Merge remote-tracking branch 'origin/master'

This commit is contained in:
Kang-min Liu 2015-11-14 00:05:44 +01:00
commit ac8759cd3f
73 changed files with 2780 additions and 1334 deletions

View file

@ -29,10 +29,10 @@ class TestBingEngine(SearxTestCase):
self.assertRaises(AttributeError, bing.response, '')
self.assertRaises(AttributeError, bing.response, '[]')
response = mock.Mock(content='<html></html>')
response = mock.Mock(text='<html></html>')
self.assertEqual(bing.response(response), [])
response = mock.Mock(content='<html></html>')
response = mock.Mock(text='<html></html>')
self.assertEqual(bing.response(response), [])
html = """
@ -54,7 +54,7 @@ class TestBingEngine(SearxTestCase):
</div>
</div>
"""
response = mock.Mock(content=html)
response = mock.Mock(text=html)
results = bing.response(response)
self.assertEqual(type(results), list)
self.assertEqual(len(results), 1)
@ -81,7 +81,7 @@ class TestBingEngine(SearxTestCase):
</div>
</li>
"""
response = mock.Mock(content=html)
response = mock.Mock(text=html)
results = bing.response(response)
self.assertEqual(type(results), list)
self.assertEqual(len(results), 1)

View file

@ -31,10 +31,10 @@ class TestBingImagesEngine(SearxTestCase):
self.assertRaises(AttributeError, bing_images.response, '')
self.assertRaises(AttributeError, bing_images.response, '[]')
response = mock.Mock(content='<html></html>')
response = mock.Mock(text='<html></html>')
self.assertEqual(bing_images.response(response), [])
response = mock.Mock(content='<html></html>')
response = mock.Mock(text='<html></html>')
self.assertEqual(bing_images.response(response), [])
html = """
@ -52,7 +52,7 @@ oh:&quot;238&quot;,tft:&quot;0&quot;,oi:&quot;http://www.image.url/Images/Test%2
</div>
"""
html = html.replace('\r\n', '').replace('\n', '').replace('\r', '')
response = mock.Mock(content=html)
response = mock.Mock(text=html)
results = bing_images.response(response)
self.assertEqual(type(results), list)
self.assertEqual(len(results), 1)
@ -75,7 +75,7 @@ oh:&quot;238&quot;,tft:&quot;0&quot;,oi:&quot;http://www.image.url/Images/Test%2
style="height:144px;" width="178" height="144"/>
</a>
"""
response = mock.Mock(content=html)
response = mock.Mock(text=html)
results = bing_images.response(response)
self.assertEqual(type(results), list)
self.assertEqual(len(results), 0)
@ -263,7 +263,7 @@ oh:&quot;238&quot;,tft:&quot;0&quot;,oi:&quot;http://www.image.url/Images/Test%2
</div>
"""
html = html.replace('\r\n', '').replace('\n', '').replace('\r', '')
response = mock.Mock(content=html)
response = mock.Mock(text=html)
results = bing_images.response(response)
self.assertEqual(type(results), list)
self.assertEqual(len(results), 10)

View file

@ -1,3 +1,4 @@
# -*- coding: utf-8 -*-
from collections import defaultdict
import mock
from searx.engines import duckduckgo
@ -30,7 +31,7 @@ class TestDuckduckgoEngine(SearxTestCase):
response = mock.Mock(text='<html></html>')
self.assertEqual(duckduckgo.response(response), [])
html = """
html = u"""
<div class="results_links results_links_deep web-result">
<div class="icon_fav" style="display: block;">
<a rel="nofollow" href="https://www.test.com/">
@ -39,7 +40,7 @@ class TestDuckduckgoEngine(SearxTestCase):
</a>
</div>
<div class="links_main links_deep"> <!-- This is the visible part -->
<a rel="nofollow" class="large" href="http://this.should.be.the.link/">
<a rel="nofollow" class="large" href="http://this.should.be.the.link/ű">
This <b>is</b> <b>the</b> title
</a>
<div class="snippet"><b>This</b> should be the content.</div>
@ -54,7 +55,7 @@ class TestDuckduckgoEngine(SearxTestCase):
self.assertEqual(type(results), list)
self.assertEqual(len(results), 1)
self.assertEqual(results[0]['title'], 'This is the title')
self.assertEqual(results[0]['url'], 'http://this.should.be.the.link/')
self.assertEqual(results[0]['url'], u'http://this.should.be.the.link/ű')
self.assertEqual(results[0]['content'], 'This should be the content.')
html = """

