def test_basic(self): pkg_id = c.pkg.id # before filter # <a href="http://www.annakarenina.com/download/x=1&y=2" target="_blank">Full text. Needs escaping: " Umlaut: u</a> self.check_named_element(self.pkg_page, 'a', '!href="%s"' % self.harvest_xml_url) self.check_named_element(self.pkg_page, 'a', '!href="%s"' % self.harvest_html_url) res = harvest_filter(HTML(self.pkg_page), c.pkg) res = res.render('html').decode('utf8') # after filter self.check_named_element(res, 'a', 'href="%s"' % self.harvest_xml_url) self.check_named_element(res, 'a', 'href="%s"' % self.harvest_html_url)
def test_basic(self): pkg = model.Package.by_name(u'annakarenina') harvest_object_id = pkg.extras.get('harvest_object_id') harvest_xml_url = '/api/2/rest/harvestobject/%s/xml' % harvest_object_id harvest_html_url = '/api/2/rest/harvestobject/%s/html' % harvest_object_id # before filter pkg_page = HTML(self.pkg_page).render() self.check_named_element(pkg_page, 'a', '!href="%s"' % harvest_xml_url) self.check_named_element(pkg_page, 'a', '!href="%s"' % harvest_html_url) anna = model.Package.by_name(u'annakarenina') res = harvest_filter(HTML(self.pkg_page), anna) res = res.render('html').decode('utf8') print res # after filter self.check_named_element(res, 'a', 'href="%s"' % harvest_xml_url) self.check_named_element(res, 'a', 'href="%s"' % harvest_html_url)