2015-01-29 20:56:57 +01:00
|
|
|
from collections import defaultdict
|
|
|
|
|
import mock
|
|
|
|
|
from searx.engines import bing_news
|
|
|
|
|
from searx.testing import SearxTestCase
|
2015-06-04 18:30:08 +02:00
|
|
|
import lxml
|
2015-01-29 20:56:57 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
class TestBingNewsEngine(SearxTestCase):
|
|
|
|
|
|
|
|
|
|
def test_request(self):
|
|
|
|
|
query = 'test_query'
|
|
|
|
|
dicto = defaultdict(dict)
|
|
|
|
|
dicto['pageno'] = 1
|
|
|
|
|
dicto['language'] = 'fr_FR'
|
|
|
|
|
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_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, '[]')
|
|
|
|
|
|
2015-10-16 12:01:35 +02:00
|
|
|
response = mock.Mock(content='<html></html>')
|
2015-01-29 20:56:57 +01:00
|
|
|
self.assertEqual(bing_news.response(response), [])
|
|
|
|
|
|
2015-10-16 12:01:35 +02:00
|
|
|
response = mock.Mock(content='<html></html>')
|
2015-01-29 20:56:57 +01:00
|
|
|
self.assertEqual(bing_news.response(response), [])
|
|
|
|
|
|
2015-06-04 18:30:08 +02:00
|
|
|
html = """<?xml version="1.0" encoding="utf-8" ?>
|
|
|
|
|
<rss version="2.0" xmlns:News="https://www.bing.com:443/news/search?q=python&setmkt=en-US&first=1&format=RSS">
|
|
|
|
|
<channel>
|
|
|
|
|
<title>python - Bing News</title>
|
|
|
|
|
<link>https://www.bing.com:443/news/search?q=python&setmkt=en-US&first=1&format=RSS</link>
|
|
|
|
|
<description>Search results</description>
|
|
|
|
|
<image>
|
|
|
|
|
<url>http://10.53.64.9/rsslogo.gif</url>
|
|
|
|
|
<title>test</title>
|
|
|
|
|
<link>https://www.bing.com:443/news/search?q=test&setmkt=en-US&first=1&format=RSS</link>
|
|
|
|
|
</image>
|
|
|
|
|
<copyright>Copyright</copyright>
|
|
|
|
|
<item>
|
|
|
|
|
<title>Title</title>
|
|
|
|
|
<link>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</link>
|
|
|
|
|
<description>Article Content</description>
|
|
|
|
|
<pubDate>Tue, 02 Jun 2015 13:37:00 GMT</pubDate>
|
|
|
|
|
<News:Source>Infoworld</News:Source>
|
|
|
|
|
<News:Image>http://a1.bing4.com/th?id=ON.13371337133713371337133713371337&pid=News</News:Image>
|
|
|
|
|
<News:ImageSize>w={0}&h={1}&c=7</News:ImageSize>
|
|
|
|
|
<News:ImageKeepOriginalRatio></News:ImageKeepOriginalRatio>
|
|
|
|
|
<News:ImageMaxWidth>620</News:ImageMaxWidth>
|
|
|
|
|
<News:ImageMaxHeight>413</News:ImageMaxHeight>
|
|
|
|
|
</item>
|
|
|
|
|
<item>
|
|
|
|
|
<title>Another Title</title>
|
|
|
|
|
<link>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</link>
|
|
|
|
|
<description>Another Article Content</description>
|
|
|
|
|
<pubDate>Tue, 02 Jun 2015 13:37:00 GMT</pubDate>
|
|
|
|
|
</item>
|
|
|
|
|
</channel>
|
|
|
|
|
</rss>""" # noqa
|
2015-10-16 12:01:35 +02:00
|
|
|
response = mock.Mock(content=html)
|
2015-01-29 20:56:57 +01:00
|
|
|
results = bing_news.response(response)
|
|
|
|
|
self.assertEqual(type(results), list)
|
2015-06-04 18:30:08 +02:00
|
|
|
self.assertEqual(len(results), 2)
|
2015-01-29 20:56:57 +01:00
|
|
|
self.assertEqual(results[0]['title'], 'Title')
|
|
|
|
|
self.assertEqual(results[0]['url'], 'http://url.of.article/')
|
|
|
|
|