View file

@ -15,7 +15,7 @@ class TestPiratebayEngine(SearxTestCase):
params = piratebay.request(query, dicto)
self.assertIn('url', params)
self.assertIn(query, params['url'])
self.assertIn('piratebay.am', params['url'])
self.assertIn('piratebay.se', params['url'])
self.assertIn('0', params['url'])
dicto['category'] = 'music'
@ -99,7 +99,7 @@ class TestPiratebayEngine(SearxTestCase):
self.assertEqual(type(results), list)
self.assertEqual(len(results), 2)
self.assertEqual(results[0]['title'], 'This is the title')
self.assertEqual(results[0]['url'], 'https://thepiratebay.am/this.is.the.link')
self.assertEqual(results[0]['url'], 'https://thepiratebay.se/this.is.the.link')
self.assertEqual(results[0]['content'], 'This is the content and should be OK')
self.assertEqual(results[0]['seed'], 13)
self.assertEqual(results[0]['leech'], 334)
@ -149,7 +149,7 @@ class TestPiratebayEngine(SearxTestCase):
self.assertEqual(type(results), list)
self.assertEqual(len(results), 1)
self.assertEqual(results[0]['title'], 'This is the title')
self.assertEqual(results[0]['url'], 'https://thepiratebay.am/this.is.the.link')
self.assertEqual(results[0]['url'], 'https://thepiratebay.se/this.is.the.link')
self.assertEqual(results[0]['content'], 'This is the content and should be OK')
self.assertEqual(results[0]['seed'], 0)
self.assertEqual(results[0]['leech'], 0)

View file

@ -42,7 +42,7 @@ class TestStartpageEngine(SearxTestCase):
</a>
<span id='title_stars_2' name='title_stars_2'> </span>
</h3>
<p class='desc'>
<p class='desc clk'>
This should be the content.
</p>
<p>
@ -78,7 +78,7 @@ class TestStartpageEngine(SearxTestCase):
</a>
<span id='title_stars_2' name='title_stars_2'> </span>
</h3>
<p class='desc'>
<p class='desc clk'>
This should be the content.
</p>
<p>
@ -101,7 +101,7 @@ class TestStartpageEngine(SearxTestCase):
<h3>
<span id='title_stars_2' name='title_stars_2'> </span>
</h3>
<p class='desc'>
<p class='desc clk'>
This should be the content.
</p>
<p>

