Utils/WikiTagUtils.php
author cavaliet
Wed, 30 Nov 2011 17:58:18 +0100
changeset 50 e967654e90cb
parent 43 54f204bceb28
child 60 984ba20c150b
permissions -rwxr-xr-x
First step of error management when Wikipedia request fails. Set up in whole list and document list.
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
     1
<?php
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
     2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
     3
namespace IRI\Bundle\WikiTagBundle\Utils;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
     4
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
     5
use IRI\Bundle\WikiTagBundle\Entity\Tag;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
     6
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
     7
class WikiTagUtils
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
     8
{
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
     9
    // Constants
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    10
    private static $WIKIPEDIA_API_URL = "http://fr.wikipedia.org/w/api.php";
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    11
    private static $WIKIPEDIA_VERSION_PERMALINK_TEMPLATE = "http://fr.wikipedia.org/w/index.php?oldid=%s";
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    12
    private static $DBPEDIA_URI_TEMPLATE = "http://dbpedia.org/resource/%s";
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    13
    
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    14
    
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    15
    /**
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    16
     * Cleans the tag label
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    17
     */
8
7d2fb5d7c9ff debug addtag for document
cavaliet
parents: 2
diff changeset
    18
    public static function normalizeTag($tag_label)
2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    19
    {
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    20
        if(strlen($tag_label)==0){
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    21
            return $tag_label;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    22
        }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    23
        $tag_label = trim($tag_label);//tag.strip()
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    24
        $tag_label = str_replace("_", " ", $tag_label);//tag.replace("_", " ")
43
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    25
        $tag_label = preg_replace('/\s+/u', ' ', $tag_label);//" ".join(tag.split())
2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    26
        $tag_label = ucfirst($tag_label);//tag[0].upper() + tag[1:]
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    27
        return $tag_label;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    28
    }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    29
    
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    30
    /**
43
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    31
     * Query wikipedia with a normalized label or a pageid
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    32
     * return an array with the form
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    33
     * array(
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    34
     *      'new_label'=>$new_label,
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    35
     *   	'alternative_label'=>$alternative_label,
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    36
     *   	'status'=>$status,
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    37
     *   	'wikipedia_url'=>$url,
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    38
     *      'wikipedia_alternative_url'=>$alternative_url,
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    39
     *   	'pageid'=>$pageid,
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    40
     *   	'alternative_pageid'=>$alternative_pageid,
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    41
     *   	'dbpedia_uri'=>$dbpedia_uri,
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    42
     *   	'revision_id'=> ,
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    43
     *   	'response'=> the original wikipedia json response
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    44
     *   	)
2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    45
     *
43
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    46
     * @param string $tag_label_normalized
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    47
     * @param bigint $page_id
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
    48
     * @return array
2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    49
     */
22
99c15cfe420b Add ModifyTag from tag list. Add Reset Wikipedia info from tag list. Enable boolean from addJavascript controller/template to switch from list or document context.
cavaliet
parents: 9
diff changeset
    50
    public static function getWikipediaInfo($tag_label_normalized, $page_id=null)
2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    51
    {
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    52
        $params = array('action'=>'query', 'prop'=>'info|categories|langlinks', 'inprop'=>'url', 'lllimit'=>'500', 'cllimit'=>'500', 'rvprop'=>'ids', 'format'=>'json');
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    53
        if($tag_label_normalized!=null){
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    54
            $params['titles'] = urlencode($tag_label_normalized);
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    55
        }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    56
        else if($page_id!=null){
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    57
            $params['pageids'] = $page_id;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    58
        }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    59
        else{
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    60
            return WikiTagUtils::returnNullResult(null);
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    61
        }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    62
        
