logo

searx

My custom branche(s) on searx, a meta-search engine git clone https://hacktivis.me/git/searx.git

test_bing_news.py (6912B)

    

  1. from collections import defaultdict

  2. import mock

  3. from searx.engines import bing_news

  4. from searx.testing import SearxTestCase

  5. import lxml

  6. 

  7. 

  8. class TestBingNewsEngine(SearxTestCase):

  9. 

  10.     def test_request(self):

  11.         bing_news.supported_languages = ['en', 'fr']

  12.         query = 'test_query'

  13.         dicto = defaultdict(dict)

  14.         dicto['pageno'] = 1

  15.         dicto['language'] = 'fr-FR'

  16.         dicto['time_range'] = ''

  17.         params = bing_news.request(query, dicto)

  18.         self.assertIn('url', params)

  19.         self.assertIn(query, params['url'])

  20.         self.assertIn('bing.com', params['url'])

  21.         self.assertIn('fr', params['url'])

  22. 

  23.     def test_no_url_in_request_year_time_range(self):

  24.         dicto = defaultdict(dict)

  25.         query = 'test_query'

  26.         dicto['time_range'] = 'year'

  27.         params = bing_news.request(query, dicto)

  28.         self.assertEqual({}, params['url'])

  29. 

  30.     def test_response(self):

  31.         self.assertRaises(AttributeError, bing_news.response, None)

  32.         self.assertRaises(AttributeError, bing_news.response, [])

  33.         self.assertRaises(AttributeError, bing_news.response, '')

  34.         self.assertRaises(AttributeError, bing_news.response, '[]')

  35. 

  36.         response = mock.Mock(content='<html></html>')

  37.         self.assertEqual(bing_news.response(response), [])

  38. 

  39.         response = mock.Mock(content='<html></html>')

  40.         self.assertEqual(bing_news.response(response), [])

  41. 

  42.         html = """<?xml version="1.0" encoding="utf-8" ?>

  43. <rss version="2.0" xmlns:News="https://www.bing.com:443/news/search?q=python&amp;setmkt=en-US&amp;first=1&amp;format=RSS">

  44.     <channel>

  45.         <title>python - Bing News</title>

  46.         <link>https://www.bing.com:443/news/search?q=python&amp;setmkt=en-US&amp;first=1&amp;format=RSS</link>

  47.         <description>Search results</description>

  48.         <image>

  49.             <url>http://10.53.64.9/rsslogo.gif</url>

  50.             <title>test</title>

  51.             <link>https://www.bing.com:443/news/search?q=test&amp;setmkt=en-US&amp;first=1&amp;format=RSS</link>

  52.         </image>

  53.         <copyright>Copyright</copyright>

  54.         <item>

  55.             <title>Title</title>

  56.             <link>https://www.bing.com/news/apiclick.aspx?ref=FexRss&amp;aid=&amp;tid=c237eccc50bd4758b106a5e3c94fce09&amp;url=http%3a%2f%2furl.of.article%2f&amp;c=xxxxxxxxx&amp;mkt=en-us</link>

  57.             <description>Article Content</description>

  58.             <pubDate>Tue, 02 Jun 2015 13:37:00 GMT</pubDate>

  59.             <News:Source>Infoworld</News:Source>

  60.             <News:Image>http://a1.bing4.com/th?id=ON.13371337133713371337133713371337&amp;pid=News</News:Image>

  61.             <News:ImageSize>w={0}&amp;h={1}&amp;c=7</News:ImageSize>

  62.             <News:ImageKeepOriginalRatio></News:ImageKeepOriginalRatio>

  63.             <News:ImageMaxWidth>620</News:ImageMaxWidth>

  64.             <News:ImageMaxHeight>413</News:ImageMaxHeight>

  65.         </item>

  66.         <item>

  67.             <title>Another Title</title>

  68.             <link>https://www.bing.com/news/apiclick.aspx?ref=FexRss&amp;aid=&amp;tid=c237eccc50bd4758b106a5e3c94fce09&amp;url=http%3a%2f%2fanother.url.of.article%2f&amp;c=xxxxxxxxx&amp;mkt=en-us</link>

  69.             <description>Another Article Content</description>

  70.             <pubDate>Tue, 02 Jun 2015 13:37:00 GMT</pubDate>

  71.         </item>

  72.     </channel>

  73. </rss>"""  # noqa

  74.         response = mock.Mock(content=html.encode('utf-8'))

