2014-11-18 20:18:53 +01:00
|
|
|
#!/usr/bin/env python
|
|
|
|
|
|
|
|
import os
|
|
|
|
import re
|
|
|
|
import time
|
|
|
|
import glob
|
|
|
|
import imp
|
2014-11-19 19:34:16 +01:00
|
|
|
import pickle
|
|
|
|
import gzip
|
2014-11-19 19:45:41 +01:00
|
|
|
|
2014-11-20 15:25:43 +01:00
|
|
|
# Default configuration
|
|
|
|
|
|
|
|
DB_ROOT = './output/'
|
|
|
|
DISPLAY_ROOT = './output/'
|
|
|
|
|
|
|
|
log_format = '$server_name:$server_port $remote_addr - $remote_user [$time_local] ' +\
|
|
|
|
'"$request" $status $body_bytes_sent ' +\
|
|
|
|
'"$http_referer" "$http_user_agent"';
|
|
|
|
|
|
|
|
time_format = '%d/%b/%Y:%H:%M:%S +0100'
|
|
|
|
|
2014-11-20 16:15:57 +01:00
|
|
|
pre_analysis_hooks = []
|
|
|
|
post_analysis_hooks = []
|
|
|
|
display_hooks = []
|
|
|
|
|
2014-11-20 15:25:43 +01:00
|
|
|
from conf import *
|
|
|
|
|
2014-11-18 20:18:53 +01:00
|
|
|
print '==> Start'
|
|
|
|
|
2014-11-20 09:37:54 +01:00
|
|
|
meta_visit = {}
|
2014-11-19 19:34:16 +01:00
|
|
|
analyse_started = False
|
2014-11-19 19:45:41 +01:00
|
|
|
current_visits = {}
|
2014-11-19 21:37:37 +01:00
|
|
|
cache_plugins = {}
|
2014-11-20 11:50:06 +01:00
|
|
|
display = {}
|
2014-11-18 20:18:53 +01:00
|
|
|
|
|
|
|
log_format_extracted = re.sub(r'([^\$\w])', r'\\\g<1>', log_format);
|
|
|
|
log_format_extracted = re.sub(r'\$(\w+)', '(?P<\g<1>>.+)', log_format_extracted)
|
|
|
|
http_request_extracted = re.compile(r'(?P<http_method>\S+) (?P<http_uri>\S+) (?P<http_version>\S+)')
|
|
|
|
|
|
|
|
log_re = re.compile(log_format_extracted)
|
2014-11-19 08:01:12 +01:00
|
|
|
uri_re = re.compile(r'(?P<extract_uri>[^\?]*)[\?(?P<extract_parameters>.*)]?')
|
2014-11-18 20:18:53 +01:00
|
|
|
pages_extensions = ['/', 'html', 'xhtml', 'py', 'pl', 'rb', 'php']
|
|
|
|
viewed_http_codes = [200]
|
|
|
|
|
2014-11-20 16:15:57 +01:00
|
|
|
HOOKS_ROOT = './plugins/'
|
|
|
|
PRE_HOOK_DIRECTORY = HOOKS_ROOT + 'pre_analysis/'
|
|
|
|
POST_HOOK_DIRECTORY = HOOKS_ROOT + 'post_analysis/'
|
|
|
|
DISPLAY_HOOK_DIRECTORY = HOOKS_ROOT + 'display/'
|
2014-11-19 19:34:16 +01:00
|
|
|
META_PATH = DB_ROOT + 'meta.db'
|
|
|
|
DB_FILENAME = 'iwla.db'
|
2014-11-19 08:01:12 +01:00
|
|
|
|
2014-11-20 16:15:57 +01:00
|
|
|
plugins = {PRE_HOOK_DIRECTORY : pre_analysis_hooks, POST_HOOK_DIRECTORY : post_analysis_hooks, DISPLAY_HOOK_DIRECTORY : display_hooks}
|
|
|
|
|
|
|
|
ANALYSIS_CLASS = 'HTTP'
|
|
|
|
API_VERSION = 1
|
|
|
|
|
|
|
|
def preloadPlugins():
|
|
|
|
for root in plugins.keys():
|
|
|
|
for plugin_name in plugins[root]:
|
|
|
|
p = root + '/' + plugin_name
|
|
|
|
try:
|
|
|
|
mod = cache_plugins[p] = imp.load_source('hook', p)
|
|
|
|
infos = mod.get_plugins_infos()
|
|
|
|
if infos['class'] != ANALYSIS_CLASS or \
|
|
|
|
API_VERSION < infos['min_version'] or\
|
|
|
|
(infos['max_version'] != -1 and (API_VERSION > infos['max_version'])):
