# -*- coding: utf-8 -*- from lxml.html import fromstring from lxml import etree from collections import defaultdict import mock from searx.engines import wikidata from searx.testing import SearxTestCase class TestWikidataEngine(SearxTestCase): def test_request(self): query = 'test_query' dicto = defaultdict(dict) dicto['language'] = 'all' params = wikidata.request(query, dicto) self.assertIn('url', params) self.assertIn(query, params['url']) self.assertIn('wikidata.org', params['url']) dicto['language'] = 'es_ES' params = wikidata.request(query, dicto) self.assertIn(query, params['url']) # successful cases are not tested here to avoid sending additional requests def test_response(self): self.assertRaises(AttributeError, wikidata.response, None) self.assertRaises(AttributeError, wikidata.response, []) self.assertRaises(AttributeError, wikidata.response, '') self.assertRaises(AttributeError, wikidata.response, '[]') wikidata.supported_languages = ['en', 'es'] wikidata.language_aliases = {} response = mock.Mock(content=''.encode("utf-8"), search_params={"language": "en"}) self.assertEqual(wikidata.response(response), []) def test_getDetail(self): response = {} results = wikidata.getDetail(response, "Q123", "en", "en-US", etree.HTMLParser()) self.assertEqual(results, []) title_html = '
Test
' html = """
""" response = {"parse": {"displaytitle": title_html, "text": html}} results = wikidata.getDetail(response, "Q123", "en", "en-US", etree.HTMLParser()) self.assertEqual(len(results), 1) self.assertEqual(results[0]['url'], 'https://en.wikipedia.org/wiki/Test') title_html = """
Test English
""" html = """
Description English
official website English
https://officialsite.com
""" response = {"parse": {"displaytitle": title_html, "text": html}} results = wikidata.getDetail(response, "Q123", "yua", "yua_MX", etree.HTMLParser()) self.assertEqual(len(results), 2) self.assertEqual(results[0]['title'], 'Official website') self.assertEqual(results[0]['url'], 'https://officialsite.com') self.assertEqual(results[1]['infobox'], 'Test') self.assertEqual(results[1]['id'], None) self.assertEqual(results[1]['content'], 'Description') self.assertEqual(results[1]['attributes'], []) self.assertEqual(results[1]['urls'][0]['title'], 'Official website') self.assertEqual(results[1]['urls'][0]['url'], 'https://officialsite.com') self.assertEqual(results[1]['urls'][1]['title'], 'Wikipedia (en)') self.assertEqual(results[1]['urls'][1]['url'], 'https://en.wikipedia.org/wiki/Test') def test_add_image(self): image_src = wikidata.add_image(fromstring("
")) self.assertEqual(image_src, None) html = u"""
image
image.png
2,687 × 3,356; 1.22 MB
""" html_etree = fromstring(html) id_cache = wikidata.get_id_cache(html_etree) image_src = wikidata.add_image(id_cache) self.assertEqual(image_src, "https://commons.wikimedia.org/wiki/Special:FilePath/image.png?width=500&height=400") html = u"""
icon
icon.png
671 × 671; 18 KB
logo
logo.png
170 × 170; 1 KB
""" html_etree = fromstring(html) id_cache = wikidata.get_id_cache(html_etree) image_src = wikidata.add_image(id_cache) self.assertEqual(image_src, "https://commons.wikimedia.org/wiki/Special:FilePath/logo.png?width=500&height=400") def test_add_attribute(self): html = u"""
country of citizenship
""" attributes = [] html_etree = fromstring(html) id_cache = wikidata.get_id_cache(html_etree) wikidata.add_attribute(attributes, id_cache, "Fail") self.assertEqual(attributes, []) wikidata.add_attribute(attributes, id_cache, "P27") self.assertEqual(len(attributes), 1) self.assertEqual(attributes[0]["label"], "Country of citizenship") self.assertEqual(attributes[0]["value"], "United Kingdom") html = u"""
