Merge pull request #1 from asciimoo/master

-

tests/unit/engines/test_base.py

@@ -0,0 +1,91 @@
+# -*- coding: utf-8 -*-
+from collections import defaultdict
+import mock
+from searx.engines import base
+from searx.testing import SearxTestCase
+
+
+class TestBaseEngine(SearxTestCase):
+
+    def test_request(self):
+        query = 'test_query'
+        dicto = defaultdict(dict)
+        dicto['pageno'] = 1
+        params = base.request(query, dicto)
+        self.assertIn('url', params)
+        self.assertIn('base-search.net', params['url'])
+
+    def test_response(self):
+        self.assertRaises(AttributeError, base.response, None)
+        self.assertRaises(AttributeError, base.response, [])
+        self.assertRaises(AttributeError, base.response, '')
+        self.assertRaises(AttributeError, base.response, '[]')
+
+        response = mock.Mock(text='<response></response>')
+        self.assertEqual(base.response(response), [])
+
+        xml_mock = """<?xml version="1.0"?>
+<response>
+  <lst name="responseHeader">
+    <int name="status">0</int>
+    <int name="QTime">1</int>
+  </lst>
+  <result name="response" numFound="1" start="0">
+    <doc>
+      <date name="dchdate">2000-01-01T01:01:01Z</date>
+      <str name="dcdocid">1</str>
+      <str name="dccontinent">cna</str>
+      <str name="dccountry">us</str>
+      <str name="dccollection">ftciteseerx</str>
+      <str name="dcprovider">CiteSeerX</str>
+      <str name="dctitle">Science and more</str>
+      <arr name="dccreator">
+        <str>Someone</str>
+      </arr>
+      <arr name="dcperson">
+        <str>Someone</str>
+      </arr>
+      <arr name="dcsubject">
+        <str>Science and more</str>
+      </arr>
+      <str name="dcdescription">Science, and even more.</str>
+      <arr name="dccontributor">
+        <str>The neighbour</str>
+      </arr>
+      <str name="dcdate">2001</str>
+      <int name="dcyear">2001</int>
+      <arr name="dctype">
+        <str>text</str>
+      </arr>
+      <arr name="dctypenorm">
+        <str>1</str>
+      </arr>
+      <arr name="dcformat">
+        <str>application/pdf</str>
+      </arr>
+      <arr name="dccontenttype">
+        <str>application/pdf</str>
+      </arr>
+      <arr name="dcidentifier">
+        <str>http://example.org/</str>
+      </arr>
+      <str name="dclink">http://example.org</str>
+      <str name="dcsource">http://example.org</str>
+      <arr name="dclanguage">
+        <str>en</str>
+      </arr>
+      <str name="dcrights">Under the example.org licence</str>
+      <int name="dcoa">1</int>
+      <arr name="dclang">
+        <str>eng</str>
+      </arr>
+    </doc>
+  </result>
+</response>"""
+
+        response = mock.Mock(text=xml_mock.encode('utf-8'))
+        results = base.response(response)
+        self.assertEqual(type(results), list)
+        self.assertEqual(len(results), 1)
+        self.assertEqual(results[0]['title'], 'Science and more')
+        self.assertEqual(results[0]['content'], 'Science, and even more.')

tests/unit/engines/test_bing_images.py

@@ -8,10 +8,12 @@ from searx.testing import SearxTestCase
 class TestBingImagesEngine(SearxTestCase):
 
     def test_request(self):
+        bing_images.supported_languages = ['fr-FR', 'en-US']
+
         query = 'test_query'
         dicto = defaultdict(dict)
         dicto['pageno'] = 1
-        dicto['language'] = 'fr_FR'
+        dicto['language'] = 'fr-FR'
         dicto['safesearch'] = 1
         dicto['time_range'] = ''
         params = bing_images.request(query, dicto)
@@ -19,12 +21,19 @@ class TestBingImagesEngine(SearxTestCase):
         self.assertTrue(query in params['url'])
         self.assertTrue('bing.com' in params['url'])
         self.assertTrue('SRCHHPGUSR' in params['cookies'])
-        self.assertTrue('fr' in params['cookies']['SRCHHPGUSR'])
+        self.assertTrue('DEMOTE' in params['cookies']['SRCHHPGUSR'])
+        self.assertTrue('_EDGE_S' in params['cookies'])
+        self.assertTrue('fr-fr' in params['cookies']['_EDGE_S'])
+
+        dicto['language'] = 'fr'
+        params = bing_images.request(query, dicto)
+        self.assertTrue('_EDGE_S' in params['cookies'])
+        self.assertTrue('fr-fr' in params['cookies']['_EDGE_S'])
 