|
|
|
|
del cache_plugins[p]
|
|
|
|
elif not mod.load():
|
|
|
|
del cache_plugins[p]
|
|
|
|
except Exception as e:
|
|
|
|
print 'Error loading \'%s\' => %s' % (p, e)
|
|
|
|
return False
|
|
|
|
return True
|
|
|
|
|
2014-11-18 20:18:53 +01:00
|
|
|
|
2014-11-19 19:45:41 +01:00
|
|
|
def createEmptyVisits():
|
|
|
|
visits = {'days_stats' : {}, 'month_stats' : {}, 'visits' : {}}
|
|
|
|
return visits
|
|
|
|
|
2014-11-19 21:37:37 +01:00
|
|
|
def createEmptyMeta():
|
2014-11-20 11:50:06 +01:00
|
|
|
meta = {'last_time' : None}
|
2014-11-19 21:37:37 +01:00
|
|
|
return meta
|
|
|
|
|
2014-11-20 11:50:06 +01:00
|
|
|
def createEmptyDisplay():
|
|
|
|
display = {}
|
|
|
|
return display
|
|
|
|
|
|
|
|
def createPage(filename, title):
|
|
|
|
page = {}
|
|
|
|
page['title'] = title;
|
|
|
|
page['blocks'] = []
|
|
|
|
display[filename] = page
|
|
|
|
|
|
|
|
return page
|
|
|
|
|
2014-11-19 19:45:41 +01:00
|
|
|
def getDBFilename(time):
|
2014-11-19 19:34:16 +01:00
|
|
|
return (DB_ROOT + '%d/%d_%s') % (time.tm_year, time.tm_mon, DB_FILENAME)
|
2014-11-19 08:01:12 +01:00
|
|
|
|
2014-11-19 19:34:16 +01:00
|
|
|
def serialize(obj, filename):
|
|
|
|
base = os.path.dirname(filename)
|
|
|
|
if not os.path.exists(base):
|
|
|
|
os.makedirs(base)
|
2014-11-19 08:01:12 +01:00
|
|
|
|
2014-11-20 08:18:31 +01:00
|
|
|
# TODO : remove return
|
|
|
|
return
|
|
|
|
|
2014-11-19 19:34:16 +01:00
|
|
|
with open(filename + '.tmp', 'wb+') as f:
|
|
|
|
pickle.dump(obj, f)
|
|
|
|
f.seek(0)
|
|
|
|
with gzip.open(filename, 'w') as fzip:
|
|
|
|
fzip.write(f.read())
|
|
|
|
os.remove(filename + '.tmp')
|
2014-11-19 08:01:12 +01:00
|
|
|
|
2014-11-19 19:34:16 +01:00
|
|
|
def deserialize(filename):
|
|
|
|
if not os.path.exists(filename):
|
|
|
|
return None
|
2014-11-19 08:01:12 +01:00
|
|
|
|
2014-11-19 19:34:16 +01:00
|
|
|
with gzip.open(filename, 'r') as f:
|
|
|
|
return pickle.load(f)
|
|
|
|
return None
|
2014-11-19 08:01:12 +01:00
|
|
|
|
2014-11-20 16:15:57 +01:00
|
|
|
def callPlugins(root, *kwargs):
|
|
|
|
print '==> Call plugins (%s)' % root
|
|
|
|
for p in plugins[root]:
|
2014-11-19 08:01:12 +01:00
|
|
|
print '\t%s' % (p)
|
2014-11-20 16:15:57 +01:00
|
|
|
mod = cache_plugins[root + '/' + p]
|
2014-11-19 08:01:12 +01:00
|
|
|
mod.hook(*kwargs)
|
|
|
|
|
2014-11-18 20:18:53 +01:00
|
|
|
def isPage(request):
|
|
|
|
for e in pages_extensions:
|
|
|
|
if request.endswith(e):
|
|
|
|
return True
|
|
|
|
|
|
|
|
return False
|
|
|
|
|
|
|
|
def appendHit(hit):
|
2014-11-20 08:18:31 +01:00
|
|
|
remote_addr = hit['remote_addr']
|
|
|
|
|
|
|
|
if not remote_addr in current_visits['visits'].keys():
|
|
|
|
createUser(hit)
|
|
|
|
return
|
|
|
|
|
|
|
|
super_hit = current_visits['visits'][remote_addr]
|
2014-11-18 20:18:53 +01:00
|
|
|
super_hit['pages'].append(hit)
|
|
|
|