View file

@ -1,204 +0,0 @@
from collections import defaultdict
import mock
from searx.engines import youtube
from searx.testing import SearxTestCase
class TestYoutubeEngine(SearxTestCase):
def test_request(self):
query = 'test_query'
dicto = defaultdict(dict)
dicto['pageno'] = 0
dicto['language'] = 'fr_FR'
params = youtube.request(query, dicto)
self.assertTrue('url' in params)
self.assertTrue(query in params['url'])
self.assertTrue('youtube.com' in params['url'])
self.assertTrue('fr' in params['url'])
dicto['language'] = 'all'
params = youtube.request(query, dicto)
self.assertFalse('fr' in params['url'])
def test_response(self):
self.assertRaises(AttributeError, youtube.response, None)
self.assertRaises(AttributeError, youtube.response, [])
self.assertRaises(AttributeError, youtube.response, '')
self.assertRaises(AttributeError, youtube.response, '[]')
response = mock.Mock(text='{}')
self.assertEqual(youtube.response(response), [])
response = mock.Mock(text='{"data": []}')
self.assertEqual(youtube.response(response), [])
json = """
{"feed":{"entry":[{
"id":{"$t":"http://gdata.youtube.com/feeds/api/videos/DIVZCPfAOeM"},
"published":{"$t":"2015-01-23T21:25:00.000Z"},
"updated":{"$t":"2015-01-26T14:38:15.000Z"},
"title":{"$t":"Title",
"type":"text"},"content":{"$t":"Description","type":"text"},
"link":[{"rel":"alternate","type":"text/html",
"href":"https://www.youtube.com/watch?v=DIVZCPfAOeM&feature=youtube_gdata"},
{"rel":"http://gdata.youtube.com/schemas/2007#video.related",
"type":"application/atom+xml",
"href":"https://gdata.youtube.com/feeds/api/videos/DIVZCPfAOeM/related"},
{"rel":"http://gdata.youtube.com/schemas/2007#mobile","type":"text/html",
"href":"https://m.youtube.com/details?v=DIVZCPfAOeM"},
{"rel":"self","type":"application/atom+xml",
"href":"https://gdata.youtube.com/feeds/api/videos/DIVZCPfAOeM"}],
"author":[{"name":{"$t":"Cauet"},
"uri":{"$t":"https://gdata.youtube.com/feeds/api/users/cauetofficiel"} }],
"gd$comments":{"gd$feedLink":{"rel":"http://gdata.youtube.com/schemas/2007#comments",
"href":"https://gdata.youtube.com/feeds/api/videos/DIVZCPfAOeM/comments",
"countHint":8} },
"media$group":{"media$category":[{"$t":"Comedy","label":"Comedy",
"scheme":"http://gdata.youtube.com/schemas/2007/categories.cat"}],
"media$content":[{"url":"https://www.youtube.com/v/DIVZCPfAOeM?version=3&f=videos&app=youtube_gdata",
"type":"application/x-shockwave-flash","medium":"video",
"isDefault":"true","expression":"full","duration":354,"yt$format":5},
{"url":"rtsp://r1---sn-cg07luel.c.youtube.com/CiILENy73wIaGQnjOcD3CFmFDBMYDSANFEgGUgZ2aWRlb3MM/0/0/0/video.3gp",
"type":"video/3gpp","medium":"video","expression":"full","duration":354,
"yt$format":1},
{"url":"rtsp://r1---sn-cg07luel.c.youtube.com/CiILENy73wIaGQnjOcD3CFmFDBMYESARFEgGUgZ2aWRlb3MM/0/0/0/video.3gp",
"type":"video/3gpp","medium":"video","expression":"full","duration":354,"yt$format":6}],
"media$description":{"$t":"Desc","type":"plain"},
"media$keywords":{},
"media$player":[{"url":"https://www.youtube.com/watch?v=DIVZCPfAOeM&feature=youtube_gdata_player"}],
"media$thumbnail":[{"url":"https://i.ytimg.com/vi/DIVZCPfAOeM/0.jpg",
"height":360,"width":480,"time":"00:02:57"},
{"url":"https://i.ytimg.com/vi/DIVZCPfAOeM/1.jpg","height":90,"width":120,"time":"00:01:28.500"},
{"url":"https://i.ytimg.com/vi/DIVZCPfAOeM/2.jpg","height":90,"width":120,"time":"00:02:57"},
{"url":"https://i.ytimg.com/vi/DIVZCPfAOeM/3.jpg","height":90,"width":120,"time":"00:04:25.500"}],
"media$title":{"$t":"Title","type":"plain"},
"yt$duration":{"seconds":"354"} },
"gd$rating":{"average":4.932159,"max":5,"min":1,"numRaters":1533,
"rel":"http://schemas.google.com/g/2005#overall"},
"yt$statistics":{"favoriteCount":"0","viewCount":"92464"} }
]
}
}
"""
response = mock.Mock(text=json)
results = youtube.response(response)
self.assertEqual(type(results), list)
self.assertEqual(len(results), 1)
self.assertEqual(results[0]['title'], 'Title')
self.assertEqual(results[0]['url'], 'https://www.youtube.com/watch?v=DIVZCPfAOeM')
self.assertEqual(results[0]['content'], 'Description')
self.assertEqual(results[0]['thumbnail'], 'https://i.ytimg.com/vi/DIVZCPfAOeM/0.jpg')
self.assertTrue('DIVZCPfAOeM' in results[0]['embedded'])
json = """
{"feed":{"entry":[{
"id":{"$t":"http://gdata.youtube.com/feeds/api/videos/DIVZCPfAOeM"},
"published":{"$t":"2015-01-23T21:25:00.000Z"},
"updated":{"$t":"2015-01-26T14:38:15.000Z"},
"title":{"$t":"Title",
"type":"text"},"content":{"$t":"Description","type":"text"},
"link":[{"rel":"http://gdata.youtube.com/schemas/2007#video.related",
"type":"application/atom+xml",
"href":"https://gdata.youtube.com/feeds/api/videos/DIVZCPfAOeM/related"},
{"rel":"self","type":"application/atom+xml",
"href":"https://gdata.youtube.com/feeds/api/videos/DIVZCPfAOeM"}],
"author":[{"name":{"$t":"Cauet"},
"uri":{"$t":"https://gdata.youtube.com/feeds/api/users/cauetofficiel"} }],
