def test_regrab_miro_bugs_refreshes_older_bugs_even_when_missing_from_csv(self, mock_csv_maker, mock_xml_opener): mock_xml_opener.return_value = open(os.path.join( settings.MEDIA_ROOT, 'sample-data', 'miro-2294-2009-08-06.xml')) # Situation: Assume there are zero bitesized bugs today. # Desire: We re-get old bugs that don't show up in the CSV. # Prereq: We have some bug with lame data: bug = Bug() bug.people_involved = 1 bug.canonical_bug_link = 'http://bugzilla.pculture.org/show_bug.cgi?id=2294' bug.date_reported = datetime.datetime.now() bug.last_touched = datetime.datetime.now() bug.last_polled = datetime.datetime.now() bug.project, _ = Project.objects.get_or_create(name='Miro') bug.save() # Prepare a fake CSV that is empty mock_csv_maker.return_value = StringIO('') # Now, do a crawl and notice that we updated the bug even though the CSV is empty mysite.customs.miro.grab_miro_bugs() # refreshes no bugs since CSV is empty! all_bugs = Bug.all_bugs.all() self.assertEqual(len(all_bugs), 1) bug = all_bugs[0] self.assertEqual(bug.people_involved, 5)
bug.canonical_bug_link = remote_bug_url bug.good_for_newcomers = 'oss-bite-size' in metadata_dict['Keywords'] bug.status = metadata_dict['State'] status_number = int(bug.status.split('-')[0]) bug.looks_closed = (status_number in (8, 10, 11)) bug.title = metadata_dict['Synopsis'] bug.importance = '' # No importance, as far as I can tell. # For description, just grab the first "message" bug.description = metadata_dict['Description'] # We are always the project called Python. bug.project, _ = Project.objects.get_or_create(name='OpenSolaris OS/Net', language='C') # How many people participated? bug.people_involved = None # This tracker has no idea. bug.last_polled = datetime.datetime.utcnow() return bug BUG_URL_PREFIX = 'http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=' def get_remote_bug_ids_to_read(): remote_bug_list = 'http://hub.opensolaris.org/bin/view/Main/oss_bite_size' tree = lxml.html.document_fromstring(urllib2.urlopen(remote_bug_list).read()) for a in tree.cssselect('a'): if BUG_URL_PREFIX in ( a.attrib.get('href', '')): yield bug_url2bug_id(a.attrib['href'])