date of birth
27 January 1832 Gregorian
""" attributes = [] html_etree = fromstring(html) id_cache = wikidata.get_id_cache(html_etree) wikidata.add_attribute(attributes, id_cache, "P569", date=True) self.assertEqual(len(attributes), 1) self.assertEqual(attributes[0]["label"], "Date of birth") self.assertEqual(attributes[0]["value"], "27 January 1832") html = u"""
head of government
""" attributes = [] html_etree = fromstring(html) id_cache = wikidata.get_id_cache(html_etree) wikidata.add_attribute(attributes, id_cache, "P6") self.assertEqual(len(attributes), 1) self.assertEqual(attributes[0]["label"], "Head of government") self.assertEqual(attributes[0]["value"], "Old Prime Minister, Actual Prime Minister") attributes = [] html_etree = fromstring(html) id_cache = wikidata.get_id_cache(html_etree) wikidata.add_attribute(attributes, id_cache, "P6", trim=True) self.assertEqual(len(attributes), 1) self.assertEqual(attributes[0]["value"], "Actual Prime Minister") def test_add_url(self): html = u"""
official website
""" urls = [] html_etree = fromstring(html) id_cache = wikidata.get_id_cache(html_etree) wikidata.add_url(urls, html_etree, id_cache, 'P856') self.assertEquals(len(urls), 1) self.assertIn({'title': 'Official website', 'url': 'https://searx.me/'}, urls) urls = [] results = [] wikidata.add_url(urls, html_etree, id_cache, 'P856', 'custom label', results=results) self.assertEquals(len(urls), 1) self.assertEquals(len(results), 1) self.assertIn({'title': 'custom label', 'url': 'https://searx.me/'}, urls) self.assertIn({'title': 'custom label', 'url': 'https://searx.me/'}, results) html = u"""
official website
""" urls = [] html_etree = fromstring(html) id_cache = wikidata.get_id_cache(html_etree) wikidata.add_url(urls, html_etree, id_cache, 'P856') self.assertEquals(len(urls), 2) self.assertIn({'title': 'Official website', 'url': 'http://www.worldofwarcraft.com'}, urls) self.assertIn({'title': 'Official website', 'url': 'http://eu.battle.net/wow/en/'}, urls) def test_get_imdblink(self): html = u"""
tt0433664
""" html_etree = fromstring(html) imdblink = wikidata.get_imdblink(html_etree, 'https://www.imdb.com/') html = u"""
nm4915994
""" html_etree = fromstring(html) imdblink = wikidata.get_imdblink(html_etree, 'https://www.imdb.com/') self.assertIn('https://www.imdb.com/name/nm4915994', imdblink) def test_get_geolink(self): html = u"""
60°N, 40°E
""" html_etree = fromstring(html) geolink = wikidata.get_geolink(html_etree) self.assertIn('https://www.openstreetmap.org/', geolink) self.assertIn('lat=60&lon=40', geolink) html = u"""
34°35'59"S, 58°22'55"W
""" html_etree = fromstring(html) geolink = wikidata.get_geolink(html_etree) self.assertIn('https://www.openstreetmap.org/', geolink) self.assertIn('lat=-34.59', geolink) self.assertIn('lon=-58.38', geolink) def test_get_wikilink(self): html = """
""" html_etree = fromstring(html) wikilink = wikidata.get_wikilink(html_etree, 'nowiki') self.assertEqual(wikilink, None) wikilink = wikidata.get_wikilink(html_etree, 'enwiki') self.assertEqual(wikilink, 'https://en.wikipedia.org/wiki/Test') wikilink = wikidata.get_wikilink(html_etree, 'arwiki') self.assertEqual(wikilink, 'https://ar.wikipedia.org/wiki/Test') wikilink = wikidata.get_wikilink(html_etree, 'enwikiquote') self.assertEqual(wikilink, 'https://en.wikiquote.org/wiki/Test')