|
123
|
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 |
|
|
135
|
10 |
#import pydevd #@UnresolvedImport |
|
|
11 |
#pydevd.settrace(suspend=False) |
|
|
12 |
|
|
|
13 |
|
|
123
|
14 |
class Command(BaseCommand): |
|
|
15 |
''' |
|
|
16 |
Command to export tags |
|
|
17 |
''' |
|
|
18 |
args = '<path_to_geojson_file>' |
|
|
19 |
options = '' |
|
|
20 |
help = """Adds semantic data to countries.geo.json files""" |
|
|
21 |
|
|
|
22 |
def handle(self, *args, **options): |
|
|
23 |
|
|
|
24 |
if len(args) == 0 or not args[0]: |
|
|
25 |
raise CommandError("Give a Geo Json File to process") |
|
|
26 |
|
|
|
27 |
geojsonfile = args[0] |
|
|
28 |
|
|
|
29 |
f = open(geojsonfile,"r") |
|
|
30 |
geojson = json.loads(f.read()) |
|
|
31 |
f.close() |
|
|
32 |
|
|
|
33 |
labels = [feature['properties']['name'] for feature in geojson['features']] |
|
|
34 |
uris = dict([(label, 'http://dbpedia.org/resource/' + urlquote(label.replace(' ','_'))) for label in labels]) |
|
|
35 |
|
|
|
36 |
uri_list = [uris[label] for label in uris] |
|
|
37 |
base_uris = dict([(uris[label],uris[label]) for label in uris]) |
|
|
38 |
|
|
|
39 |
grp = 10 |
|
|
40 |
endpoint = SPARQLWrapper("http://dbpedia.org/sparql") |
|
|
41 |
endpoint.setReturnFormat(JSON) |
|
|
42 |
|
|
|
43 |
start = 0 |
|
|
44 |
while start < len(uri_list): |
|
|
45 |
sparql = """ |
|
|
46 |
PREFIX dbpedia-owl: <http://dbpedia.org/ontology/> |
|
|
47 |
SELECT ?source ?target |
|
|
48 |
WHERE { |
|
|
49 |
?source dbpedia-owl:wikiPageRedirects ?target . |
|
|
50 |
Filter (?source in (<%s>)) . |
|
|
51 |
} |
|
|
52 |
""" % '>, <'.join(uri_list[start:start + grp]) |
|
|
53 |
endpoint.setQuery(sparql) |
|
|
54 |
results = endpoint.query().convert() |
|
|
55 |
|
|
|
56 |
for r in results["results"]["bindings"]: |
|
|
57 |
source = r["source"]["value"] |
|
|
58 |
target = r["target"]["value"] |
|
|
59 |
base_uris[source] = target |
|
|
60 |
#print "%s redirects to %s"%(source, target) |
|
|
61 |
start += grp |
|
|
62 |
|
|
|
63 |
base_uri_list = [base_uris[uri] for uri in base_uris] |
|
|
64 |
|
|
|
65 |
french_labels = {} |
|
135
|
66 |
labels = {} |
|
123
|
67 |
|
|
|
68 |
start = 0 |
|
|
69 |
while start < len(base_uri_list): |
|
|
70 |
|
|
|
71 |
sparql = """ |
|
|
72 |
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#> |
|
|
73 |
SELECT ?country ?label WHERE { |
|
|
74 |
?country rdfs:label ?label . |
|
135
|
75 |
FILTER (?country in (<%s>)) |
|
123
|
76 |
} |
|
|
77 |
""" % '>, <'.join(base_uri_list[start:start + grp]) |
|
|
78 |
|
|
|
79 |
endpoint.setQuery(sparql) |
|
|
80 |
results = endpoint.query().convert() |
|
|
81 |
|
|
135
|
82 |
print repr(results) |
|
|
83 |
|
|
123
|
84 |
for r in results["results"]["bindings"]: |
|
|
85 |
country = r["country"]["value"] |
|
|
86 |
label = r["label"]["value"] |
|
135
|
87 |
if country not in labels: |
|
|
88 |
labels[country] = {} |
|
|
89 |
labels[country][r["label"]["xml:lang"]] = label |
|
|
90 |
if r["label"]["xml:lang"] == 'fr': |
|
|
91 |
french_labels[country] = label |
|
123
|
92 |
print "%s label: %s"%(country, label) |
|
|
93 |
|
|
|
94 |
start += grp |
|
|
95 |
|
|
|
96 |
for feature in geojson['features']: |
|
|
97 |
base_uri = base_uris[uris[feature['properties']['name']]] |
|
|
98 |
feature['properties']['dbpedia_uri'] = base_uri |
|
135
|
99 |
|
|
|
100 |
if labels.has_key(base_uri): |
|
|
101 |
feature['properties']['labels'] = labels[base_uri] |
|
123
|
102 |
if french_labels.has_key(base_uri): |
|
|
103 |
feature['properties']['label_fr'] = french_labels[base_uri] |
|
|
104 |
else: |
|
|
105 |
print "No label for %s"%base_uri |
|
|
106 |
|
|
|
107 |
f = open(geojsonfile,"w") |
|
|
108 |
f.write(json.dumps(geojson)) |
|
|
109 |
f.close() |