def test_add_violation_when_sitemap_is_too_large(self):
page = PageFactory.create(url='http://globo.com')
reviewer = Reviewer(api_url='http://localhost:2368',
page_uuid=page.uuid,
page_url=page.url,
page_score=0.0,
config=Config(),
validators=[])
validator = SitemapValidator(reviewer)
validator.review.data['sitemap.files.size'] = {
'http://g1.globo.com/sitemap.xml': 10241
}
validator.review.data['sitemap.data'] = {
'http://g1.globo.com/sitemap.xml': Mock(status_code=200,
text='data')
}
validator.review.data['sitemap.files.urls'] = {
'http://g1.globo.com/sitemap.xml': 10
}
validator.review.data['sitemap.urls'] = {
'http://g1.globo.com/sitemap.xml': []
}
validator.add_violation = Mock()
validator.validate()
validator.add_violation.assert_called_once_with(
key='total.size.sitemap',
value={
'url': 'http://g1.globo.com/sitemap.xml',
'size': 10.0009765625
},
points=10)
def test_add_violation_when_sitemap_with_good_link(self):
page = PageFactory.create(url='http://globo.com')
reviewer = Reviewer(api_url='http://localhost:2368',
page_uuid=page.uuid,
page_url=page.url,
page_score=0.0,
config=Config(),
validators=[])
validator = SitemapValidator(reviewer)
validator.review.data['sitemap.files.size'] = {
'http://g1.globo.com/sitemap.xml': 10
}
validator.review.data['sitemap.data'] = {
'http://g1.globo.com/sitemap.xml':
Mock(status_code=200,
text='data',
url='http://g1.globo.com/%C3%BCmlat.php&q=name')
}
validator.review.data['sitemap.files.urls'] = {
'http://g1.globo.com/sitemap.xml': 20
}
validator.review.data['sitemap.urls'] = {
'http://g1.globo.com/sitemap.xml':
['http://g1.globo.com/%C3%BCmlat.php&q=name']
}
validator.add_violation = Mock()
validator.flush = Mock()
validator.validate()
expect(validator.add_violation.call_count).to_equal(0)
expect(validator.flush.call_count).to_equal(1)
def test_add_violation_when_404(self):
page = PageFactory.create(url='http://globo.com')
reviewer = Reviewer(api_url='http://localhost:2368',
page_uuid=page.uuid,
page_url=page.url,
page_score=0.0,
config=Config(),
validators=[])
validator = SitemapValidator(reviewer)
validator.review.data['sitemap.files.size'] = {
'http://g1.globo.com/sitemap.xml': 10
}
validator.review.data['sitemap.data'] = {
'http://g1.globo.com/sitemap.xml': Mock(status_code=404, text=None)
}
validator.add_violation = Mock()
validator.validate()
validator.add_violation.assert_called_once_with(
key='sitemap.not_found',
value='http://g1.globo.com/sitemap.xml',
points=100)
def test_add_violation_when_sitemap_has_links_that_not_need_to_be_encoded(
self):
page = PageFactory.create(url='http://globo.com')
reviewer = Reviewer(api_url='http://localhost:2368',
page_uuid=page.uuid,
page_url=page.url,
page_score=0.0,
config=Config(),
validators=[])
validator = SitemapValidator(reviewer)
validator.review.data['sitemap.files.size'] = {
'http://g1.globo.com/sitemap.xml': 10
}
validator.review.data['sitemap.data'] = {
'http://g1.globo.com/sitemap.xml': Mock(status_code=200,
text='data')
}
validator.review.data['sitemap.files.urls'] = {
'http://g1.globo.com/sitemap.xml': 20
}
validator.review.data['sitemap.urls'] = {
'http://g1.globo.com/sitemap.xml': ['http://g1.globo.com/1.html']
}
validator.add_violation = Mock()
validator.validate()
expect(validator.add_violation.call_count).to_equal(0)
def test_add_violation_when_sitemap_is_too_large(self):
page = PageFactory.create(url='http://globo.com')
reviewer = Reviewer(
api_url='http://localhost:2368',
page_uuid=page.uuid,
page_url=page.url,
page_score=0.0,
config=Config(),
validators=[]
)
validator = SitemapValidator(reviewer)
validator.review.data['sitemap.files.size'] = {'http://g1.globo.com/sitemap.xml': 10241}
validator.review.data['sitemap.data'] = {'http://g1.globo.com/sitemap.xml': Mock(status_code=200, text='data')}
validator.review.data['sitemap.files.urls'] = {'http://g1.globo.com/sitemap.xml': 10}
validator.review.data['sitemap.urls'] = {'http://g1.globo.com/sitemap.xml': []}
validator.add_violation = Mock()
validator.validate()
validator.add_violation.assert_called_once_with(
key='total.size.sitemap',
value={
'url': 'http://g1.globo.com/sitemap.xml',
'size': 10.0009765625
},
points=10
)
def test_add_violation_when_sitemap_has_links_that_need_to_be_encoded_with_amp(self):
page = PageFactory.create(url='http://globo.com')
reviewer = Reviewer(
api_url='http://localhost:2368',
page_uuid=page.uuid,
page_url=page.url,
page_score=0.0,
config=Config(),
validators=[]
)
validator = SitemapValidator(reviewer)
validator.review.data['sitemap.files.size'] = {'http://g1.globo.com/sitemap.xml': 10}
validator.review.data['sitemap.data'] = {'http://g1.globo.com/sitemap.xml': Mock(status_code=200, text='data')}
validator.review.data['sitemap.files.urls'] = {'http://g1.globo.com/sitemap.xml': 20}
validator.review.data['sitemap.urls'] = {'http://g1.globo.com/sitemap.xml': ['http://g1.globo.com/%C3%BCmlat.php&q=name']}