50
e967654e90cb First step of error management when Wikipedia request fails. Set up in whole list and document list.
cavaliet
parents: 43
diff changeset
    63
        try {
e967654e90cb First step of error management when Wikipedia request fails. Set up in whole list and document list.
cavaliet
parents: 43
diff changeset
    64
            $ar = WikiTagUtils::requestWikipedia($params);
e967654e90cb First step of error management when Wikipedia request fails. Set up in whole list and document list.
cavaliet
parents: 43
diff changeset
    65
        }
e967654e90cb First step of error management when Wikipedia request fails. Set up in whole list and document list.
cavaliet
parents: 43
diff changeset
    66
        catch (\Exception $e){
e967654e90cb First step of error management when Wikipedia request fails. Set up in whole list and document list.
cavaliet
parents: 43
diff changeset
    67
            throw new \Exception($e->getMessage());
e967654e90cb First step of error management when Wikipedia request fails. Set up in whole list and document list.
cavaliet
parents: 43
diff changeset
    68
        }
2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    69
        $res = $ar[0];
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    70
        $original_response = $res;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    71
        $pages = $ar[1];
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    72
        // If there 0 or more than 1 result, the query has failed
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    73
        if(count($pages)>1 || count($pages)==0){
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    74
            return WikiTagUtils::returnNullResult($res);
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    75
        }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    76
        // get first result
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    77
        $page = reset($pages);
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    78
        // Unknow entry ?
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    79
        if(array_key_exists('missing', $page) || array_key_exists('invalid', $page)){
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    80
            return WikiTagUtils::returnNullResult($res);
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    81
        }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    82
        // The entry exists, we get the datas.
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    83
        $url = array_key_exists('fullurl', $page) ? $page['fullurl'] : null;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    84
        $pageid = array_key_exists('pageid', $page) ? $page['pageid'] : null;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    85
        $new_label = array_key_exists('title', $page) ? $page['title'] : null;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    86
        // We test the status (redirect first because a redirect has no categories key)
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    87
        if(array_key_exists('redirect', $page)){
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    88
            //return " REDIRECT";
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    89
            $status = Tag::$TAG_URL_STATUS_DICT["redirection"];
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    90
        }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    91
        else if(WikiTagUtils::isHomonymy($page)){
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    92
            //return " HOMONYMY";
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    93
            $status = Tag::$TAG_URL_STATUS_DICT["homonyme"];
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    94
        }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    95
        else{
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    96
            //return " MATCH";
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    97
            $status = Tag::$TAG_URL_STATUS_DICT["match"];
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    98
        }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
    99
        // In redirection, we have to get more datas by adding redirects=true to the params
42
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   100
        $alternative_label = null;
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   101
        $alternative_url = null;
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   102
        $alternative_pageid = null;
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   103
        if($status==Tag::$TAG_URL_STATUS_DICT["redirection"])
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   104
        {
2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   105
            $params['redirects'] = "true";
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   106
            $ar = WikiTagUtils::requestWikipedia($params);
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   107
            $res = $ar[0];
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   108
            $pages = $ar[1];
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   109
            #we know that we have at least one answer
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   110
            if(count($pages)>1 || count($pages)==0){
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   111
                return WikiTagUtils::returnNullResult($res);
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   112
            }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   113
            // get first result
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   114
            $page = reset($pages);
42
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   115
            $alternative_label = array_key_exists('title', $page) ? $page['title'] : null;
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   116
            $alternative_url = array_key_exists('fullurl', $page) ? $page['fullurl'] : null;
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   117
            $alternative_pageid = array_key_exists('pageid', $page) ? $page['pageid'] : null;
2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   118
        }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   119
        
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   120
        $revision_id = $page['lastrevid'];
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   121
        
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   122
        // process language to extract the english label
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   123
        $english_label = null;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   124
        if($status==Tag::$TAG_URL_STATUS_DICT["match"] || $status==Tag::$TAG_URL_STATUS_DICT["redirection"]){
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   125
            if(array_key_exists("langlinks", $page)){
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   126
                foreach ($page["langlinks"] as $ar) {
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   127
                    if($ar["lang"]=="en"){
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   128
                        $english_label = $ar["*"];
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   129
                        break;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   130
                    }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   131
                }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   132
            }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   133
        }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   134
        // We create the dbpedia uri.
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   135
        $dbpedia_uri = null;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   136
        if($english_label!=null && strpos($english_label, '#')===false){
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   137
            $dbpedia_uri = WikiTagUtils::getDbpediaUri($english_label);
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   138
        }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   139
        
42
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   140
        $wp_response = array(
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   141
            'new_label'=>$new_label,
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   142
        	'alternative_label'=>$alternative_label,
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   143
        	'status'=>$status,
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   144
        	'wikipedia_url'=>$url,
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   145
            'wikipedia_alternative_url'=>$alternative_url,
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   146
        	'pageid'=>$pageid,
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   147
        	'alternative_pageid'=>$alternative_pageid,
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   148
        	'dbpedia_uri'=>$dbpedia_uri,
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   149
        	'revision_id'=>$revision_id,
0e57c730bb18 Documentation and add alternative wp url and label + migrations
ymh <ymh.work@gmail.com>
parents: 32
diff changeset
   150
        	'response'=>$original_response);
2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   151
        //return $url." <br/>RES =  ".$res/*." <br/>DUMP =  ".var_dump($pages)*/." <br/>COUNT =  ".count($pages)." <br/>page =  ".var_dump($page);
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   152
        return $wp_response;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   153
    }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   154
    
