access processor now handles extra fields in the log line

This commit is contained in:
John Dickinson 2011-01-17 11:07:58 -06:00
parent b3c914e875
commit d8810a1ed7
2 changed files with 46 additions and 4 deletions

View File

@ -40,7 +40,7 @@ class AccessLogProcessor(object):
'''given a raw access log line, return a dict of the good parts'''
d = {}
try:
(_,
(unused,
server,
client_ip,
lb_ip,
@ -57,7 +57,8 @@ class AccessLogProcessor(object):
etag,
trans_id,
headers,
processing_time) = (unquote(x) for x in raw_log[16:].split(' '))
processing_time) = (unquote(x) for x in
raw_log[16:].split(' ')[:18])
except ValueError:
self.logger.debug(_('Bad line data: %s') % repr(raw_log))
return {}

View File

@ -21,8 +21,49 @@ from swift.stats import access_processor
class TestAccessProcessor(unittest.TestCase):
def test_placeholder(self):
pass
def test_log_line_parser_field_count(self):
p = access_processor.AccessLogProcessor({})
# too few fields
log_line = [str(x) for x in range(17)]
log_line[1] = 'proxy-server'
log_line[4] = '1/Jan/3/4/5/6'
log_line[6] = '/v1/a/c/o'
log_line = 'x'*16 + ' '.join(log_line)
res = p.log_line_parser(log_line)
expected = {}
self.assertEquals(res, expected)
# right amount of fields
log_line = [str(x) for x in range(18)]
log_line[1] = 'proxy-server'
log_line[4] = '1/Jan/3/4/5/6'
log_line[6] = '/v1/a/c/o'
log_line = 'x'*16 + ' '.join(log_line)
res = p.log_line_parser(log_line)
expected = {'code': 8, 'processing_time': '17', 'auth_token': '11',
'month': '01', 'second': '6', 'year': '3', 'tz': '+0000',
'http_version': '7', 'object_name': 'o', 'etag': '14',
'method': '5', 'trans_id': '15', 'client_ip': '2',
'bytes_out': 13, 'container_name': 'c', 'day': '1',
'minute': '5', 'account': 'a', 'hour': '4',
'referrer': '9', 'request': '/v1/a/c/o',
'user_agent': '10', 'bytes_in': 12, 'lb_ip': '3'}
self.assertEquals(res, expected)
# too many fields
log_line = [str(x) for x in range(19)]
log_line[1] = 'proxy-server'
log_line[4] = '1/Jan/3/4/5/6'
log_line[6] = '/v1/a/c/o'
log_line = 'x'*16 + ' '.join(log_line)
res = p.log_line_parser(log_line)
expected = {'code': 8, 'processing_time': '17', 'auth_token': '11',
'month': '01', 'second': '6', 'year': '3', 'tz': '+0000',
'http_version': '7', 'object_name': 'o', 'etag': '14',
'method': '5', 'trans_id': '15', 'client_ip': '2',
'bytes_out': 13, 'container_name': 'c', 'day': '1',
'minute': '5', 'account': 'a', 'hour': '4',
'referrer': '9', 'request': '/v1/a/c/o',
'user_agent': '10', 'bytes_in': 12, 'lb_ip': '3'}
self.assertEquals(res, expected)
if __name__ == '__main__':