         dicto['language'] = 'all'
         params = bing_images.request(query, dicto)
-        self.assertIn('SRCHHPGUSR', params['cookies'])
-        self.assertIn('en', params['cookies']['SRCHHPGUSR'])
+        self.assertTrue('_EDGE_S' in params['cookies'])
+        self.assertTrue('en-us' in params['cookies']['_EDGE_S'])
 
     def test_response(self):
         self.assertRaises(AttributeError, bing_images.response, None)
@@ -82,3 +91,28 @@ class TestBingImagesEngine(SearxTestCase):
         self.assertEqual(results[0]['content'], '')
         self.assertEqual(results[0]['thumbnail_src'], 'thumb_url')
         self.assertEqual(results[0]['img_src'], 'img_url')
+
+    def test_fetch_supported_languages(self):
+        html = """
+        <div>
+            <div id="region-section-content">
+                <ul class="b_vList">
+                    <li>
+                        <a href="https://bing...&setmkt=de-DE&s...">Germany</a>
+                        <a href="https://bing...&setmkt=nb-NO&s...">Norway</a>
+                    </li>
+                </ul>
+                <ul class="b_vList">
+                    <li>
+                        <a href="https://bing...&setmkt=es-AR&s...">Argentina</a>
+                    </li>
+                </ul>
+            </div>
+        </div>
+        """
+        response = mock.Mock(text=html)
+        languages = list(bing_images._fetch_supported_languages(response))
+        self.assertEqual(len(languages), 3)
+        self.assertIn('de-DE', languages)
+        self.assertIn('no-NO', languages)
+        self.assertIn('es-AR', languages)

tests/unit/engines/test_bing_videos.py

@@ -8,6 +8,8 @@ from searx.testing import SearxTestCase
 class TestBingVideosEngine(SearxTestCase):
 
     def test_request(self):
+        bing_videos.supported_languages = ['fr-FR', 'en-US']
+
         query = 'test_query'
         dicto = defaultdict(dict)
         dicto['pageno'] = 1

tests/unit/engines/test_blekko_images.py

@@ -1,71 +0,0 @@
-from collections import defaultdict
-import mock
-from searx.engines import blekko_images
-from searx.testing import SearxTestCase
-
-
-class TestBlekkoImagesEngine(SearxTestCase):
-
-    def test_request(self):
-        query = 'test_query'
-        dicto = defaultdict(dict)
-        dicto['pageno'] = 0
-        dicto['safesearch'] = 1
-        params = blekko_images.request(query, dicto)
-        self.assertIn('url', params)
-        self.assertIn(query, params['url'])
-        self.assertIn('blekko.com', params['url'])
-        self.assertIn('page', params['url'])
-
-        dicto['pageno'] = 1
-        params = blekko_images.request(query, dicto)
-        self.assertNotIn('page', params['url'])
-
-    def test_response(self):
-        self.assertRaises(AttributeError, blekko_images.response, None)
-        self.assertRaises(AttributeError, blekko_images.response, [])
-        self.assertRaises(AttributeError, blekko_images.response, '')
-        self.assertRaises(AttributeError, blekko_images.response, '[]')
-
-        response = mock.Mock(text='[]')
-        self.assertEqual(blekko_images.response(response), [])
-
-        json = """
-        [
-            {
-                "c": 1,
-                "page_url": "http://result_url.html",
-                "title": "Photo title",
-                "tn_url": "http://ts1.mm.bing.net/th?id=HN.608050619474382748&pid=15.1",
-                "url": "http://result_image.jpg"
-            },
-            {
-                "c": 2,
-                "page_url": "http://companyorange.simpsite.nl/OSM",
-                "title": "OSM",
-                "tn_url": "http://ts2.mm.bing.net/th?id=HN.608048068264919461&pid=15.1",
-                "url": "http://simpsite.nl/userdata2/58985/Home/OSM.bmp"
-            },
-            {
-                "c": 3,
-                "page_url": "http://invincible.webklik.nl/page/osm",
-                "title": "OSM",
-                "tn_url": "http://ts1.mm.bing.net/th?id=HN.608024514657649476&pid=15.1",
-                "url": "http://www.webklik.nl/user_files/2009_09/65324/osm.gif"
-            },
-            {
-                "c": 4,
-                "page_url": "http://www.offshorenorway.no/event/companyDetail/id/12492",
-                "title": "Go to OSM Offshore AS homepage",
-                "tn_url": "http://ts2.mm.bing.net/th?id=HN.608054265899847285&pid=15.1",
-                "url": "http://www.offshorenorway.no/firmalogo/OSM-logo.png"
-            }
-        ]
-        """
-        response = mock.Mock(text=json)
-        results = blekko_images.response(response)
-        self.assertEqual(type(results), list)
-        self.assertEqual(len(results), 4)
-        self.assertEqual(results[0]['title'], 'Photo title')
-        self.assertEqual(results[0]['url'], 'http://result_url.html')
-        self.assertEqual(results[0]['img_src'], 'http://result_image.jpg')

