@@ -1,13 +1,12 @@
11import json
2−from datetime import datetime
32from requests import get
43from urllib import urlencode
4+from datetime import datetime
55
66resultCount=2
7−urlSearch = 'https://www.wikidata.org/w/api.php?action=query&list=search&format=json&srnamespace=0&srprop=sectionsnippet&{query}'
7+urlSearch = 'https://www.wikidata.org/w/api.php?action=query&list=search&format=json&srnamespace=0&srprop=sectiontitle&{query}'
88urlDetail = 'https://www.wikidata.org/w/api.php?action=wbgetentities&format=json&props=labels%7Cinfo%7Csitelinks%7Csitelinks%2Furls%7Cdescriptions%7Cclaims&{query}'
9−# find the right URL for urlMap
10−urlMap = 'http://www.openstreetmap.org/?lat={latitude}&lon={longitude}&zoom={zoom}&layers=M'
9+urlMap = 'https://www.openstreetmap.org/?lat={latitude}&lon={longitude}&zoom={zoom}&layers=M'
1110
1211def request(query, params):
1312 params['url'] = urlSearch.format(query=urlencode({'srsearch': query, 'srlimit': resultCount}))
@@ -18,24 +17,27 @@ def request(query, params):
1817def response(resp):
1918 results = []
2019 search_res = json.loads(resp.text)
21− # TODO parallel http queries
22− before = datetime.now()
20+
21+ wikidata_ids = set()
2322 for r in search_res.get('query', {}).get('search', {}):
24− wikidata_id = r.get('title', '')
25− results = results + getDetail(wikidata_id)
26− after = datetime.now()
27− print str(after - before) + " second(s)"
23+ wikidata_ids.add(r.get('title', ''))
2824
29− return results
25+ language = resp.search_params['language'].split('_')[0]
26+ if language == 'all':
27+ language = 'en'
28+ url = urlDetail.format(query=urlencode({'ids': '|'.join(wikidata_ids), 'languages': language + '|en'}))
29+
30+ before = datetime.now()
31+ htmlresponse = get(url)
32+ print datetime.now() - before
33+ jsonresponse = json.loads(htmlresponse.content)
34+ for wikidata_id in wikidata_ids:
35+ results = results + getDetail(jsonresponse, wikidata_id, language)
3036
31−def getDetail(wikidata_id):
32− language = 'fr'
37+ return results
3338
34− url = urlDetail.format(query=urlencode({'ids': wikidata_id, 'languages': language + '|en'}))
35− print url
36− response = get(url)
37− result = json.loads(response.content)
38− result = result.get('entities', {}).get(wikidata_id, {})
39+def getDetail(jsonresponse, wikidata_id, language):
40+ result = jsonresponse.get('entities', {}).get(wikidata_id, {})
3941
4042 title = result.get('labels', {}).get(language, {}).get('value', None)
4143 if title == None:
@@ -50,7 +52,6 @@ def getDetail(wikidata_id):
5052
5153 claims = result.get('claims', {})
5254 official_website = get_string(claims, 'P856', None)
53− print official_website
5455 if official_website != None:
5556 urls.append({ 'title' : 'Official site', 'url': official_website })
5657 results.append({ 'title': title, 'url' : official_website })
@@ -98,10 +99,12 @@ def getDetail(wikidata_id):
9899
99100 return results
100101
102+
101103def add_url(urls, title, url):
102104 if url != None:
103105 urls.append({'title' : title, 'url' : url})
104106
107+
105108def get_mainsnak(claims, propertyName):
106109 propValue = claims.get(propertyName, {})
107110 if len(propValue) == 0:
@@ -110,6 +113,7 @@ def get_mainsnak(claims, propertyName):
110113 propValue = propValue[0].get('mainsnak', None)
111114 return propValue
112115
116+
113117def get_string(claims, propertyName, defaultValue=None):
114118 propValue = claims.get(propertyName, {})
115119 if len(propValue) == 0:
@@ -129,6 +133,7 @@ def get_string(claims, propertyName, defaultValue=None):
129133 else:
130134 return ', '.join(result)
131135
136+
132137def get_time(claims, propertyName, defaultValue=None):
133138 propValue = claims.get(propertyName, {})
134139 if len(propValue) == 0:
@@ -149,6 +154,7 @@ def get_time(claims, propertyName, defaultValue=None):
149154 else:
150155 return ', '.join(result)
151156
157+
152158def get_geolink(claims, propertyName, defaultValue=''):
153159 mainsnak = get_mainsnak(claims, propertyName)
154160
@@ -182,6 +188,7 @@ def get_geolink(claims, propertyName, defaultValue=''):
182188
183189 return url
184190
191+
185192def get_wikilink(result, wikiid):
186193 url = result.get('sitelinks', {}).get(wikiid, {}).get('url', None)
187194 if url == None: