|
1 # -*- coding: utf-8 -*- |
|
2 ''' |
|
3 @author: raphv |
|
4 ''' |
|
5 from django.core.management.base import BaseCommand, CommandError |
|
6 from django.utils.http import urlquote |
|
7 import django.utils.simplejson as json |
|
8 from SPARQLWrapper import SPARQLWrapper, JSON |
|
9 |
|
10 class Command(BaseCommand): |
|
11 ''' |
|
12 Command to export tags |
|
13 ''' |
|
14 args = '<path_to_geojson_file>' |
|
15 options = '' |
|
16 help = """Adds semantic data to countries.geo.json files""" |
|
17 |
|
18 def handle(self, *args, **options): |
|
19 |
|
20 if len(args) == 0 or not args[0]: |
|
21 raise CommandError("Give a Geo Json File to process") |
|
22 |
|
23 geojsonfile = args[0] |
|
24 |
|
25 f = open(geojsonfile,"r") |
|
26 geojson = json.loads(f.read()) |
|
27 f.close() |
|
28 |
|
29 labels = [feature['properties']['name'] for feature in geojson['features']] |
|
30 uris = dict([(label, 'http://dbpedia.org/resource/' + urlquote(label.replace(' ','_'))) for label in labels]) |
|
31 |
|
32 uri_list = [uris[label] for label in uris] |
|
33 base_uris = dict([(uris[label],uris[label]) for label in uris]) |
|
34 |
|
35 grp = 10 |
|
36 endpoint = SPARQLWrapper("http://dbpedia.org/sparql") |
|
37 endpoint.setReturnFormat(JSON) |
|
38 |
|
39 start = 0 |
|
40 while start < len(uri_list): |
|
41 sparql = """ |
|
42 PREFIX dbpedia-owl: <http://dbpedia.org/ontology/> |
|
43 SELECT ?source ?target |
|
44 WHERE { |
|
45 ?source dbpedia-owl:wikiPageRedirects ?target . |
|
46 Filter (?source in (<%s>)) . |
|
47 } |
|
48 """ % '>, <'.join(uri_list[start:start + grp]) |
|
49 endpoint.setQuery(sparql) |
|
50 results = endpoint.query().convert() |
|
51 |
|
52 for r in results["results"]["bindings"]: |
|
53 source = r["source"]["value"] |
|
54 target = r["target"]["value"] |
|
55 base_uris[source] = target |
|
56 #print "%s redirects to %s"%(source, target) |
|
57 start += grp |
|
58 |
|
59 base_uri_list = [base_uris[uri] for uri in base_uris] |
|
60 |
|
61 french_labels = {} |
|
62 |
|
63 start = 0 |
|
64 while start < len(base_uri_list): |
|
65 |
|
66 sparql = """ |
|
67 PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#> |
|
68 SELECT ?country ?label WHERE { |
|
69 ?country rdfs:label ?label . |
|
70 FILTER (?country in (<%s>)) . |
|
71 FILTER langMatches( lang(?label), "fr" ) |
|
72 } |
|
73 """ % '>, <'.join(base_uri_list[start:start + grp]) |
|
74 |
|
75 endpoint.setQuery(sparql) |
|
76 results = endpoint.query().convert() |
|
77 |
|
78 for r in results["results"]["bindings"]: |
|
79 country = r["country"]["value"] |
|
80 label = r["label"]["value"] |
|
81 french_labels[country] = label |
|
82 print "%s label: %s"%(country, label) |
|
83 |
|
84 start += grp |
|
85 |
|
86 for feature in geojson['features']: |
|
87 base_uri = base_uris[uris[feature['properties']['name']]] |
|
88 feature['properties']['dbpedia_uri'] = base_uri |
|
89 if french_labels.has_key(base_uri): |
|
90 feature['properties']['label_fr'] = french_labels[base_uri] |
|
91 else: |
|
92 print "No label for %s"%base_uri |
|
93 |
|
94 f = open(geojsonfile,"w") |
|
95 f.write(json.dumps(geojson)) |
|
96 f.close() |