tests/unit/engines/test_faroo.py

@@ -40,9 +40,6 @@ class TestFarooEngine(SearxTestCase):
         response = mock.Mock(text='{"data": []}')
         self.assertEqual(faroo.response(response), [])
 
-        response = mock.Mock(text='{"data": []}', status_code=401)
-        self.assertRaises(Exception, faroo.response, response)
-
         response = mock.Mock(text='{"data": []}', status_code=429)
         self.assertRaises(Exception, faroo.response, response)
 
@@ -98,14 +95,14 @@ class TestFarooEngine(SearxTestCase):
         response = mock.Mock(text=json)
         results = faroo.response(response)
         self.assertEqual(type(results), list)
-        self.assertEqual(len(results), 4)
+        self.assertEqual(len(results), 3)
         self.assertEqual(results[0]['title'], 'This is the title')
         self.assertEqual(results[0]['url'], 'http://this.is.the.url/')
         self.assertEqual(results[0]['content'], 'This is the content')
         self.assertEqual(results[1]['title'], 'This is the title2')
         self.assertEqual(results[1]['url'], 'http://this.is.the.url2/')
         self.assertEqual(results[1]['content'], 'This is the content2')
-        self.assertEqual(results[3]['img_src'], 'http://upload.wikimedia.org/optimized.jpg')
+        self.assertEqual(results[2]['thumbnail'], 'http://upload.wikimedia.org/optimized.jpg')
 
         json = """
         {}

tests/unit/engines/test_nyaa.py

@@ -13,38 +13,92 @@ class TestNyaaEngine(SearxTestCase):
         params = nyaa.request(query, dic)
         self.assertTrue('url' in params)
         self.assertTrue(query in params['url'])
-        self.assertTrue('nyaa.se' in params['url'])
+        self.assertTrue('nyaa.si' in params['url'])
 
     def test_response(self):
         resp = mock.Mock(text='<html></html>')
         self.assertEqual(nyaa.response(resp), [])
 
