diff --git a/searx/engines/kickass.py b/searx/engines/kickass.py index ac349283d..8b89e1f47 100644 --- a/searx/engines/kickass.py +++ b/searx/engines/kickass.py @@ -13,6 +13,7 @@ from cgi import escape from urllib import quote from lxml import html from operator import itemgetter +from searx.engines.xpath import extract_text # engine dependent config categories = ['videos', 'music', 'files'] @@ -56,9 +57,8 @@ def response(resp): for result in search_res[1:]: link = result.xpath('.//a[@class="cellMainLink"]')[0] href = urljoin(url, link.attrib['href']) - title = ' '.join(link.xpath('.//text()')) - content = escape(html.tostring(result.xpath(content_xpath)[0], - method="text")) + title = extract_text(link) + content = escape(extract_text(result.xpath(content_xpath))) seed = result.xpath('.//td[contains(@class, "green")]/text()')[0] leech = result.xpath('.//td[contains(@class, "red")]/text()')[0] filesize = result.xpath('.//td[contains(@class, "nobr")]/text()')[0] @@ -88,7 +88,7 @@ def response(resp): filesize = int(filesize * 1024 * 1024 * 1024) elif filesize_multiplier == 'MB': filesize = int(filesize * 1024 * 1024) - elif filesize_multiplier == 'kb': + elif filesize_multiplier == 'KB': filesize = int(filesize * 1024) except: filesize = None diff --git a/searx/tests/engines/test_kickass.py b/searx/tests/engines/test_kickass.py new file mode 100644 index 000000000..3c20a97e7 --- /dev/null +++ b/searx/tests/engines/test_kickass.py @@ -0,0 +1,398 @@ +# -*- coding: utf-8 -*- +from collections import defaultdict +import mock +from searx.engines import kickass +from searx.testing import SearxTestCase + + +class TestKickassEngine(SearxTestCase): + + def test_request(self): + query = 'test_query' + dicto = defaultdict(dict) + dicto['pageno'] = 1 + params = kickass.request(query, dicto) + self.assertIn('url', params) + self.assertIn(query, params['url']) + self.assertIn('kickass.so', params['url']) + self.assertIn('verify', params) + self.assertFalse(params['verify']) + + def test_response(self): + self.assertRaises(AttributeError, kickass.response, None) + self.assertRaises(AttributeError, kickass.response, []) + self.assertRaises(AttributeError, kickass.response, '') + self.assertRaises(AttributeError, kickass.response, '[]') + + response = mock.Mock(text='') + self.assertEqual(kickass.response(response), []) + + html = """ + + + + + + + + + + + + + + + + + +
torrent name + size + + files + + age + + seed + + leech +
+
+ + 3 + + + + + + + + + + + + + + +
+
+ + +
+ + This should be the title + + + Posted by + riri in + + Other > Unsorted + + +
+
449 bytes42 years101
+ """ + response = mock.Mock(text=html) + results = kickass.response(response) + self.assertEqual(type(results), list) + self.assertEqual(len(results), 1) + self.assertEqual(results[0]['title'], 'This should be the title') + self.assertEqual(results[0]['url'], 'https://kickass.so/url.html') + self.assertEqual(results[0]['content'], 'Posted by riri in Other > Unsorted') + self.assertEqual(results[0]['seed'], 10) + self.assertEqual(results[0]['leech'], 1) + self.assertEqual(results[0]['filesize'], 449) + self.assertEqual(results[0]['files'], 4) + self.assertEqual(results[0]['magnetlink'], 'magnet:?xt=urn:btih:MAGNETURL&dn=test') + self.assertEqual(results[0]['torrentfile'], 'http://torcache.net/torrent/53917.torrent?title=test') + + html = """ + + + + + + + + + +
torrent name + size + + files + + age + + seed + + leech +
+ """ + response = mock.Mock(text=html) + results = kickass.response(response) + self.assertEqual(type(results), list) + self.assertEqual(len(results), 0) + + html = """ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
torrent name + size + + files + + age + + seed + + leech +
+
+ + 3 + + + + + + + + + + + + + + +
+
+ + +
+ + This should be the title + + + Posted by + riri in + + Other > Unsorted + + +
+
1 KB42 years101
+
+ + 3 + + + + + + + + + + + + + + +
+
+ + +
+ + This should be the title + + + Posted by + riri in + + Other > Unsorted + + +
+
1 MB42 years91
+
+ + 3 + + + + + + + + + + + + + + +
+
+ + +
+ + This should be the title + + + Posted by + riri in + + Other > Unsorted + + +
+
1 GB42 years81
+
+ + 3 + + + + + + + + + + + + + + +
+
+ + +
+ + This should be the title + + + Posted by + riri in + + Other > Unsorted + + +
+
1 TB42 years71
+
+ + 3 + + + + + + + + + + + + + + +
+
+ + +
+ + This should be the title + + + Posted by + riri in + + Other > Unsorted + + +
+
z bytesr2 yearsat
+ """ + response = mock.Mock(text=html) + results = kickass.response(response) + self.assertEqual(type(results), list) + self.assertEqual(len(results), 5) + self.assertEqual(results[0]['title'], 'This should be the title') + self.assertEqual(results[0]['url'], 'https://kickass.so/url.html') + self.assertEqual(results[0]['content'], 'Posted by riri in Other > Unsorted') + self.assertEqual(results[0]['seed'], 10) + self.assertEqual(results[0]['leech'], 1) + self.assertEqual(results[0]['files'], 4) + self.assertEqual(results[0]['magnetlink'], 'magnet:?xt=urn:btih:MAGNETURL&dn=test') + self.assertEqual(results[0]['torrentfile'], 'http://torcache.net/torrent/53917.torrent?title=test') + self.assertEqual(results[0]['filesize'], 1024) + self.assertEqual(results[1]['filesize'], 1048576) + self.assertEqual(results[2]['filesize'], 1073741824) + self.assertEqual(results[3]['filesize'], 1099511627776) + self.assertEqual(results[4]['seed'], 0) + self.assertEqual(results[4]['leech'], 0) + self.assertEqual(results[4]['files'], None) + self.assertEqual(results[4]['filesize'], None) diff --git a/searx/tests/test_engines.py b/searx/tests/test_engines.py index b07444e42..66f8fbff7 100644 --- a/searx/tests/test_engines.py +++ b/searx/tests/test_engines.py @@ -9,6 +9,7 @@ from searx.tests.engines.test_digg import * # noqa from searx.tests.engines.test_dummy import * # noqa from searx.tests.engines.test_flickr import * # noqa from searx.tests.engines.test_github import * # noqa +from searx.tests.engines.test_kickass import * # noqa from searx.tests.engines.test_mixcloud import * # noqa from searx.tests.engines.test_searchcode_code import * # noqa from searx.tests.engines.test_searchcode_doc import * # noqa