super_hit['bandwith'] += int(hit['body_bytes_sent'])
|
2014-11-20 08:18:31 +01:00
|
|
|
super_hit['last_access'] = meta_visit['last_time']
|
|
|
|
|
2014-11-18 20:18:53 +01:00
|
|
|
request = hit['extract_request']
|
|
|
|
|
|
|
|
if 'extract_uri' in request.keys():
|
|
|
|
uri = request['extract_uri']
|
|
|
|
else:
|
|
|
|
uri = request['http_uri']
|
|
|
|
|
|
|
|
hit['is_page'] = isPage(uri)
|
|
|
|
|
2014-11-19 21:37:37 +01:00
|
|
|
# Don't count 3xx status
|
|
|
|
status = int(hit['status'])
|
|
|
|
if status >= 300 and status < 400: return
|
2014-11-18 20:18:53 +01:00
|
|
|
|
|
|
|
if super_hit['robot'] or\
|
2014-11-20 09:37:54 +01:00
|
|
|
not status in viewed_http_codes:
|
2014-11-18 20:18:53 +01:00
|
|
|
page_key = 'not_viewed_pages'
|
|
|
|
hit_key = 'not_viewed_hits'
|
|
|
|
else:
|
|
|
|
page_key = 'viewed_pages'
|
|
|
|
hit_key = 'viewed_hits'
|
|
|
|
|
|
|
|
if hit['is_page']:
|
|
|
|
super_hit[page_key] += 1
|
|
|
|
else:
|
|
|
|
super_hit[hit_key] += 1
|
|
|
|
|
2014-11-19 08:01:12 +01:00
|
|
|
def createUser(hit):
|
2014-11-19 21:37:37 +01:00
|
|
|
super_hit = current_visits['visits'][hit['remote_addr']] = {}
|
2014-11-18 20:18:53 +01:00
|
|
|
super_hit['viewed_pages'] = 0;
|
|
|
|
super_hit['viewed_hits'] = 0;
|
|
|
|
super_hit['not_viewed_pages'] = 0;
|
|
|
|
super_hit['not_viewed_hits'] = 0;
|
|
|
|
super_hit['bandwith'] = 0;
|
2014-11-20 08:18:31 +01:00
|
|
|
super_hit['last_access'] = meta_visit['last_time']
|
2014-11-18 20:18:53 +01:00
|
|
|
super_hit['pages'] = [];
|
2014-11-20 16:15:57 +01:00
|
|
|
super_hit['robot'] = False
|
2014-11-20 14:09:01 +01:00
|
|
|
super_hit['hit_only'] = 0;
|
2014-11-18 20:18:53 +01:00
|
|
|
appendHit(hit)
|
|
|
|
|
2014-11-19 19:45:41 +01:00
|
|
|
def decodeHTTPRequest(hit):
|
2014-11-18 20:18:53 +01:00
|
|
|
if not 'request' in hit.keys(): return False
|
|
|
|
|
|
|
|
groups = http_request_extracted.match(hit['request'])
|
|
|
|
|
|
|
|
if groups:
|
|
|
|
hit['extract_request'] = groups.groupdict()
|
|
|
|
uri_groups = uri_re.match(hit['extract_request']['http_uri']);
|
|
|
|
if uri_groups:
|
2014-11-19 08:01:12 +01:00
|
|
|
d = uri_groups.groupdict()
|
|
|
|
hit['extract_request']['extract_uri'] = d['extract_uri']
|
|
|
|
if 'extract_parameters' in d.keys():
|
|
|
|
hit['extract_request']['extract_parameters'] = d['extract_parameters']
|
2014-11-18 20:18:53 +01:00
|
|
|
else:
|
|
|
|
print "Bad request extraction " + hit['request']
|
|
|
|
return False
|
|
|
|
|
|
|
|
referer_groups = uri_re.match(hit['http_referer']);
|
|
|
|
if referer_groups:
|
2014-11-19 08:01:12 +01:00
|
|
|
referer = hit['extract_referer'] = referer_groups.groupdict()
|
2014-11-18 20:18:53 +01:00
|
|
|
return True
|
|
|
|
|
2014-11-19 19:45:41 +01:00
|
|
|
def decodeTime(hit):
|
2014-11-18 20:18:53 +01:00
|
|
|
t = hit['time_local']
|
|
|
|
|
|
|
|
hit['time_decoded'] = time.strptime(t, time_format)