         html = """
-        <table class="tlist">
-          <tbody>
-            <tr class="trusted tlistrow">
-              <td class="tlisticon">
-                <a href="//www.nyaa.se" title="English-translated Anime">
-                   <img src="//files.nyaa.se" alt="English-translated Anime">
-                </a>
-              </td>
-              <td class="tlistname">
-                <a href="//www.nyaa.se/?page3">
-                  Sample torrent title
-                </a>
-              </td>
-              <td class="tlistdownload">
-                <a href="//www.nyaa.se/?page_dl" title="Download">
-                  <img src="//files.nyaa.se/www-dl.png" alt="DL">
-                </a>
-              </td>
-              <td class="tlistsize">10 MiB</td>
-              <td class="tlistsn">1</td>
-              <td class="tlistln">3</td>
-              <td class="tlistdn">666</td>
-              <td class="tlistmn">0</td>
-            </tr>
-          </tbody>
+        <table class="table table-bordered table-hover table-striped torrent-list">
+        <thead>
+        <tr>
+        <th class="hdr-category text-center" style="width:80px;">
+        <div>Category</div>
+        </th>
+        <th class="hdr-name" style="width:auto;">
+        <div>Name</div>
+        </th>
+        <th class="hdr-comments sorting text-center" title="Comments" style="width:50px;">
+        <a href="/?f=0&amp;c=0_0&amp;q=Death+Parade&amp;s=comments&amp;o=desc"></a>
+        <i class="fa fa-comments-o"></i>
+        </th>
+        <th class="hdr-link text-center" style="width:70px;">
+        <div>Link</div>
+        </th>
+        <th class="hdr-size sorting text-center" style="width:100px;">
+        <a href="/?f=0&amp;c=0_0&amp;q=Death+Parade&amp;s=size&amp;o=desc"></a>
+        <div>Size</div>
+        </th>
+        <th class="hdr-date sorting_desc text-center" title="In local time" style="width:140px;">
+        <a href="/?f=0&amp;c=0_0&amp;q=Death+Parade&amp;s=id&amp;o=asc"></a>
+        <div>Date</div>
+        </th>
+        <th class="hdr-seeders sorting text-center" title="Seeders" style="width:50px;">
+        <a href="/?f=0&amp;c=0_0&amp;q=Death+Parade&amp;s=seeders&amp;o=desc"></a>
+        <i class="fa fa-arrow-up" aria-hidden="true"></i>
+        </th>
+        <th class="hdr-leechers sorting text-center" title="Leechers" style="width:50px;">
+        <a href="/?f=0&amp;c=0_0&amp;q=Death+Parade&amp;s=leechers&amp;o=desc"></a>
+        <i class="fa fa-arrow-down" aria-hidden="true"></i>
+        </th>
+        <th class="hdr-downloads sorting text-center" title="Completed downloads" style="width:50px;">
+        <a href="/?f=0&amp;c=0_0&amp;q=Death+Parade&amp;s=downloads&amp;o=desc"></a>
+        <i class="fa fa-check" aria-hidden="true"></i>
+        </th>
+        </tr>
+        </thead>
+        <tbody>
+        <tr class="default">
+        <td style="padding:0 4px;">
+        <a href="/?c=1_2" title="Anime - English-translated">
+        <img src="/static/img/icons/nyaa/1_2.png" alt="Anime - English-translated">
+        </a>
+        </td>
+        <td colspan="2">
+        <a href="/view/1" title="Sample title 1">Sample title 1</a>
+        </td>
+        <td class="text-center" style="white-space: nowrap;">
+        <a href="/download/1.torrent"><i class="fa fa-fw fa-download"></i></a>
+        <a href="magnet:?xt=urn:btih:2"><i class="fa fa-fw fa-magnet"></i></a>
+        </td>
+        <td class="text-center">723.7 MiB</td>
+        <td class="text-center" data-timestamp="1503307456" title="1 week 3
+        days 9 hours 44 minutes 39 seconds ago">2017-08-21 11:24</td>
+        <td class="text-center" style="color: green;">1</td>
+        <td class="text-center" style="color: red;">3</td>
+        <td class="text-center">12</td>
+        </tr>
+        <tr class="default">
+        <td style="padding:0 4px;">
+        <a href="/?c=1_2" title="Anime - English-translated">
+        <img src="/static/img/icons/nyaa/1_2.png" alt="Anime - English-translated">
+        </a>
+        </td>
+        <td colspan="2">
+        <a href="/view/2" title="Sample title 2">Sample title 2</a>
+        </td>
+        <td class="text-center" style="white-space: nowrap;">
+        <a href="magnet:?xt=urn:btih:2"><i class="fa fa-fw fa-magnet"></i></a>
+        </td>
+        <td class="text-center">8.2 GiB</td>
+        <td class="text-center" data-timestamp="1491608400" title="4 months 3
+        weeks 4 days 19 hours 28 minutes 55 seconds ago">2017-04-08 01:40</td>
+        <td class="text-center" style="color: green;">10</td>
+        <td class="text-center" style="color: red;">1</td>
+        <td class="text-center">206</td>
+        </tr>
+        </tbody>
         </table>
         """
 
@@ -52,15 +106,19 @@ class TestNyaaEngine(SearxTestCase):
         results = nyaa.response(resp)
 
         self.assertEqual(type(results), list)
-        self.assertEqual(len(results), 1)
+        self.assertEqual(len(results), 2)
 
         r = results[0]
-        self.assertTrue(r['url'].find('www.nyaa.se/?page3') >= 0)
-        self.assertTrue(r['torrentfile'].find('www.nyaa.se/?page_dl') >= 0)
-        self.assertTrue(r['content'].find('English-translated Anime') >= 0)
-        self.assertTrue(r['content'].find('Downloaded 666 times.') >= 0)
+        self.assertTrue(r['url'].find('1') >= 0)
+        self.assertTrue(r['torrentfile'].find('1.torrent') >= 0)
+        self.assertTrue(r['content'].find('Anime - English-translated') >= 0)
+        self.assertTrue(r['content'].find('Downloaded 12 times.') >= 0)
 
