9 // Constants |
9 // Constants |
10 private static $WIKIPEDIA_API_URL = "http://fr.wikipedia.org/w/api.php"; |
10 private static $WIKIPEDIA_API_URL = "http://fr.wikipedia.org/w/api.php"; |
11 private static $WIKIPEDIA_VERSION_PERMALINK_TEMPLATE = "http://fr.wikipedia.org/w/index.php?oldid=%s"; |
11 private static $WIKIPEDIA_VERSION_PERMALINK_TEMPLATE = "http://fr.wikipedia.org/w/index.php?oldid=%s"; |
12 private static $DBPEDIA_URI_TEMPLATE = "http://dbpedia.org/resource/%s"; |
12 private static $DBPEDIA_URI_TEMPLATE = "http://dbpedia.org/resource/%s"; |
13 |
13 |
14 |
|
15 /** |
|
16 * Get or create tag. Returns an array(tag:WikiTagTag, revision_id=int, created:Boolean) |
|
17 */ |
|
18 |
|
19 /** |
|
20 * |
|
21 * Enter description here ... |
|
22 * @param unknown_type $tag_label |
|
23 * @param unknown_type $doctrine |
|
24 * @return multitype:boolean Ambigous <NULL, \IRI\Bundle\WikiTagBundle\Entity\Tag> Ambigous <NULL, unknown, mixed, string> (array(\IRI\Bundle\WikiTagBundle\Model\TagInterface, revision_id=int, created:Boolean)) |
|
25 */ |
|
26 public static function getOrCreateTag($tag_label, $doctrine) |
|
27 { |
|
28 $tag_label_normalized = WikiTagUtils::normalizeTag($tag_label); |
|
29 // We get the wikipedia references for the tag_label |
|
30 // We get or create the tag object |
|
31 $tags = $doctrine->getRepository('WikiTagBundle:Tag')->findBy(array('normalizedLabel' => $tag_label_normalized)); |
|
32 $tag = null; |
|
33 foreach ($tags as $t){ |
|
34 if($tag==null || $t->getUrlStatus()!=Tag::$TAG_URL_STATUS_DICT['null_result']){ |
|
35 $tag = $t; |
|
36 if($t->getUrlStatus()!=Tag::$TAG_URL_STATUS_DICT['null_result']){ |
|
37 break; |
|
38 } |
|
39 } |
|
40 } |
|
41 $wp_request_done = false; |
|
42 if($tag==null){ |
|
43 $tag = new Tag(); |
|
44 $tag->setLabel($tag_label_normalized); |
|
45 $tag->setOriginalLabel($tag_label); |
|
46 $tag->setNormalizedLabel($tag_label_normalized); |
|
47 $created = true; |
|
48 } |
|
49 else{ |
|
50 $created = false; |
|
51 $match_exists = false; |
|
52 // Even if a tag with the normalised label exists, IF this tag is not wikipedia semantised, |
|
53 // we search if a wikipedia semantised version exists in the base |
|
54 foreach ($tags as $t){ |
|
55 if($t->getUrlStatus()==Tag::$TAG_URL_STATUS_DICT['match']){ |
|
56 $tag = $t; |
|
57 $match_exists = true; |
|
58 break; |
|
59 } |
|
60 } |
|
61 if($match_exists==false){ |
|
62 $wp_response = WikiTagUtils::getWikipediaInfo($tag_label_normalized); |
|
63 $status = $wp_response['status']; |
|
64 if($status==Tag::$TAG_URL_STATUS_DICT['match']){ |
|
65 $tag = new Tag(); |
|
66 $tag->setLabel($tag_label_normalized); |
|
67 $tag->setOriginalLabel($tag_label); |
|
68 $tag->setNormalizedLabel($tag_label_normalized); |
|
69 $created = true; |
|
70 $wp_request_done = true; |
|
71 } |
|
72 } |
|
73 } |
|
74 |
|
75 // We request Wikipedia if the tag is created |
|
76 if($created==true){ |
|
77 if($wp_request_done==false){ |
|
78 $wp_response = WikiTagUtils::getWikipediaInfo($tag_label_normalized); |
|
79 } |
|
80 $new_label = $wp_response['new_label']; |
|
81 $status = $wp_response['status']; |
|
82 $url = $wp_response['wikipedia_url']; |
|
83 $pageid = $wp_response['pageid']; |
|
84 $dbpedia_uri = $wp_response["dbpedia_uri"]; |
|
85 $wikipedia_revision_id = $wp_response['revision_id']; |
|
86 |
|
87 # We save the datas |
|
88 if($new_label!=null){ |
|
89 $tag->setLabel($new_label); |
|
90 } |
|
91 if($status!=null){ |
|
92 $tag->setUrlStatus($status); |
|
93 } |
|
94 $tag->setWikipediaUrl($url); |
|
95 $tag->setWikipediaPageId($pageid); |
|
96 $tag->setDbpediaUri($dbpedia_uri); |
|
97 |
|
98 // Save datas. |
|
99 $em = $doctrine->getEntityManager(); |
|
100 $em->persist($tag); |
|
101 $em->flush(); |
|
102 |
|