|
|
|
|
|
2014-11-20 11:50:06 +01:00
|
|
|
def buildPages():
|
|
|
|
for filename in display.keys():
|
|
|
|
page = display[filename]
|
|
|
|
with open(DISPLAY_ROOT + filename, 'w') as f:
|
|
|
|
f.write('<html><title>%s</title><body>' % (page['title']))
|
|
|
|
for block in page['blocks']:
|
|
|
|
if block['type'] == 'html':
|
|
|
|
f.write(block['value'])
|
|
|
|
elif block['type'] == 'table':
|
|
|
|
f.write('<table>')
|
|
|
|
f.write('<tr>')
|
|
|
|
for title in block['cols']:
|
|
|
|
f.write('<th>%s</th>' % (title))
|
|
|
|
f.write('</tr>')
|
|
|
|
for row in block['rows']:
|
|
|
|
f.write('<tr>')
|
|
|
|
for v in row:
|
|
|
|
f.write('<td>%s</td>' % (v))
|
|
|
|
f.write('</tr>')
|
|
|
|
f.write('</table>')
|
|
|
|
f.write('</body></html>')
|
|
|
|
|
|
|
|
def generateDisplayDaysStat():
|
|
|
|
cur_time = meta_visit['last_time']
|
|
|
|
title = 'Stats %d/%d' % (cur_time.tm_mon, cur_time.tm_year)
|
|
|
|
filename = '%d/index_%d.html' % (cur_time.tm_year, cur_time.tm_mon)
|
|
|
|
page = createPage(filename, title)
|
|
|
|
|
|
|
|
days = {'type' : 'table', 'title' : 'By day'}
|
|
|
|
days['cols'] = ['Day', 'Visits', 'Pages', 'Hits', 'Bandwith', 'Robot Bandwith']
|
|
|
|
days['rows'] = []
|
|
|
|
keys = current_visits['days_stats'].keys()
|
|
|
|
keys.sort()
|
|
|
|
nb_visits = 0
|
|
|
|
for k in keys:
|
|
|
|
stats = current_visits['days_stats'][k]
|
|
|
|
row = [k, stats['nb_visitors'], stats['viewed_pages'], stats['viewed_hits'], stats['viewed_bandwidth'], stats['not_viewed_bandwidth']]
|
|
|
|
row = map(lambda(v): str(v), row)
|
|
|
|
days['rows'].append(row)
|
|
|
|
nb_visits += stats['nb_visitors']
|
|
|
|
|
|
|
|
stats = current_visits['month_stats']
|
|
|
|
|
|
|
|
nb_days = len(keys)
|
|
|
|
row = [0, nb_visits, stats['viewed_pages'], stats['viewed_hits'], stats['viewed_bandwidth'], stats['not_viewed_bandwidth']]
|
|
|
|
if nb_days:
|
|
|
|
row = map(lambda(v): str(int(v/nb_days)), row)
|
|
|
|
else:
|
|
|
|
row = map(lambda(v): '0', row)
|
|
|
|
|
|
|
|
row[0] = 'Average'
|
|
|
|
days['rows'].append(row)
|
|
|
|
|
|
|
|
row = ['Total', nb_visits, stats['viewed_pages'], stats['viewed_hits'], stats['viewed_bandwidth'], stats['not_viewed_bandwidth']]
|
|
|
|
row = map(lambda(v): str(v), row)
|
|
|
|
days['rows'].append(row)
|
|
|
|
page['blocks'].append(days)
|
|
|
|
|
|
|
|
def generateDisplay():
|
|
|
|
generateDisplayDaysStat()
|
|
|
|
callPlugins(DISPLAY_HOOK_DIRECTORY, current_visits, display)
|
|
|
|
buildPages()
|
2014-11-18 20:18:53 +01:00
|
|
|
|
2014-11-19 21:37:37 +01:00
|
|
|
def generateStats(visits):
|
2014-11-19 19:34:16 +01:00
|
|
|
stats = {}
|
|
|
|
stats['viewed_bandwidth'] = 0
|
|
|
|
stats['not_viewed_bandwidth'] = 0
|
|
|
|
stats['viewed_pages'] = 0
|
|
|
|
stats['viewed_hits'] = 0
|
2014-11-19 21:37:37 +01:00
|
|
|
#stats['pages'] = set()
|
|
|
|
