from collections import defaultdict import mock from searx.engines import bing_news from searx.testing import SearxTestCase import lxml class TestBingNewsEngine(SearxTestCase): def test_request(self): query = 'test_query' dicto = defaultdict(dict) dicto['pageno'] = 1 dicto['language'] = 'fr_FR' dicto['time_range'] = '' params = bing_news.request(query, dicto) self.assertIn('url', params) self.assertIn(query, params['url']) self.assertIn('bing.com', params['url']) self.assertIn('fr', params['url']) dicto['language'] = 'all' params = bing_news.request(query, dicto) self.assertIn('en', params['url']) def test_no_url_in_request_year_time_range(self): dicto = defaultdict(dict) query = 'test_query' dicto['time_range'] = 'year' params = bing_news.request(query, dicto) self.assertEqual({}, params['url']) def test_response(self): self.assertRaises(AttributeError, bing_news.response, None) self.assertRaises(AttributeError, bing_news.response, []) self.assertRaises(AttributeError, bing_news.response, '') self.assertRaises(AttributeError, bing_news.response, '[]') response = mock.Mock(text='') self.assertEqual(bing_news.response(response), []) response = mock.Mock(text='') self.assertEqual(bing_news.response(response), []) html = """ python - Bing News https://www.bing.com:443/news/search?q=python&setmkt=en-US&first=1&format=RSS Search results http://10.53.64.9/rsslogo.gif test https://www.bing.com:443/news/search?q=test&setmkt=en-US&first=1&format=RSS Copyright Title https://www.bing.com/news/apiclick.aspx?ref=FexRss&aid=&tid=c237eccc50bd4758b106a5e3c94fce09&url=http%3a%2f%2furl.of.article%2f&c=xxxxxxxxx&mkt=en-us Article Content Tue, 02 Jun 2015 13:37:00 GMT Infoworld http://a1.bing4.com/th?id=ON.13371337133713371337133713371337&pid=News w={0}&h={1}&c=7 620 413 Another Title https://www.bing.com/news/apiclick.aspx?ref=FexRss&aid=&tid=c237eccc50bd4758b106a5e3c94fce09&url=http%3a%2f%2fanother.url.of.article%2f&c=xxxxxxxxx&mkt=en-us Another Article Content Tue, 02 Jun 2015 13:37:00 GMT """ # noqa response = mock.Mock(text=html.encode('utf-8')) results = bing_news.response(response) self.assertEqual(type(results), list) self.assertEqual(len(results), 2) self.assertEqual(results[0]['title'], 'Title') self.assertEqual(results[0]['url'], 'http://url.of.article/') self.assertEqual(results[0]['content'], 'Article Content') self.assertEqual(results[0]['img_src'], 'https://www.bing.com/th?id=ON.13371337133713371337133713371337') self.assertEqual(results[1]['title'], 'Another Title') self.assertEqual(results[1]['url'], 'http://another.url.of.article/') self.assertEqual(results[1]['content'], 'Another Article Content') self.assertNotIn('img_src', results[1]) html = """ python - Bing News https://www.bing.com:443/news/search?q=python&setmkt=en-US&first=1&format=RSS Search results http://10.53.64.9/rsslogo.gif test https://www.bing.com:443/news/search?q=test&setmkt=en-US&first=1&format=RSS Copyright Title http://another.url.of.article/ Article Content garbage Infoworld http://another.bing.com/image w={0}&h={1}&c=7 620 413 """ # noqa response = mock.Mock(text=html.encode('utf-8')) results = bing_news.response(response) self.assertEqual(type(results), list) self.assertEqual(len(results), 1) self.assertEqual(results[0]['title'], 'Title') self.assertEqual(results[0]['url'], 'http://another.url.of.article/') self.assertEqual(results[0]['content'], 'Article Content') self.assertEqual(results[0]['img_src'], 'http://another.bing.com/image') html = """ python - Bing News https://www.bing.com:443/news/search?q=python&setmkt=en-US&first=1&format=RSS Search results http://10.53.64.9/rsslogo.gif test https://www.bing.com:443/news/search?q=test&setmkt=en-US&first=1&format=RSS """ # noqa response = mock.Mock(text=html.encode('utf-8')) results = bing_news.response(response) self.assertEqual(type(results), list) self.assertEqual(len(results), 0) html = """gabarge""" response = mock.Mock(text=html.encode('utf-8')) self.assertRaises(lxml.etree.XMLSyntaxError, bing_news.response, response)