAG_URL_STATUS_DICT["match"];
}
// In redirection, we have to get more datas by adding redirects=true to the params
- if($status==Tag::$TAG_URL_STATUS_DICT["redirection"]){
+ $alternative_label = null;
+ $alternative_url = null;
+ $alternative_pageid = null;
+ if($status==Tag::$TAG_URL_STATUS_DICT["redirection"])
+ {
$params['redirects'] = "true";
$ar = WikiTagUtils::requestWikipedia($params);
$res = $ar[0];
@@ -192,6 +107,9 @@
}
// get first result
$page = reset($pages);
+ $alternative_label = array_key_exists('title', $page) ? $page['title'] : null;
+ $alternative_url = array_key_exists('fullurl', $page) ? $page['fullurl'] : null;
+ $alternative_pageid = array_key_exists('pageid', $page) ? $page['pageid'] : null;
}
$revision_id = $page['lastrevid'];
@@ -214,17 +132,27 @@
$dbpedia_uri = WikiTagUtils::getDbpediaUri($english_label);
}
- $wp_response = array('new_label'=>$new_label, 'status'=>$status, 'wikipedia_url'=>$url, 'pageid'=>$pageid, 'dbpedia_uri'=>$dbpedia_uri, 'revision_id'=>$revision_id, 'response'=>$original_response);
+ $wp_response = array(
+ 'new_label'=>$new_label,
+ 'alternative_label'=>$alternative_label,
+ 'status'=>$status,
+ 'wikipedia_url'=>$url,
+ 'wikipedia_alternative_url'=>$alternative_url,
+ 'pageid'=>$pageid,
+ 'alternative_pageid'=>$alternative_pageid,
+ 'dbpedia_uri'=>$dbpedia_uri,
+ 'revision_id'=>$revision_id,
+ 'response'=>$original_response);
//return $url." <br/>RES = ".$res/*." <br/>DUMP = ".var_dump($pages)*/." <br/>COUNT = ".count($pages)." <br/>page = ".var_dump($page);
return $wp_response;
}
/**
+ * build and do the request to Wikipedia.
*
- * TODO : Enter description here ...
- * @param unknown_type $params
- * @return multitype:unknown mixed
+ * @param array $params
+ * @return array
*/
private static function requestWikipedia($params)
{