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   155
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   156
    /**
43
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
   157
     * build and do the request to Wikipedia.
2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   158
     *
43
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
   159
     * @param array $params
54f204bceb28 Correct migration
ymh <ymh.work@gmail.com>
parents: 42
diff changeset
   160
     * @return array
2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   161
     */
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   162
    private static function requestWikipedia($params)
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   163
    {
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   164
        $params_str = '';
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   165
        foreach ($params as $key => $value) {
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   166
            if ($params_str==''){
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   167
                $params_str = $key.'='.$value;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   168
            }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   169
            else{
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   170
                $params_str .= '&'.$key.'='.$value;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   171
            }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   172
        }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   173
        
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   174
        $url = WikiTagUtils::$WIKIPEDIA_API_URL.'?'.$params_str;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   175
        
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   176
        $ch = curl_init();
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   177
        curl_setopt($ch, CURLOPT_URL, $url);
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   178
        curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   179
        curl_setopt($ch, CURLOPT_USERAGENT, 'http://www.iri.centrepompidou.fr');
50
e967654e90cb First step of error management when Wikipedia request fails. Set up in whole list and document list.
cavaliet
parents: 43
diff changeset
   180
        curl_setopt($ch, CURLOPT_TIMEOUT_MS, 5000);
2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   181
        $res = curl_exec($ch);
50
e967654e90cb First step of error management when Wikipedia request fails. Set up in whole list and document list.
cavaliet
parents: 43
diff changeset
   182
        $curl_errno = curl_errno($ch);
e967654e90cb First step of error management when Wikipedia request fails. Set up in whole list and document list.
cavaliet
parents: 43
diff changeset
   183
        $curl_error = curl_error($ch);
2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   184
        curl_close($ch);
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   185
        
50
e967654e90cb First step of error management when Wikipedia request fails. Set up in whole list and document list.
cavaliet
parents: 43
diff changeset
   186
        if ($curl_errno > 0) {
e967654e90cb First step of error management when Wikipedia request fails. Set up in whole list and document list.
cavaliet
parents: 43
diff changeset
   187
            throw new \Exception("Wikipedia request failed. cURLError #$curl_errno: $curl_error\n");
e967654e90cb First step of error management when Wikipedia request fails. Set up in whole list and document list.
cavaliet
parents: 43
diff changeset
   188
        }
e967654e90cb First step of error management when Wikipedia request fails. Set up in whole list and document list.
cavaliet
parents: 43
diff changeset
   189
        
2
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   190
        $val = json_decode($res, true);
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   191
        $pages = $val["query"]["pages"];
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   192
        return array($res, $pages);
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   193
    }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   194
    
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   195
    /**
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   196
     * Returns tag with a null result, usually used after a failed request on Wikipedia
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   197
     */
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   198
    private static function returnNullResult($response)
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   199
    {
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   200
        return array('new_label'=>null, 'status'=>Tag::$TAG_URL_STATUS_DICT['null_result'], 'wikipedia_url'=>null, 'pageid'=>null, 'dbpedia_uri'=>null, 'revision_id'=>null, 'response'=>$response);
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   201
    }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   202
    
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   203
    /**
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   204
     * Returns tag with a null result, usually used after a failed request on Wikipedia
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   205
     */
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   206
    private static function isHomonymy($page)
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   207
    {
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   208
        //$s = "";
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   209
        foreach ($page["categories"] as $ar) {
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   210
            //$s .= ", b : ".$ar." - title = ".$ar["title"].", strpos = ".strpos($ar["title"], 'Catégorie:Homonymie');
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   211
            // Strict test because false can be seen as "O".
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   212
            if(strpos($ar["title"], 'Catégorie:Homonymie')!==false || strpos($ar["title"], 'Category:Disambiguation')!==false){
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   213
                //$s .= "TRUE";
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   214
                return true;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   215
            }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   216
        }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   217
        return false;
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   218
    }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   219
    
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   220
    /**
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   221
     * Builds DbPedia URI
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   222
     */
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   223
    private static function getDbpediaUri($english_label)
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   224
    {
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   225
        return sprintf(WikiTagUtils::$DBPEDIA_URI_TEMPLATE, WikiTagUtils::urlize_for_wikipedia($english_label));
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   226
    }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   227
    
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   228
    /**
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   229
     * URLencode label for wikipedia
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   230
     */
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   231
    private static function urlize_for_wikipedia($label){
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   232
        return urlencode(str_replace(" ", "_", $label));
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   233
    }
13f43f53d0ba first implementation
ymh <ymh.work@gmail.com>
parents:
diff changeset
   234
}