| | var WIKIPEDIA = function() {
2 | var my = {}; |
| DBPedia SPARQL endpoint | my.endpoint = 'http://dbpedia.org/sparql/'; |
getData
3 |
4 | Return structured information (via callback) on the provided Wikipedia URL by querying
5 | the DBPedia SPARQL endpoint and then tidying the data up.
6 |
7 | Data is return in the form of the following hash:
8 |
9 | {
10 | raw: the-raw-json-from-dbpedia,
11 | summary: a-cleaned-up-set-of-the-properties (see extractSummary),
12 | dbpediaUrl: dbpedia-resource-url e.g. http://dbpedia.org/resource/WorldWarII
13 | }
14 |
15 | Function is asynchronous as we have to call out to DBPedia to get the
16 | info. | my.getData = function(wikipediaUrl, callback, error) {
17 | var url = my._getDbpediaUrl(wikipediaUrl);
18 | function onSuccess(data) {
19 | callback({
20 | raw: data,
21 | dbpediaUrl: url,
22 | summary: my.extractSummary(url, data)
23 | })
24 | }
25 | my.getRawJSON(url, onSuccess, error);
26 | } |
_getDbpediaUrl
27 |
28 | Convert a Wikipedia url convert to DBPedia url | my._getDbpediaUrl = function(url) {
29 | if (url.indexOf('wikipedia')) {
30 | var parts = url.split('/');
31 | var title = parts[parts.length-1];
32 | url = 'http://dbpedia.org/resource/' + title;
33 | }
34 | return url;
35 | }; |
getRawJSON
36 |
37 | get raw RDF JSON for DBPedia resource from DBPedia SPARQL endpoint | my.getRawJSON = function(url, callback, error) {
38 | var sparqlQuery = 'DESCRIBE <{{url}}>'.replace('{{url}}', url);
39 | var jqxhr = $.ajax({
40 | url: my.endpoint,
41 | data: {
42 | query: sparqlQuery, |
| format: 'application/x-json+ld' | format: 'application/rdf+json'
43 | },
44 | success: callback,
45 | error: error
46 | });
47 | }; |
| Standard RDF namespace prefixes for use in lookupProperty function | my.PREFIX = {
48 | rdf: "http://www.w3.org/1999/02/22-rdf-syntax-ns#",
49 | rdfs: "http://www.w3.org/2000/01/rdf-schema#",
50 | xsd: "http://www.w3.org/2001/XMLSchema#",
51 | owl: "http://www.w3.org/2002/07/owl#",
52 | dc: "http://purl.org/dc/terms/",
53 | foaf: "http://xmlns.com/foaf/0.1/",
54 | vcard: "http://www.w3.org/2006/vcard/ns#",
55 | dbp: "http://dbpedia.org/property/",
56 | dbo: "http://dbpedia.org/ontology/",
57 | geo: "http://www.geonames.org/ontology#",
58 | wgs: "http://www.w3.org/2003/01/geo/wgs84_pos#"
59 | }; |
lookupProperty
60 |
61 | lookup a property value given a standard RDF/JSON property dictionary
62 | e.g. something like ...
63 |
64 | ...
65 | "http://dbpedia.org/property/regent": [
66 | {
67 | "type": "uri",
68 | "value": "http://dbpedia.org/resource/Richard_I_of_England"
69 | }
70 | ],
71 | ...
72 |
| my._lookupProperty = function(dict, property) { |
| first expand namespace | for(key in WIKIPEDIA.PREFIX) {
73 | if (property.indexOf(key + ':') == 0) {
74 | property = WIKIPEDIA.PREFIX[key] + property.slice(key.length + 1);
75 | }
76 | }
77 | var values = dict[property];
78 | for (idx in values) { |
| only take english values if lang is present | if (!values[idx]['lang'] || values[idx].lang == 'en') {
79 | return values[idx].value;
80 | }
81 | }
82 | }; |
| Extract a standard set of attributes (e.g. title, description, dates etc
83 | etc) from rdfJson and the given subject uri (url) e.g.
84 |
85 | extractSummary('http://dbpedia.org/resource/Rufus_Pollock', rdfJson object from dbpedia)
86 |
| my.extractSummary = function(subjectUri, rdfJson) {
87 | var properties = rdfJson[subjectUri];
88 | function lkup(attribs) {
89 | if (attribs instanceof Array) {
90 | var out = [];
91 | for (idx in attribs) {
92 | var _tmp = my._lookupProperty(properties, attribs[idx]);
93 | if (_tmp) {
94 | out.push(_tmp);
95 | }
96 | }
97 | return out;
98 | } else {
99 | return my._lookupProperty(properties, attribs);
100 | }
101 | }
102 |
103 | var summaryInfo = {
104 | title: lkup('rdfs:label'),
105 | description: lkup('dbo:abstract'),
106 | summary: lkup('rdfs:comment'),
107 | birthDate: lkup('dbp:birthDate'),
108 | deathDate: lkup('dbp:deathDate'), |
| both dbp:date and dbo:date are usually present but dbp:date is
109 | frequently "bad" (e.g. just a single integer rather than a date)
110 | whereas ontology value is better | date: lkup('dbo:date'),
111 | place: lkup('dbp:place'),
112 | birthPlace: lkup('dpb:birthPlace'),
113 | deathPlace: lkup('dpb:deathPlace'),
114 | source: lkup('foaf:page'),
115 | images: lkup(['dbo:thumbnail', 'foaf:depiction', 'foaf:img']),
116 | location: {
117 | lat: lkup('wgs:lat'),
118 | lon: lkup('wgs:lon')
119 | }
120 | };
121 |
122 | summaryInfo.start = summaryInfo.birthDate || summaryInfo.date;
123 | summaryInfo.end = summaryInfo.deathDate;
124 | summaryInfo.location.title = summaryInfo.place || summaryInfo.birthPlace ||
125 | summaryInfo.deathPlace;
126 | summaryInfo.image = summaryInfo.images ? summaryInfo.images[0] : null;
127 |
128 | return summaryInfo;
129 | };
130 |
131 | return my;
132 | }();
133 |
134 | |
49 |
50 |
51 | 
94 |