  75.         results = bing_news.response(response)

  76.         self.assertEqual(type(results), list)

  77.         self.assertEqual(len(results), 2)

  78.         self.assertEqual(results[0]['title'], 'Title')

  79.         self.assertEqual(results[0]['url'], 'http://url.of.article/')

  80.         self.assertEqual(results[0]['content'], 'Article Content')

  81.         self.assertEqual(results[0]['img_src'], 'https://www.bing.com/th?id=ON.13371337133713371337133713371337')

  82.         self.assertEqual(results[1]['title'], 'Another Title')

  83.         self.assertEqual(results[1]['url'], 'http://another.url.of.article/')

  84.         self.assertEqual(results[1]['content'], 'Another Article Content')

  85.         self.assertNotIn('img_src', results[1])

  86. 

  87.         html = """<?xml version="1.0" encoding="utf-8" ?>

  88. <rss version="2.0" xmlns:News="https://www.bing.com:443/news/search?q=python&amp;setmkt=en-US&amp;first=1&amp;format=RSS">

  89.     <channel>

  90.         <title>python - Bing News</title>

  91.         <link>https://www.bing.com:443/news/search?q=python&amp;setmkt=en-US&amp;first=1&amp;format=RSS</link>

  92.         <description>Search results</description>

  93.         <image>

  94.             <url>http://10.53.64.9/rsslogo.gif</url>

  95.             <title>test</title>

  96.             <link>https://www.bing.com:443/news/search?q=test&amp;setmkt=en-US&amp;first=1&amp;format=RSS</link>

  97.         </image>

  98.         <copyright>Copyright</copyright>

  99.         <item>

  100.             <title>Title</title>

  101.             <link>http://another.url.of.article/</link>

  102.             <description>Article Content</description>

  103.             <pubDate>garbage</pubDate>

  104.             <News:Source>Infoworld</News:Source>

  105.             <News:Image>http://another.bing.com/image</News:Image>

  106.             <News:ImageSize>w={0}&amp;h={1}&amp;c=7</News:ImageSize>

  107.             <News:ImageKeepOriginalRatio></News:ImageKeepOriginalRatio>

  108.             <News:ImageMaxWidth>620</News:ImageMaxWidth>

  109.             <News:ImageMaxHeight>413</News:ImageMaxHeight>

  110.         </item>

  111.     </channel>

  112. </rss>"""  # noqa

  113.         response = mock.Mock(content=html.encode('utf-8'))

  114.         results = bing_news.response(response)

  115.         self.assertEqual(type(results), list)

  116.         self.assertEqual(len(results), 1)

  117.         self.assertEqual(results[0]['title'], 'Title')

  118.         self.assertEqual(results[0]['url'], 'http://another.url.of.article/')

  119.         self.assertEqual(results[0]['content'], 'Article Content')

  120.         self.assertEqual(results[0]['img_src'], 'http://another.bing.com/image')

  121. 

  122.         html = """<?xml version="1.0" encoding="utf-8" ?>

  123. <rss version="2.0" xmlns:News="https://www.bing.com:443/news/search?q=python&amp;setmkt=en-US&amp;first=1&amp;format=RSS">

  124.     <channel>

  125.         <title>python - Bing News</title>

  126.         <link>https://www.bing.com:443/news/search?q=python&amp;setmkt=en-US&amp;first=1&amp;format=RSS</link>

  127.         <description>Search results</description>

  128.         <image>

  129.             <url>http://10.53.64.9/rsslogo.gif</url>

  130.             <title>test</title>

  131.             <link>https://www.bing.com:443/news/search?q=test&amp;setmkt=en-US&amp;first=1&amp;format=RSS</link>

  132.         </image>

  133.     </channel>

  134. </rss>"""  # noqa

  135. 

  136.         response = mock.Mock(content=html.encode('utf-8'))

  137.         results = bing_news.response(response)

  138.         self.assertEqual(type(results), list)

  139.         self.assertEqual(len(results), 0)

  140. 

  141.         html = """<?xml version="1.0" encoding="utf-8" ?>gabarge"""

  142.         response = mock.Mock(content=html.encode('utf-8'))

  143.         self.assertRaises(lxml.etree.XMLSyntaxError, bing_news.response, response)