validator.add_violation = Mock()
validator.validate()
validator.add_violation.assert_called_once_with(
key='sitemap.links.not_encoded',
value={'url': 'http://g1.globo.com/sitemap.xml', 'links': 1},
points=10
)
def test_return_if_page_url_is_not_root_of_domain(self):
page = PageFactory.create(url='http://globo.com/1')
reviewer = Reviewer(
api_url='http://localhost:2368',
page_uuid=page.uuid,
page_url=page.url,
page_score=0.0,
config=Config(),
validators=[]
)
validator = SitemapValidator(reviewer)
validator.review.data['sitemap.files.size'] = {'http://g1.globo.com/sitemap.xml': 10}
validator.review.data['sitemap.data'] = {'http://g1.globo.com/sitemap.xml': Mock(status_code=404, text=None)}
validator.add_violation = Mock()
validator.validate()
expect(validator.add_violation.call_count).to_equal(0)
def test_add_violation_when_sitemap_has_links_that_need_to_be_encoded(
self):
page = PageFactory.create(url='http://globo.com')
reviewer = Reviewer(api_url='http://localhost:2368',
page_uuid=page.uuid,
page_url=page.url,
page_score=0.0,
config=Config(),
validators=[])
validator = SitemapValidator(reviewer)
validator.review.data['sitemap.files.size'] = {
'http://g1.globo.com/sitemap.xml': 10
}
validator.review.data['sitemap.data'] = {
'http://g1.globo.com/sitemap.xml': Mock(status_code=200,
text='data')
}
validator.review.data['sitemap.files.urls'] = {
'http://g1.globo.com/sitemap.xml': 20
}
validator.review.data['sitemap.urls'] = {
'http://g1.globo.com/sitemap.xml': [
'http://g1.globo.com/ümlat.php',
u'http://g1.globo.com/ümlat.php'
]
}
validator.add_violation = Mock()
# validator.send_url = Mock()
validator.validate()
validator.add_violation.assert_called_once_with(
key='sitemap.links.not_encoded',
value={
'url': 'http://g1.globo.com/sitemap.xml',
'links': 2
},
points=10)
def test_return_if_page_url_is_not_root_of_domain(self):
page = PageFactory.create(url='http://globo.com/1')
reviewer = Reviewer(api_url='http://localhost:2368',
page_uuid=page.uuid,
page_url=page.url,
page_score=0.0,
config=Config(),
validators=[])
validator = SitemapValidator(reviewer)
validator.review.data['sitemap.files.size'] = {
'http://g1.globo.com/sitemap.xml': 10
}
validator.review.data['sitemap.data'] = {
'http://g1.globo.com/sitemap.xml': Mock(status_code=404, text=None)
}
validator.add_violation = Mock()
validator.validate()
expect(validator.add_violation.call_count).to_equal(0)
def test_add_violation_when_sitemap_has_links_that_not_need_to_be_encoded(self):
page = PageFactory.create(url='http://globo.com')
reviewer = Reviewer(
api_url='http://localhost:2368',
page_uuid=page.uuid,
page_url=page.url,
page_score=0.0,
config=Config(),
validators=[]
)
validator = SitemapValidator(reviewer)
validator.review.data['sitemap.files.size'] = {'http://g1.globo.com/sitemap.xml': 10}
validator.review.data['sitemap.data'] = {'http://g1.globo.com/sitemap.xml': Mock(status_code=200, text='data')}
validator.review.data['sitemap.files.urls'] = {'http://g1.globo.com/sitemap.xml': 20}
validator.review.data['sitemap.urls'] = {'http://g1.globo.com/sitemap.xml': ['http://g1.globo.com/1.html']}
validator.add_violation = Mock()
validator.validate()
expect(validator.add_violation.call_count).to_equal(0)
def test_add_violation_when_404(self):
page = PageFactory.create(url='http://globo.com')
reviewer = Reviewer(
api_url='http://localhost:2368',
page_uuid=page.uuid,
page_url=page.url,
page_score=0.0,
config=Config(),
validators=[]
)
validator = SitemapValidator(reviewer)
validator.review.data['sitemap.files.size'] = {'http://g1.globo.com/sitemap.xml': 10}
validator.review.data['sitemap.data'] = {'http://g1.globo.com/sitemap.xml': Mock(status_code=404, text=None)}
validator.add_violation = Mock()
validator.validate()
validator.add_violation.assert_called_once_with(
key='sitemap.not_found',
value='http://g1.globo.com/sitemap.xml',
points=100
)
def test_add_violation_when_sitemap_with_good_link(self):
page = PageFactory.create(url='http://globo.com')
reviewer = Reviewer(
api_url='http://localhost:2368',
page_uuid=page.uuid,
page_url=page.url,
page_score=0.0,
config=Config(),
validators=[]
)
validator = SitemapValidator(reviewer)
validator.review.data['sitemap.files.size'] = {'http://g1.globo.com/sitemap.xml': 10}
validator.review.data['sitemap.data'] = {'http://g1.globo.com/sitemap.xml': Mock(status_code=200, text='data', url='http://g1.globo.com/%C3%BCmlat.php&q=name')}
validator.review.data['sitemap.files.urls'] = {'http://g1.globo.com/sitemap.xml': 20}
validator.review.data['sitemap.urls'] = {'http://g1.globo.com/sitemap.xml': ['http://g1.globo.com/%C3%BCmlat.php&q=name']}
validator.add_violation = Mock()
validator.flush = Mock()
validator.validate()
expect(validator.add_violation.call_count).to_equal(0)
expect(validator.flush.call_count).to_equal(1)