103 } |
|
104 else if($tag!=null && $tag->getWikipediaPageId()!=null){ |
|
105 $wp_response = WikiTagUtils::getWikipediaInfo(null, $tag->getWikipediaPageId()); |
|
106 $wikipedia_revision_id = $wp_response['revision_id']; |
|
107 } |
|
108 else{ |
|
109 $wikipedia_revision_id = null; |
|
110 } |
|
111 |
|
112 return array($tag, $wikipedia_revision_id, $created);//, $wpReponse); |
|
113 } |
|
114 |
14 |
115 /** |
15 /** |
116 * Cleans the tag label |
16 * Cleans the tag label |
117 */ |
17 */ |
118 public static function normalizeTag($tag_label) |
18 public static function normalizeTag($tag_label) |
179 else{ |
79 else{ |
180 //return " MATCH"; |
80 //return " MATCH"; |
181 $status = Tag::$TAG_URL_STATUS_DICT["match"]; |
81 $status = Tag::$TAG_URL_STATUS_DICT["match"]; |
182 } |
82 } |
183 // In redirection, we have to get more datas by adding redirects=true to the params |
83 // In redirection, we have to get more datas by adding redirects=true to the params |
184 if($status==Tag::$TAG_URL_STATUS_DICT["redirection"]){ |
84 $alternative_label = null; |
|
85 $alternative_url = null; |
|
86 $alternative_pageid = null; |
|
87 if($status==Tag::$TAG_URL_STATUS_DICT["redirection"]) |
|
88 { |
|
89 //TODO: add alternative label |
185 $params['redirects'] = "true"; |
90 $params['redirects'] = "true"; |
186 $ar = WikiTagUtils::requestWikipedia($params); |
91 $ar = WikiTagUtils::requestWikipedia($params); |
187 $res = $ar[0]; |
92 $res = $ar[0]; |
188 $pages = $ar[1]; |
93 $pages = $ar[1]; |
189 #we know that we have at least one answer |
94 #we know that we have at least one answer |
190 if(count($pages)>1 || count($pages)==0){ |
95 if(count($pages)>1 || count($pages)==0){ |
191 return WikiTagUtils::returnNullResult($res); |
96 return WikiTagUtils::returnNullResult($res); |
192 } |
97 } |
193 // get first result |
98 // get first result |
194 $page = reset($pages); |
99 $page = reset($pages); |
|
100 $alternative_label = array_key_exists('title', $page) ? $page['title'] : null; |
|
101 $alternative_url = array_key_exists('fullurl', $page) ? $page['fullurl'] : null; |
|
102 $alternative_pageid = array_key_exists('pageid', $page) ? $page['pageid'] : null; |
195 } |
103 } |
196 |
104 |
197 $revision_id = $page['lastrevid']; |
105 $revision_id = $page['lastrevid']; |
198 |
106 |
199 // process language to extract the english label |
107 // process language to extract the english label |
212 $dbpedia_uri = null; |
120 $dbpedia_uri = null; |
213 if($english_label!=null && strpos($english_label, '#')===false){ |
121 if($english_label!=null && strpos($english_label, '#')===false){ |
214 $dbpedia_uri = WikiTagUtils::getDbpediaUri($english_label); |
122 $dbpedia_uri = WikiTagUtils::getDbpediaUri($english_label); |
215 } |
123 } |
216 |
124 |
217 $wp_response = array('new_label'=>$new_label, 'status'=>$status, 'wikipedia_url'=>$url, 'pageid'=>$pageid, 'dbpedia_uri'=>$dbpedia_uri, 'revision_id'=>$revision_id, 'response'=>$original_response); |
125 $wp_response = array( |
|
126 'new_label'=>$new_label, |
|
127 'alternative_label'=>$alternative_label, |
|
128 'status'=>$status, |
|
129 'wikipedia_url'=>$url, |
|
130 'wikipedia_alternative_url'=>$alternative_url, |
|
131 'pageid'=>$pageid, |
|
132 'alternative_pageid'=>$alternative_pageid, |
|
133 'dbpedia_uri'=>$dbpedia_uri, |
|
134 'revision_id'=>$revision_id, |
|
135 'response'=>$original_response); |
218 //return $url." <br/>RES = ".$res/*." <br/>DUMP = ".var_dump($pages)*/." <br/>COUNT = ".count($pages)." <br/>page = ".var_dump($page); |
136 //return $url." <br/>RES = ".$res/*." <br/>DUMP = ".var_dump($pages)*/." <br/>COUNT = ".count($pages)." <br/>page = ".var_dump($page); |
219 return $wp_response; |
137 return $wp_response; |
220 } |
138 } |
221 |
139 |
222 |
140 |