stats['nb_visitors'] = 0
|
2014-11-19 19:34:16 +01:00
|
|
|
|
2014-11-19 21:37:37 +01:00
|
|
|
for k in visits.keys():
|
|
|
|
super_hit = visits[k]
|
2014-11-19 19:34:16 +01:00
|
|
|
if super_hit['robot']:
|
|
|
|
stats['not_viewed_bandwidth'] += super_hit['bandwith']
|
|
|
|
continue
|
|
|
|
|
2014-11-19 21:37:37 +01:00
|
|
|
print "[%s] =>\t%d/%d" % (k, super_hit['viewed_pages'], super_hit['viewed_hits'])
|
|
|
|
|
2014-11-20 14:09:01 +01:00
|
|
|
if not super_hit['hit_only']:
|
|
|
|
stats['nb_visitors'] += 1
|
2014-11-19 19:34:16 +01:00
|
|
|
stats['viewed_bandwidth'] += super_hit['bandwith']
|
|
|
|
stats['viewed_pages'] += super_hit['viewed_pages']
|
|
|
|
stats['viewed_hits'] += super_hit['viewed_hits']
|
|
|
|
|
2014-11-19 21:37:37 +01:00
|
|
|
# for p in super_hit['pages']:
|
|
|
|
# if not p['is_page']: continue
|
|
|
|
# req = p['extract_request']
|
|
|
|
# stats['pages'].add(req['extract_uri'])
|
2014-11-19 19:34:16 +01:00
|
|
|
|
2014-11-19 21:37:37 +01:00
|
|
|
return stats
|
|
|
|
|
|
|
|
def generateMonthStats():
|
2014-11-20 11:50:06 +01:00
|
|
|
display = createEmptyDisplay()
|
|
|
|
|
2014-11-19 21:37:37 +01:00
|
|
|
visits = current_visits['visits']
|
|
|
|
|
|
|
|
stats = generateStats(visits)
|
|
|
|
|
2014-11-19 19:34:16 +01:00
|
|
|
cur_time = meta_visit['last_time']
|
|
|
|
print "== Stats for %d/%d ==" % (cur_time.tm_year, cur_time.tm_mon)
|
|
|
|
print stats
|
|
|
|
|
2014-11-19 21:37:37 +01:00
|
|
|
valid_visitors = {k: v for (k,v) in visits.items() if not visits[k]['robot']}
|
|
|
|
callPlugins(POST_HOOK_DIRECTORY, valid_visitors)
|
|
|
|
|
|
|
|
current_visits['month_stats'] = stats
|
|
|
|
|
2014-11-19 19:45:41 +01:00
|
|
|
path = getDBFilename(cur_time)
|
2014-11-19 19:34:16 +01:00
|
|
|
if os.path.exists(path):
|
|
|
|
os.remove(path)
|
|
|
|
|
|
|
|
print "==> Serialize to %s" % path
|
|
|
|
|
2014-11-19 19:45:41 +01:00
|
|
|
serialize(current_visits, path)
|
2014-11-19 19:34:16 +01:00
|
|
|
|
2014-11-20 11:50:06 +01:00
|
|
|
generateDisplay()
|
|
|
|
|
2014-11-19 21:37:37 +01:00
|
|
|
def generateDayStats():
|
|
|
|
visits = current_visits['visits']
|
|
|
|
|
|
|
|
callPlugins(PRE_HOOK_DIRECTORY, visits)
|
|
|
|
|
|
|
|
stats = generateStats(visits)
|
|
|
|
|
|
|
|
cur_time = meta_visit['last_time']
|
|
|
|
print "== Stats for %d/%d/%d ==" % (cur_time.tm_year, cur_time.tm_mon, cur_time.tm_mday)
|
|
|
|
|
|
|
|
if cur_time.tm_mday > 1:
|
|
|
|
last_day = cur_time.tm_mday - 1
|
|
|
|
while last_day:
|
|
|
|
if last_day in current_visits['days_stats'].keys():
|
|
|
|
break
|
|
|
|
last_day -= 1
|
|
|
|
if last_day:
|
|
|
|
for k in stats.keys():
|
|
|
|
stats[k] -= current_visits['days_stats'][last_day][k]
|
2014-11-20 08:18:31 +01:00
|
|
|
stats['nb_visitors'] = 0
|
|
|
|
for k in visits.keys():
|
|
|
|
if visits[k]['robot']: continue
|
|
|
|
