on master help sign in
[fix][mod] wikidata date handling refactor - fixes #387
commit362c849797e2e6f0e232642c23744c47a75cdfd4 authorAdam Tauber <asciimoo@gmail.com> dateMon 7 Sep 2015, 22:39 (11 years ago) parent7580852
2 files changed+36−26
searx/engines/wikidata.py+26−8 searx/utils.py+10−18
searx/engines/wikidata.py+26−8
@@ -1,8 +1,15 @@
11import json
2−from urllib import urlencode
2+
3+from searx import logger
34from searx.poolrequests import get
45from searx.utils import format_date_by_locale
56
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')
613result_count = 1
714wikidata_host = 'https://www.wikidata.org'
815wikidata_api = wikidata_host + '/w/api.php'
@@ -164,14 +171,12 @@ def getDetail(jsonresponse, wikidata_id, language, locale):
164171 if postal_code is not None:
165172 attributes.append({'label': 'Postal code(s)', 'value': postal_code})
166173
167− date_of_birth = get_time(claims, 'P569', None)
174+ date_of_birth = get_time(claims, 'P569', locale, None)
168175 if date_of_birth is not None:
169− date_of_birth = format_date_by_locale(date_of_birth[8:], locale)
170176 attributes.append({'label': 'Date of birth', 'value': date_of_birth})
171177
172− date_of_death = get_time(claims, 'P570', None)
178+ date_of_death = get_time(claims, 'P570', locale, None)
173179 if date_of_death is not None:
174− date_of_death = format_date_by_locale(date_of_death[8:], locale)
175180 attributes.append({'label': 'Date of death', 'value': date_of_death})
176181
177182 if len(attributes) == 0 and len(urls) == 2 and len(description) == 0:
@@ -229,7 +234,7 @@ def get_string(claims, propertyName, defaultValue=None):
229234 return result[0]
230235
231236
232−def get_time(claims, propertyName, defaultValue=None):
237+def get_time(claims, propertyName, locale, defaultValue=None):
233238 propValue = claims.get(propertyName, {})
234239 if len(propValue) == 0:
235240 return defaultValue
@@ -244,9 +249,22 @@ def get_time(claims, propertyName, defaultValue=None):
244249 result.append(value.get('time', ''))
245250
246251 if len(result) == 0:
247− return defaultValue
252+ date_string = defaultValue
248253 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)
250268
251269
252270def get_geolink(claims, propertyName, defaultValue=''):
searx/utils.py+10−18
@@ -1,11 +1,10 @@
11# import htmlentitydefs
2−import locale
3−import dateutil.parser
42import cStringIO
53import csv
64import os
75import re
86
7+from babel.dates import format_date
98from codecs import getincrementalencoder
109from HTMLParser import HTMLParser
1110from random import choice
@@ -195,23 +194,16 @@ def get_result_templates(base_path):
195194 return result_templates
196195
197196
198−def format_date_by_locale(date_string, locale_string):
197+def format_date_by_locale(date, locale_string):
199198 # strftime works only on dates after 1900
200− parsed_date = dateutil.parser.parse(date_string)
201− if parsed_date.year <= 1900:
202− return parsed_date.isoformat().split('T')[0]
203−
204− orig_locale = locale.getlocale()[0]
205− try:
206− locale.setlocale(locale.LC_ALL, locale_string)
207− except:
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
199+
200+ if date.year <= 1900:
201+ return date.isoformat().split('T')[0]
202+
203+ if locale_string == 'all':
204+ locale_string = settings['ui']['default_locale'] or 'en_US'
205+
206+ return format_date(date, locale=locale_string)
215207
216208
217209def dict_subset(d, properties):