def test_articles_skips_short_articles(self):
extracted_data = MagicMock()
extracted_data.cleaned_text = 'short full text'
self.create_patch('argos.core.membrane.extractor.extract_entry_data', return_value=(extracted_data, 'short full text'))
articles = []
collector.get_articles(self.feed, lambda a: articles.append(a))
self.assertEquals(len(articles), 0)
def test_articles_skips_unreachable_articles(self):
from urllib import error
self.create_patch('argos.core.membrane.extractor.extract_entry_data',
side_effect=[error.URLError('unreachable')])
articles = []
collector.get_articles(self.feed, lambda a: articles.append(a))
self.assertEquals(len(articles), 0)
def test_articles_ignores_existing(self):
self.patch_extraction()
articles = []
collector.get_articles(self.feed, lambda a: articles.append(a))
collector.get_articles(self.feed, lambda a: articles.append(a))
self.assertEquals(Article.query.count(), 1)
def test_articles_skips_short_articles(self):
extracted_data = MagicMock()
extracted_data.cleaned_text = 'short full text'
self.create_patch('argos.core.membrane.extractor.extract_entry_data',
return_value=(extracted_data, 'short full text'))
articles = []
collector.get_articles(self.feed, lambda a: articles.append(a))
self.assertEquals(len(articles), 0)
def test_articles_ignores_existing(self):
self.patch_extraction()
articles = []
collector.get_articles(self.feed, lambda a: articles.append(a))
collector.get_articles(self.feed, lambda a: articles.append(a))
self.assertEquals(Article.query.count(), 1)
def test_articles_skips_404_articles(self):
from urllib import error
self.create_patch('argos.core.membrane.extractor.extract_entry_data',
side_effect=[
error.HTTPError(url=None,
code=404,
msg=None,
hdrs=None,
fp=None)
])
articles = []
collector.get_articles(self.feed, lambda a: articles.append(a))
self.assertEquals(len(articles), 0)
def test_collect_ignores_same_source_and_title_if_urls_differ(self):
self.patch_extraction()
articles = []
collector.get_articles(self.feed, lambda a: articles.append(a))
# Setup the same article, but with a different link.
article = {
'links': [{
'href': 'http://another_url.com'
}],
'title': 'some title',
'published': 'Thu, 09 Jan 2014 14:00:00 GMT',
'updated': 'Thu, 09 Jan 2014 14:00:00 GMT'
}
data = MagicMock(entries=[article], bozo=False)
self.mock_parse.return_value = data
collector.get_articles(self.feed, lambda a: articles.append(a))
self.assertEquals(Article.query.count(), 1)
def test_collect_ignores_same_source_and_title_if_urls_differ(self):
self.patch_extraction()
articles = []
collector.get_articles(self.feed, lambda a: articles.append(a))
# Setup the same article, but with a different link.
article = {
'links': [{'href': 'http://another_url.com'}],
'title': 'some title',
'published': 'Thu, 09 Jan 2014 14:00:00 GMT',
'updated': 'Thu, 09 Jan 2014 14:00:00 GMT'
}
data = MagicMock(
entries=[article],
bozo=False
)
self.mock_parse.return_value = data
collector.get_articles(self.feed, lambda a: articles.append(a))
self.assertEquals(Article.query.count(), 1)
def test_articles_skips_unreachable_articles(self):
from urllib import error
self.create_patch('argos.core.membrane.extractor.extract_entry_data', side_effect=[error.URLError('unreachable')])
articles = []
collector.get_articles(self.feed, lambda a: articles.append(a))
self.assertEquals(len(articles), 0)
def test_articles_skips_404_articles(self):
from urllib import error
self.create_patch('argos.core.membrane.extractor.extract_entry_data', side_effect=[error.HTTPError(url=None, code=404, msg=None, hdrs=None, fp=None)])
articles = []
collector.get_articles(self.feed, lambda a: articles.append(a))
self.assertEquals(len(articles), 0)
def test_articles(self):
self.patch_extraction()
articles = []
collector.get_articles(self.feed, lambda a: articles.append(a))
self.assertEquals(len(articles), 1)
def test_articles(self):
self.patch_extraction()
articles = []
collector.get_articles(self.feed, lambda a: articles.append(a))
self.assertEquals(len(articles), 1)