"gd$comments":{"gd$feedLink":{"rel":"http://gdata.youtube.com/schemas/2007#comments",
"href":"https://gdata.youtube.com/feeds/api/videos/DIVZCPfAOeM/comments",
"countHint":8} },
"media$group":{"media$category":[{"$t":"Comedy","label":"Comedy",
"scheme":"http://gdata.youtube.com/schemas/2007/categories.cat"}],
"media$content":[{"url":"https://www.youtube.com/v/DIVZCPfAOeM?version=3&f=videos&app=youtube_gdata",
"type":"application/x-shockwave-flash","medium":"video",
"isDefault":"true","expression":"full","duration":354,"yt$format":5},
{"url":"rtsp://r1---sn-cg07luel.c.youtube.com/CiILENy73wIaGQnjOcD3CFmFDBMYDSANFEgGUgZ2aWRlb3MM/0/0/0/video.3gp",
"type":"video/3gpp","medium":"video","expression":"full","duration":354,
"yt$format":1},
{"url":"rtsp://r1---sn-cg07luel.c.youtube.com/CiILENy73wIaGQnjOcD3CFmFDBMYESARFEgGUgZ2aWRlb3MM/0/0/0/video.3gp",
"type":"video/3gpp","medium":"video","expression":"full","duration":354,"yt$format":6}],
"media$description":{"$t":"Desc","type":"plain"},
"media$keywords":{},
"media$player":[{"url":"https://www.youtube.com/watch?v=DIVZCPfAOeM&feature=youtube_gdata_player"}],
"media$thumbnail":[{"url":"https://i.ytimg.com/vi/DIVZCPfAOeM/0.jpg",
"height":360,"width":480,"time":"00:02:57"},
{"url":"https://i.ytimg.com/vi/DIVZCPfAOeM/1.jpg","height":90,"width":120,"time":"00:01:28.500"},
{"url":"https://i.ytimg.com/vi/DIVZCPfAOeM/2.jpg","height":90,"width":120,"time":"00:02:57"},
{"url":"https://i.ytimg.com/vi/DIVZCPfAOeM/3.jpg","height":90,"width":120,"time":"00:04:25.500"}],
"media$title":{"$t":"Title","type":"plain"},
"yt$duration":{"seconds":"354"} },
"gd$rating":{"average":4.932159,"max":5,"min":1,"numRaters":1533,
"rel":"http://schemas.google.com/g/2005#overall"},
"yt$statistics":{"favoriteCount":"0","viewCount":"92464"} }
]
}
}
"""
response = mock.Mock(text=json)
results = youtube.response(response)
self.assertEqual(type(results), list)
self.assertEqual(len(results), 0)
json = """
{"feed":{"entry":[{
"id":{"$t":"http://gdata.youtube.com/feeds/api/videos/DIVZCPfAOeM"},
"published":{"$t":"2015-01-23T21:25:00.000Z"},
"updated":{"$t":"2015-01-26T14:38:15.000Z"},
"title":{"$t":"Title",
"type":"text"},"content":{"$t":"Description","type":"text"},
"link":[{"rel":"alternate","type":"text/html",
"href":"https://www.youtube.com/watch?v=DIVZCPfAOeM"},
{"rel":"http://gdata.youtube.com/schemas/2007#video.related",
"type":"application/atom+xml",
"href":"https://gdata.youtube.com/feeds/api/videos/DIVZCPfAOeM/related"},
{"rel":"http://gdata.youtube.com/schemas/2007#mobile","type":"text/html",
"href":"https://m.youtube.com/details?v=DIVZCPfAOeM"},
{"rel":"self","type":"application/atom+xml",
"href":"https://gdata.youtube.com/feeds/api/videos/DIVZCPfAOeM"}],
"author":[{"name":{"$t":"Cauet"},
"uri":{"$t":"https://gdata.youtube.com/feeds/api/users/cauetofficiel"} }],
"gd$comments":{"gd$feedLink":{"rel":"http://gdata.youtube.com/schemas/2007#comments",
"href":"https://gdata.youtube.com/feeds/api/videos/DIVZCPfAOeM/comments",
"countHint":8} },
"media$group":{"media$category":[{"$t":"Comedy","label":"Comedy",
"scheme":"http://gdata.youtube.com/schemas/2007/categories.cat"}],
"media$content":[{"url":"https://www.youtube.com/v/DIVZCPfAOeM?version=3&f=videos&app=youtube_gdata",
"type":"application/x-shockwave-flash","medium":"video",
"isDefault":"true","expression":"full","duration":354,"yt$format":5},
{"url":"rtsp://r1---sn-cg07luel.c.youtube.com/CiILENy73wIaGQnjOcD3CFmFDBMYDSANFEgGUgZ2aWRlb3MM/0/0/0/video.3gp",
"type":"video/3gpp","medium":"video","expression":"full","duration":354,
"yt$format":1},
{"url":"rtsp://r1---sn-cg07luel.c.youtube.com/CiILENy73wIaGQnjOcD3CFmFDBMYESARFEgGUgZ2aWRlb3MM/0/0/0/video.3gp",
"type":"video/3gpp","medium":"video","expression":"full","duration":354,"yt$format":6}],
"media$description":{"$t":"Desc","type":"plain"},
"media$keywords":{},
"media$player":[{"url":"https://www.youtube.com/watch?v=DIVZCPfAOeM&feature=youtube_gdata_player"}],
"media$title":{"$t":"Title","type":"plain"},
"yt$duration":{"seconds":"354"} },
"gd$rating":{"average":4.932159,"max":5,"min":1,"numRaters":1533,
"rel":"http://schemas.google.com/g/2005#overall"},
"yt$statistics":{"favoriteCount":"0","viewCount":"92464"} }
]
}
}
"""
response = mock.Mock(text=json)
results = youtube.response(response)
self.assertEqual(type(results), list)
self.assertEqual(len(results), 1)
self.assertEqual(results[0]['title'], 'Title')
self.assertEqual(results[0]['url'], 'https://www.youtube.com/watch?v=DIVZCPfAOeM')
self.assertEqual(results[0]['content'], 'Description')
self.assertEqual(results[0]['thumbnail'], '')
self.assertTrue('DIVZCPfAOeM' in results[0]['embedded'])
json = """
{"toto":{"entry":[]
}
}
"""
response = mock.Mock(text=json)
results = youtube.response(response)
self.assertEqual(type(results), list)
self.assertEqual(len(results), 0)