-        self.assertEqual(r['title'], 'Sample torrent title')
+        self.assertEqual(r['title'], 'Sample title 1')
         self.assertEqual(r['seed'], 1)
         self.assertEqual(r['leech'], 3)
-        self.assertEqual(r['filesize'], 10 * 1024 * 1024)
+        self.assertEqual(r['filesize'], 723700000)
+
+        r = results[1]
+        self.assertTrue(r['url'].find('2') >= 0)
+        self.assertTrue(r['magnetlink'].find('magnet:') >= 0)

tests/unit/engines/test_swisscows.py

@@ -139,9 +139,9 @@ class TestSwisscowsEngine(SearxTestCase):
             <div id="regions-popup">
                 <div>
                     <ul>
-                        <li><a data-val="browser"></a></li>
-                        <li><a data-val="de-CH"></a></li>
-                        <li><a data-val="fr-CH"></a></li>
+                        <li><a data-search-language="browser"></a></li>
+                        <li><a data-search-language="de-CH"></a></li>
+                        <li><a data-search-language="fr-CH"></a></li>
                     </ul>
                 </div>
             </div>

tests/unit/engines/test_torrentz.py

@@ -14,7 +14,7 @@ class TestTorrentzEngine(SearxTestCase):
         params = torrentz.request(query, dic)
         self.assertTrue('url' in params)
         self.assertTrue(query in params['url'])
-        self.assertTrue('torrentz.eu' in params['url'])
+        self.assertTrue('torrentz2.eu' in params['url'])
 
     def test_response(self):
         resp = mock.Mock(text='<html></html>')
@@ -30,13 +30,11 @@ class TestTorrentzEngine(SearxTestCase):
               books ebooks
             </dt>
             <dd>
-              <span class="v">1</span>
-              <span class="a">
-                <span title="Sun, 22 Nov 2015 03:01:42">4 months</span>
-              </span>
-              <span class="s">30 MB</span>
-              <span class="u">14</span>
-              <span class="d">1</span>
+              <span>1</span>
+              <span title="1503595924">5 hours</span>
+              <span>30 MB</span>
+              <span>14</span>
+              <span>1</span>
             </dd>
           </dl>
 
@@ -48,13 +46,11 @@ class TestTorrentzEngine(SearxTestCase):
               books ebooks
             </dt>
             <dd>
-              <span class="v">1</span>
-              <span class="a">
-                <span title="Sun, 2124091j0j190gm42">4 months</span>
-              </span>
-              <span class="s">30MB</span>
-              <span class="u">5,555</span>
-              <span class="d">1,234,567</span>
+              <span>1</span>
+              <span title="1503595924 aaa">5 hours</span>
+              <span>30MB</span>
+              <span>5,555</span>
+              <span>1,234,567</span>
             </dd>
           </dl>
         </div>
@@ -68,10 +64,10 @@ class TestTorrentzEngine(SearxTestCase):
 
         # testing against the first result
         r = results[0]
-        self.assertEqual(r['url'], 'https://torrentz.eu/4362e08b1d80e1820fb2550b752f9f3126fe76d6')
+        self.assertEqual(r['url'], 'https://torrentz2.eu/4362e08b1d80e1820fb2550b752f9f3126fe76d6')
         self.assertEqual(r['title'], 'Completely valid info books ebooks')
         # 22 Nov 2015 03:01:42
-        self.assertEqual(r['publishedDate'], datetime(2015, 11, 22, 3, 1, 42))
+        self.assertEqual(r['publishedDate'], datetime.fromtimestamp(1503595924))
         self.assertEqual(r['seed'], 14)
         self.assertEqual(r['leech'], 1)
         self.assertEqual(r['filesize'], 30 * 1024 * 1024)
@@ -79,7 +75,7 @@ class TestTorrentzEngine(SearxTestCase):
 
         # testing against the second result
         r = results[1]
-        self.assertEqual(r['url'], 'https://torrentz.eu/poaskdpokaspod')
+        self.assertEqual(r['url'], 'https://torrentz2.eu/poaskdpokaspod')
         self.assertEqual(r['title'], 'Invalid hash and date and filesize books ebooks')
         self.assertEqual(r['seed'], 5555)
         self.assertEqual(r['leech'], 1234567)