on master help sign in
[fix][mod] wikidata date handling refactor - fixes #387
362c849Adam Tauber
searx/engines/wikidata.py+26−8
@@ -1,8 +1,15 @@
1import json
1import json
2+
2−from urllib import urlencode
3+from searx import logger
3from searx.poolrequests import get
4from searx.poolrequests import get
4from searx.utils import format_date_by_locale
5from searx.utils import format_date_by_locale
5
6
7+from datetime import datetime
8+from dateutil.parser import parse as dateutil_parse
9+from urllib import urlencode
10+
11+
12+logger = logger.getChild('wikidata')
6result_count = 1
13result_count = 1
7wikidata_host = 'https://www.wikidata.org'
14wikidata_host = 'https://www.wikidata.org'
8wikidata_api = wikidata_host + '/w/api.php'
15wikidata_api = wikidata_host + '/w/api.php'
@@ -164,14 +171,12 @@ def getDetail(jsonresponse, wikidata_id, language, locale):
164 if postal_code is not None:
171 if postal_code is not None:
165 attributes.append({'label': 'Postal code(s)', 'value': postal_code})
172 attributes.append({'label': 'Postal code(s)', 'value': postal_code})
166
173
167− date_of_birth = get_time(claims, 'P569', None)
174+ date_of_birth = get_time(claims, 'P569', locale, None)
168 if date_of_birth is not None:
175 if date_of_birth is not None:
169− date_of_birth = format_date_by_locale(date_of_birth[8:], locale)
170 attributes.append({'label': 'Date of birth', 'value': date_of_birth})
176 attributes.append({'label': 'Date of birth', 'value': date_of_birth})
171
177
172− date_of_death = get_time(claims, 'P570', None)
178+ date_of_death = get_time(claims, 'P570', locale, None)
173 if date_of_death is not None:
179 if date_of_death is not None:
174− date_of_death = format_date_by_locale(date_of_death[8:], locale)
175 attributes.append({'label': 'Date of death', 'value': date_of_death})
180 attributes.append({'label': 'Date of death', 'value': date_of_death})
176
181
177 if len(attributes) == 0 and len(urls) == 2 and len(description) == 0:
182 if len(attributes) == 0 and len(urls) == 2 and len(description) == 0:
@@ -229,7 +234,7 @@ def get_string(claims, propertyName, defaultValue=None):
229 return result[0]
234 return result[0]
230
235
231
236
232−def get_time(claims, propertyName, defaultValue=None):
237+def get_time(claims, propertyName, locale, defaultValue=None):
233 propValue = claims.get(propertyName, {})
238 propValue = claims.get(propertyName, {})
234 if len(propValue) == 0:
239 if len(propValue) == 0:
235 return defaultValue
240 return defaultValue
@@ -244,9 +249,22 @@ def get_time(claims, propertyName, defaultValue=None):
244 result.append(value.get('time', ''))
249 result.append(value.get('time', ''))
245
250
246 if len(result) == 0:
251 if len(result) == 0:
247− return defaultValue
252+ date_string = defaultValue
248 else:
253 else:
249− return ', '.join(result)
254+ date_string = ', '.join(result)
255+
256+ try:
257+ parsed_date = datetime.strptime(date_string, "+%Y-%m-%dT%H:%M:%SZ")
258+ except:
259+ if date_string.startswith('-'):
260+ return date_string.split('T')[0]
261+ try:
262+ parsed_date = dateutil_parse(date_string, fuzzy=False, default=False)
263+ except:
264+ logger.debug('could not parse date %s', date_string)
265+ return date_string.split('T')[0]
266+
267+ return format_date_by_locale(parsed_date, locale)
250
268
251
269
252def get_geolink(claims, propertyName, defaultValue=''):
270def get_geolink(claims, propertyName, defaultValue=''):
searx/utils.py+10−18
@@ -1,11 +1,10 @@
1# import htmlentitydefs
1# import htmlentitydefs
2−import locale
3−import dateutil.parser
4import cStringIO
2import cStringIO
5import csv
3import csv
6import os
4import os
7import re
5import re
8
6
7+from babel.dates import format_date
9from codecs import getincrementalencoder
8from codecs import getincrementalencoder
10from HTMLParser import HTMLParser
9from HTMLParser import HTMLParser
11from random import choice
10from random import choice
@@ -195,23 +194,16 @@ def get_result_templates(base_path):
195 return result_templates
194 return result_templates
196
195
197
196
198−def format_date_by_locale(date_string, locale_string):
197+def format_date_by_locale(date, locale_string):
199 # strftime works only on dates after 1900
198 # strftime works only on dates after 1900
200− parsed_date = dateutil.parser.parse(date_string)
199+
201− if parsed_date.year <= 1900:
200+ if date.year <= 1900:
202− return parsed_date.isoformat().split('T')[0]
201+ return date.isoformat().split('T')[0]
203−
202+
204− orig_locale = locale.getlocale()[0]
203+ if locale_string == 'all':
205− try:
204+ locale_string = settings['ui']['default_locale'] or 'en_US'
206− locale.setlocale(locale.LC_ALL, locale_string)
205+
207− except:
206+ return format_date(date, locale=locale_string)
208− logger.warning('cannot set locale: {0}'.format(locale_string))
209− formatted_date = parsed_date.strftime(locale.nl_langinfo(locale.D_FMT))
210− try:
211− locale.setlocale(locale.LC_ALL, orig_locale)
212− except:
213− logger.warning('cannot set original locale: {0}'.format(orig_locale))
214− return formatted_date
215
207
216
208
217def dict_subset(d, properties):
209def dict_subset(d, properties):