View file

@ -40,7 +40,6 @@ from searx.tests.engines.test_www1x import * # noqa
from searx.tests.engines.test_www500px import * # noqa
from searx.tests.engines.test_yacy import * # noqa
from searx.tests.engines.test_yahoo import * # noqa
from searx.tests.engines.test_youtube import * # noqa
from searx.tests.engines.test_youtube_api import * # noqa
from searx.tests.engines.test_youtube_noapi import * # noqa
from searx.tests.engines.test_yahoo_news import * # noqa

View file

@ -5,6 +5,12 @@ from searx import plugins
from mock import Mock
def get_search_mock(query, **kwargs):
return {'search': Mock(query=query,
result_container=Mock(answers=set()),
**kwargs)}
class PluginStoreTest(SearxTestCase):
def test_PluginStore_init(self):
@ -46,23 +52,23 @@ class SelfIPTest(SearxTestCase):
request = Mock(user_plugins=store.plugins,
remote_addr='127.0.0.1')
request.headers.getlist.return_value = []
ctx = {'search': Mock(answers=set(),
query='ip')}
ctx = get_search_mock(query='ip')
store.call('post_search', request, ctx)
self.assertTrue('127.0.0.1' in ctx['search'].answers)
self.assertTrue('127.0.0.1' in ctx['search'].result_container.answers)
# User agent test
request = Mock(user_plugins=store.plugins,
user_agent='Mock')
request.headers.getlist.return_value = []
ctx = {'search': Mock(answers=set(),
query='user-agent')}
ctx = get_search_mock(query='user-agent')
store.call('post_search', request, ctx)
self.assertTrue('Mock' in ctx['search'].answers)
ctx = {'search': Mock(answers=set(),
query='user agent')}
self.assertTrue('Mock' in ctx['search'].result_container.answers)
ctx = get_search_mock(query='user-agent')
store.call('post_search', request, ctx)
self.assertTrue('Mock' in ctx['search'].answers)
ctx = {'search': Mock(answers=set(),
query='What is my User-Agent?')}
self.assertTrue('Mock' in ctx['search'].result_container.answers)
ctx = get_search_mock(query='What is my User-Agent?')
store.call('post_search', request, ctx)
self.assertTrue('Mock' in ctx['search'].result_container.answers)