if visits[k]['last_access'].tm_mday == cur_time.tm_mday:
|
|
|
|
stats['nb_visitors'] += 1
|
2014-11-19 21:37:37 +01:00
|
|
|
print stats
|
|
|
|
|
|
|
|
current_visits['days_stats'][cur_time.tm_mday] = stats
|
|
|
|
|
2014-11-19 19:34:16 +01:00
|
|
|
def newHit(hit):
|
2014-11-19 19:45:41 +01:00
|
|
|
global current_visits
|
2014-11-19 19:34:16 +01:00
|
|
|
global analyse_started
|
2014-11-18 20:18:53 +01:00
|
|
|
|
2014-11-19 19:45:41 +01:00
|
|
|
decodeTime(hit)
|
2014-11-18 20:18:53 +01:00
|
|
|
|
|
|
|
t = hit['time_decoded']
|
|
|
|
|
2014-11-19 19:34:16 +01:00
|
|
|
cur_time = meta_visit['last_time']
|
2014-11-18 20:18:53 +01:00
|
|
|
|
|
|
|
if cur_time == None:
|
2014-11-20 09:37:54 +01:00
|
|
|
current_visits = deserialize(getDBFilename(t)) or createEmptyVisits()
|
2014-11-19 19:34:16 +01:00
|
|
|
analyse_started = True
|
2014-11-18 20:18:53 +01:00
|
|
|
else:
|
2014-11-19 19:34:16 +01:00
|
|
|
if not analyse_started:
|
|
|
|
if time.mktime(cur_time) >= time.mktime(t):
|
|
|
|
return
|
|
|
|
else:
|
|
|
|
analyse_started = True
|
2014-11-20 09:37:54 +01:00
|
|
|
current_visits = deserialize(getDBFilename(t)) or createEmptyVisits()
|
2014-11-19 19:34:16 +01:00
|
|
|
if cur_time.tm_mon != t.tm_mon:
|
2014-11-19 19:45:41 +01:00
|
|
|
generateMonthStats()
|
2014-11-20 09:37:54 +01:00
|
|
|
current_visits = deserialize(getDBFilename(t)) or createEmptyVisits()
|
2014-11-19 21:37:37 +01:00
|
|
|
elif cur_time.tm_mday != t.tm_mday:
|
|
|
|
generateDayStats()
|
2014-11-19 19:34:16 +01:00
|
|
|
|
|
|
|
meta_visit['last_time'] = t
|
|
|
|
|
2014-11-19 19:45:41 +01:00
|
|
|
if not decodeHTTPRequest(hit): return False
|
2014-11-19 19:34:16 +01:00
|
|
|
|
|
|
|
for k in hit.keys():
|
|
|
|
if hit[k] == '-': hit[k] = ''
|
2014-11-18 20:18:53 +01:00
|
|
|
|
2014-11-20 08:18:31 +01:00
|
|
|
appendHit(hit)
|
2014-11-18 20:18:53 +01:00
|
|
|
|
|
|
|
return True
|
|
|
|
|
2014-11-20 16:15:57 +01:00
|
|
|
preloadPlugins()
|
|
|
|
|
2014-11-18 20:18:53 +01:00
|
|
|
print '==> Analysing log'
|
2014-11-19 19:34:16 +01:00
|
|
|
|
2014-11-20 09:37:54 +01:00
|
|
|
meta_visit = deserialize(META_PATH) or createEmptyMeta()
|
2014-11-19 21:37:37 +01:00
|
|
|
|
|
|
|
current_visits = createEmptyVisits()
|
2014-11-19 19:34:16 +01:00
|
|
|
|
2014-11-20 15:25:43 +01:00
|
|
|
f = open(analyzed_filename)
|
2014-11-18 20:18:53 +01:00
|
|
|
for l in f:
|
|
|
|
# print "line " + l;
|
|
|
|
|
|
|
|
groups = log_re.match(l)
|
|
|
|
|
|
|
|
if groups:
|
|
|
|
if not newHit(groups.groupdict()):
|
|
|
|
break
|
|
|
|
else:
|
|
|
|
print "No match " + l
|
|
|
|
f.close();
|
|
|
|
|
2014-11-19 19:45:41 +01:00
|
|
|
if analyse_started:
|
2014-11-20 08:18:31 +01:00
|
|
|
generateDayStats()
|
2014-11-19 19:45:41 +01:00
|
|
|
generateMonthStats()
|
|
|
|
serialize(meta_visit, META_PATH)
|
|
|
|
else:
|
|
|
|
print '==> Analyse not started : nothing to do'
|