def aggregate_data(top_dir, file_pattern):
'''Given the diretory path for top directory `top_dir` and the filename
pattern to search, return a tuple of all required information to write to
report. The report includes the following info:
* how many bytes were sent
* the top 100 largest cumulative size requests
* the number of all unique host IP addresses per week
* all 404-status requests sorted by its frequency
'''
Data = namedtuple('Data', 'total_sent, cumulative_size_requests, \
weekly_uniq_hosts, status_404_requests')
total_sent = 0
cumulative_size_requests= {}
status_404_requests = {}
weekly_uniq_hosts = {}
# +++your code here+++
apache_data = apache_log_infos(top_dir, file_pattern)
total_sent = get_bytes(apache_data)
cumulative_size_request = get_cumulative_size_request(apache_data)
status_404_request = get_status_404_request(apache_data)
weekly_uniq_hosts = get_weekly_uniq_hosts(apache_data)
return Data(total_sent, cumulative_size_requests,
weekly_uniq_hosts, status_404_requests)
def aggregate_info(top_dir, file_pattern):
'''Given the diretory path for top directory `top_dir` and the filename
pattern to search, return a tuple of all required information to write to
report. The report includes the following info:
* how many bytes were sent
* the top 100 largest cumulative size requests
* the number of all unique host IP addresses per week
* all 404-status requests sorted by its frequency
'''
Data = namedtuple('Data', 'total_sent, cumulative_size_requests, \
weekly_uniq_hosts, status_404_requests')
total_sent = 0
cumulative_size_requests = {}
status_404_requests = {}
weekly_uniq_hosts = {}
#+++your code here+++
newdict = apache_log_infos(top_dir, file_pattern)
for info in newdict:
total_sent += int(info['size'])
#Find the 404 statuses
if info['status'] == '404':
if info['request'] in status_404_requests:
status_404_requests[info['request']] += 1
else:
status_404_requests[info['request']] = 1
#Get the unique hosts
if info['week'] in weekly_uniq_hosts:
weekly_uniq_hosts[info['week']].add(info['host'])
else:
weekly_uniq_hosts[info['week']] = {info['host']}
#Get the cumulative size
if info['request'] in cumulative_size_requests:
cumulative_size_requests[info['request']] += int(info['size'])
else:
cumulative_size_requests[info['request']] = int(info['size'])
cumulative_size_requests = sorted(cumulative_size_requests.items(), key=itemgetter(1), reverse=True)
status_404_requests = sorted(status_404_requests.items(), key=itemgetter(1), reverse=True)
return Data(total_sent, cumulative_size_requests,
weekly_uniq_hosts, status_404_requests)
