class TestHtmlParser(unittest.TestCase): def setUp(self): f = codecs.open('test-page1.html', 'r', 'utf-8') self.html = f.read() self.reader = HtmlParser() def testParseMatchesInPlayerPage(self): matches = self.reader.find_matches(self.html) answer = [Match(656830280, datetime(2014, 5, 12, 18, 40), (28, 4)), Match(656759204, datetime(2014, 5, 12, 18, 0), (36,54)), Match(655653242, datetime(2014, 5, 12, 5, 10), (42,25)), Match(655639078, datetime(2014, 5, 12, 4, 16), (5,39)), Match(655610044, datetime(2014, 5, 12, 4, 4), (34,1)), Match(655198197, datetime(2014, 5, 11, 19, 31), (46,15)), Match(654944424, datetime(2014, 5, 11, 17, 5), (54,24)), Match(654717322, datetime(2014, 5, 11, 14, 53), (39,38)), Match(653725177, datetime(2014, 5, 11, 5, 12), (38,24)), Match(653221484, datetime(2014, 5, 10, 20, 16), (42,8)), Match(653097369, datetime(2014, 5, 10, 19, 1), (42,29)), Match(652982870, datetime(2014, 5, 10, 17, 50), (44,8)), Match(652756205, datetime(2014, 5, 10, 15, 41), (29,57)), Match(652288108, datetime(2014, 5, 10, 11, 47), (40,8)), Match(651421025, datetime(2014, 5, 9, 21, 29), (41,8)), Match(651256378, datetime(2014, 5, 9, 19, 22), (43,51)), Match(651206198, datetime(2014, 5, 9, 18, 29), (26,29)), Match(650785626, datetime(2014, 5, 9, 14, 47), (54,27)), Match(650717825, datetime(2014, 5, 9, 13, 44), (28,34)), Match(650014223, datetime(2014, 5, 9, 4, 43), (44,5))] length = min(len(answer), len(matches)) for i in range(0, length): expected_match = answer[i] match = matches[i] self.assertEqual(expected_match.match_id, match.match_id) self.assertEqual(expected_match.datetime, match.datetime) self.assertEqual(expected_match.duration, match.duration) self.assertEqual(20, len(matches)) def testStringToDatetime(self): date = datetime.strptime('May 12, 2014 18:40 +00:00', '%b %d, %Y %H:%M +00:00') self.assertEqual(2014, date.year) self.assertEqual(5, date.month) self.assertEqual(12, date.day) self.assertEqual(18, date.hour) self.assertEqual(40, date.minute)
def setUp(self): f = codecs.open('test-page1.html', 'r', 'utf-8') self.html = f.read() self.reader = HtmlParser()