2015-09-07 19:22:01 +02:00
|
|
|
# -*- coding: utf-8 -*-
|
2014-07-12 15:46:55 +02:00
|
|
|
import mock
|
2016-11-30 18:43:03 +01:00
|
|
|
import sys
|
2014-07-12 15:46:55 +02:00
|
|
|
from searx.testing import SearxTestCase
|
|
|
|
|
from searx import utils
|
|
|
|
|
|
2016-11-30 18:43:03 +01:00
|
|
|
if sys.version_info[0] == 3:
|
|
|
|
|
unicode = str
|
|
|
|
|
|
2014-07-12 15:46:55 +02:00
|
|
|
|
|
|
|
|
class TestUtils(SearxTestCase):
|
|
|
|
|
|
|
|
|
|
def test_gen_useragent(self):
|
|
|
|
|
self.assertIsInstance(utils.gen_useragent(), str)
|
|
|
|
|
self.assertIsNotNone(utils.gen_useragent())
|
|
|
|
|
self.assertTrue(utils.gen_useragent().startswith('Mozilla'))
|
|
|
|
|
|
2015-01-27 20:03:33 +01:00
|
|
|
def test_searx_useragent(self):
|
|
|
|
|
self.assertIsInstance(utils.searx_useragent(), str)
|
|
|
|
|
self.assertIsNotNone(utils.searx_useragent())
|
|
|
|
|
self.assertTrue(utils.searx_useragent().startswith('searx'))
|
|
|
|
|
|
2014-07-12 15:46:55 +02:00
|
|
|
def test_highlight_content(self):
|
|
|
|
|
self.assertEqual(utils.highlight_content(0, None), None)
|
|
|
|
|
self.assertEqual(utils.highlight_content(None, None), None)
|
|
|
|
|
self.assertEqual(utils.highlight_content('', None), None)
|
|
|
|
|
self.assertEqual(utils.highlight_content(False, None), None)
|
|
|
|
|
|
|
|
|
|
contents = [
|
|
|
|
|
'<html></html>'
|
|
|
|
|
'not<'
|
|
|
|
|
]
|
|
|
|
|
for content in contents:
|
|
|
|
|
self.assertEqual(utils.highlight_content(content, None), content)
|
|
|
|
|
|
|
|
|
|
content = 'a'
|
2016-11-30 18:43:03 +01:00
|
|
|
query = b'test'
|
2014-07-12 15:46:55 +02:00
|
|
|
self.assertEqual(utils.highlight_content(content, query), content)
|
2016-11-30 18:43:03 +01:00
|
|
|
query = b'a test'
|
2014-07-12 15:46:55 +02:00
|
|
|
self.assertEqual(utils.highlight_content(content, query), content)
|
|
|
|
|
|
2015-01-27 20:03:33 +01:00
|
|
|
def test_html_to_text(self):
|
|
|
|
|
html = """
|
|
|
|
|
<a href="/testlink" class="link_access_account">
|
|
|
|
|
<span class="toto">
|
|
|
|
|
<span>
|
|
|
|
|
<img src="test.jpg" />
|
|
|
|
|
</span>
|
|
|
|
|
</span>
|
|
|
|
|
<span class="titi">
|
|
|
|
|
Test text
|
|
|
|
|
</span>
|
|
|
|
|
</a>
|
|
|
|
|
"""
|
|
|
|
|
self.assertIsInstance(utils.html_to_text(html), unicode)
|
|
|
|
|
self.assertIsNotNone(utils.html_to_text(html))
|
|
|
|
|
self.assertEqual(utils.html_to_text(html), "Test text")
|
|
|
|
|
|
2015-09-07 19:22:01 +02:00
|
|
|
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)
|
|
|
|
|
|
2014-07-12 15:46:55 +02:00
|
|
|
|
|
|
|
|
class TestHTMLTextExtractor(SearxTestCase):
|
|
|
|
|
|
|
|
|
|
def setUp(self):
|
|
|
|
|
self.html_text_extractor = utils.HTMLTextExtractor()
|
|
|
|
|
|
|
|
|
|
def test__init__(self):
|
|
|
|
|
self.assertEqual(self.html_text_extractor.result, [])
|
|
|
|
|
|
|
|
|
|
def test_handle_charref(self):
|
|
|
|
|
self.html_text_extractor.handle_charref('xF')
|
|
|
|
|
self.assertIn(u'\x0f', self.html_text_extractor.result)
|
|
|
|
|
self.html_text_extractor.handle_charref('XF')
|
|
|
|
|
self.assertIn(u'\x0f', self.html_text_extractor.result)
|
|
|
|
|
|
|
|
|
|
self.html_text_extractor.handle_charref('97')
|
|
|
|
|
self.assertIn(u'a', self.html_text_extractor.result)
|
|
|
|
|
|
|
|
|
|
def test_handle_entityref(self):
|
|
|
|
|
entity = 'test'
|
|
|
|
|
self.html_text_extractor.handle_entityref(entity)
|
|
|
|
|
self.assertIn(entity, self.html_text_extractor.result)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class TestUnicodeWriter(SearxTestCase):
|
|
|
|
|
|
|
|
|
|
def setUp(self):
|
|
|
|
|
self.unicode_writer = utils.UnicodeWriter(mock.MagicMock())
|
|
|
|
|
|
|
|
|
|
def test_write_row(self):
|
|
|
|
|
row = [1, 2, 3]
|
|
|
|
|
self.assertEqual(self.unicode_writer.writerow(row), None)
|
|
|
|
|
|
|
|
|
|
def test_write_rows(self):
|
|
|
|
|
self.unicode_writer.writerow = mock.MagicMock()
|
|
|
|
|
rows = [1, 2, 3]
|
|
|
|
|
self.unicode_writer.writerows(rows)
|
|
|
|
|
self.assertEqual(self.unicode_writer.writerow.call_count, len(rows))
|