View file

@ -0,0 +1,41 @@
# -*- coding: utf-8 -*-
from searx.results import ResultContainer
from searx.testing import SearxTestCase
def fake_result(url='https://aa.bb/cc?dd=ee#ff',
title='aaa',
content='bbb',
engine='wikipedia', **kwargs):
result = {'url': url,
'title': title,
'content': content,
'engine': engine}
result.update(kwargs)
return result
# TODO
class ResultContainerTestCase(SearxTestCase):
def test_empty(self):
c = ResultContainer()
self.assertEqual(c.get_ordered_results(), [])
def test_one_result(self):
c = ResultContainer()
c.extend('wikipedia', [fake_result()])
self.assertEqual(c.results_length(), 1)
def test_one_suggestion(self):
c = ResultContainer()
c.extend('wikipedia', [fake_result(suggestion=True)])
self.assertEqual(len(c.suggestions), 1)
self.assertEqual(c.results_length(), 0)
def test_result_merge(self):
c = ResultContainer()
c.extend('wikipedia', [fake_result()])
c.extend('wikidata', [fake_result(), fake_result(url='https://example.com/')])
self.assertEqual(c.results_length(), 2)

View file

@ -1,25 +1,10 @@
# -*- coding: utf-8 -*-
from searx.search import score_results
from searx.testing import SearxTestCase
def fake_result(url='https://aa.bb/cc?dd=ee#ff',
title='aaa',
content='bbb',
engine='wikipedia'):
return {'url': url,
'title': title,
'content': content,
'engine': engine}
# TODO
class SearchTestCase(SearxTestCase):
class ScoreResultsTestCase(SearxTestCase):
def test_empty(self):
self.assertEqual(score_results(dict()), [])
def test_urlparse(self):
results = score_results(dict(a=[fake_result(url='https://aa.bb/cc?dd=ee#ff')]))
parsed_url = results[0]['parsed_url']
self.assertEqual(parsed_url.query, 'dd=ee')
def test_(self):
pass

View file

@ -1,3 +1,4 @@
# -*- coding: utf-8 -*-
import mock
from searx.testing import SearxTestCase
from searx import utils
@ -51,6 +52,15 @@ class TestUtils(SearxTestCase):
self.assertIsNotNone(utils.html_to_text(html))
self.assertEqual(utils.html_to_text(html), "Test text")
def test_prettify_url(self):
data = (('https://searx.me/', 'https://searx.me/'),
(u'https://searx.me/ű', u'https://searx.me/ű'),
('https://searx.me/' + (100 * 'a'), 'https://searx.me/[...]aaaaaaaaaaaaaaaaa'),
(u'https://searx.me/' + (100 * u'ű'), u'https://searx.me/[...]űűűűűűűűűűűűűűűűű'))
for test_url, expected in data:
self.assertEqual(utils.prettify_url(test_url, max_length=32), expected)
class TestHTMLTextExtractor(SearxTestCase):

View file

@ -1,6 +1,7 @@
# -*- coding: utf-8 -*-
import json
from mock import Mock
from urlparse import ParseResult
from searx import webapp
from searx.testing import SearxTestCase
@ -33,7 +34,12 @@ class ViewsTestCase(SearxTestCase):
]
def search_mock(search_self, *args):
search_self.results = self.test_results
search_self.result_container = Mock(get_ordered_results=lambda: self.test_results,
answers=set(),
suggestions=set(),
infoboxes=[],
results=self.test_results,
results_length=lambda: len(self.test_results))
webapp.Search.search = search_mock