self.assertEqual(results[0]['content'], 'Article Content')
|
2015-06-04 18:30:08 +02:00
|
|
|
self.assertEqual(results[0]['thumbnail'], '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('thumbnail', results[1])
|
2015-01-29 20:56:57 +01:00
|
|
|
|
2015-06-04 18:30:08 +02:00
|
|
|
html = """<?xml version="1.0" encoding="utf-8" ?>
|
|
|
|
|
<rss version="2.0" xmlns:News="https://www.bing.com:443/news/search?q=python&setmkt=en-US&first=1&format=RSS">
|
|
|
|
|
<channel>
|
|
|
|
|
<title>python - Bing News</title>
|
|
|
|
|
<link>https://www.bing.com:443/news/search?q=python&setmkt=en-US&first=1&format=RSS</link>
|
|
|
|
|
<description>Search results</description>
|
|
|
|
|
<image>
|
|
|
|
|
<url>http://10.53.64.9/rsslogo.gif</url>
|
|
|
|
|
<title>test</title>
|
|
|
|
|
<link>https://www.bing.com:443/news/search?q=test&setmkt=en-US&first=1&format=RSS</link>
|
|
|
|
|
</image>
|
|
|
|
|
<copyright>Copyright</copyright>
|
|
|
|
|
<item>
|
|
|
|
|
<title>Title</title>
|
|
|
|
|
<link>http://another.url.of.article/</link>
|
|
|
|
|
<description>Article Content</description>
|
|
|
|
|
<pubDate>garbage</pubDate>
|
|
|
|
|
<News:Source>Infoworld</News:Source>
|
|
|
|
|
<News:Image>http://another.bing.com/image</News:Image>
|
|
|
|
|
<News:ImageSize>w={0}&h={1}&c=7</News:ImageSize>
|
|
|
|
|
<News:ImageKeepOriginalRatio></News:ImageKeepOriginalRatio>
|
|
|
|
|
<News:ImageMaxWidth>620</News:ImageMaxWidth>
|
|
|
|
|
<News:ImageMaxHeight>413</News:ImageMaxHeight>
|
|
|
|
|
</item>
|
|
|
|
|
</channel>
|
|
|
|
|
</rss>""" # noqa
|
2015-10-16 12:01:35 +02:00
|
|
|
response = mock.Mock(content=html)
|
2015-01-29 20:56:57 +01:00
|
|
|
results = bing_news.response(response)
|
|
|
|
|
self.assertEqual(type(results), list)
|
2015-06-04 18:30:08 +02:00
|
|
|
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]['thumbnail'], 'http://another.bing.com/image')
|
|
|
|
|
|
|
|
|
|
html = """<?xml version="1.0" encoding="utf-8" ?>
|
|
|
|
|
<rss version="2.0" xmlns:News="https://www.bing.com:443/news/search?q=python&setmkt=en-US&first=1&format=RSS">
|
|
|
|
|
<channel>
|
|
|
|
|
<title>python - Bing News</title>
|
|
|
|
|
<link>https://www.bing.com:443/news/search?q=python&setmkt=en-US&first=1&format=RSS</link>
|
|
|
|
|
<description>Search results</description>
|
|
|
|
|
<image>
|
|
|
|
|
<url>http://10.53.64.9/rsslogo.gif</url>
|
|
|
|
|
<title>test</title>
|
|
|
|
|
<link>https://www.bing.com:443/news/search?q=test&setmkt=en-US&first=1&format=RSS</link>
|
|
|
|
|
</image>
|
|
|
|
|
</channel>
|
|
|
|
|
</rss>""" # noqa
|
2015-01-29 20:56:57 +01:00
|
|
|
|
2015-10-16 12:01:35 +02:00
|
|
|
response = mock.Mock(content=html)
|
2015-01-29 20:56:57 +01:00
|
|
|
results = bing_news.response(response)
|
|
|
|
|
self.assertEqual(type(results), list)
|
|
|
|
|
self.assertEqual(len(results), 0)
|
2015-06-04 18:30:08 +02:00
|
|
|
|
|
|
|
|
html = """<?xml version="1.0" encoding="utf-8" ?>gabarge"""
|
2015-10-16 12:01:35 +02:00
|
|
|
response = mock.Mock(content=html)
|
2015-06-04 18:30:08 +02:00
|
|
|
self.assertRaises(lxml.etree.XMLSyntaxError, bing_news.response, response)
|
