import time
import traceback
import glob
+import json
from optparse import OptionParser
import paramiko
import socket
import logging
-import common, metadata
+import common
+import metadata
import subprocess
from collections import Counter
+
def carbon_send(key, value):
s = socket.socket()
s.connect((config['carbon_host'], config['carbon_port']))
options = None
config = None
+
def main():
global options, config
help="Restrict output to warnings and errors")
parser.add_option("-d", "--download", action="store_true", default=False,
help="Download logs we don't have")
+ parser.add_option("--recalc", action="store_true", default=False,
+ help="Recalculate aggregate stats - use when changes "
+ "have been made that would invalidate old cached data.")
parser.add_option("--nologs", action="store_true", default=False,
help="Don't do anything logs-related")
(options, args) = parser.parse_args()
sys.exit(1)
# Get all metadata-defined apps...
- metaapps = metadata.read_metadata(options.verbose)
+ metaapps = [a for a in metadata.read_metadata().itervalues() if not a['Disabled']]
statsdir = 'stats'
logsdir = os.path.join(statsdir, 'logs')
logging.info('Retrieving logs')
ssh = paramiko.SSHClient()
ssh.load_system_host_keys()
- ssh.connect('f-droid.org', username='fdroid', timeout=10,
- key_filename=config['webserver_keyfile'])
+ ssh.connect(config['stats_server'], username=config['stats_user'],
+ timeout=10, key_filename=config['webserver_keyfile'])
ftp = ssh.open_sftp()
ftp.get_channel().settimeout(60)
logging.info("...connected")
traceback.print_exc()
sys.exit(1)
finally:
- #Disconnect
+ # Disconnect
if ftp is not None:
ftp.close()
if ssh is not None:
logging.info('Processing logs...')
appscount = Counter()
appsvercount = Counter()
- logexpr = '(?P<ip>[.:0-9a-fA-F]+) - - \[(?P<time>.*?)\] "GET (?P<uri>.*?) HTTP/1.\d" (?P<statuscode>\d+) \d+ "(?P<referral>.*?)" "(?P<useragent>.*?)"'
+ logexpr = '(?P<ip>[.:0-9a-fA-F]+) - - \[(?P<time>.*?)\] ' + \
+ '"GET (?P<uri>.*?) HTTP/1.\d" (?P<statuscode>\d+) ' + \
+ '\d+ "(?P<referral>.*?)" "(?P<useragent>.*?)"'
logsearch = re.compile(logexpr).search
- for logfile in glob.glob(os.path.join(logsdir,'access-*.log.gz')):
+ for logfile in glob.glob(os.path.join(logsdir, 'access-*.log.gz')):
logging.debug('...' + logfile)
- if options.verbose:
- print '...' + logfile
- p = subprocess.Popen(["zcat", logfile], stdout = subprocess.PIPE)
- matches = (logsearch(line) for line in p.stdout)
- for match in matches:
- if match and match.group('statuscode') == '200':
+
+ # Get the date for this log - e.g. 2012-02-28
+ thisdate = os.path.basename(logfile)[7:-7]
+
+ agg_path = os.path.join(datadir, thisdate + '.json')
+ if not options.recalc and os.path.exists(agg_path):
+ # Use previously calculated aggregate data
+ with open(agg_path, 'r') as f:
+ today = json.load(f)
+
+ else:
+ # Calculate from logs...
+
+ today = {
+ 'apps': Counter(),
+ 'appsver': Counter(),
+ 'unknown': []
+ }
+
+ p = subprocess.Popen(["zcat", logfile], stdout=subprocess.PIPE)
+ matches = (logsearch(line) for line in p.stdout)
+ for match in matches:
+ if not match:
+ continue
+ if match.group('statuscode') != '200':
+ continue
+ if match.group('ip') in config['stats_ignore']:
+ continue
uri = match.group('uri')
- if uri.endswith('.apk'):
- _, apkname = os.path.split(uri)
- app = knownapks.getapp(apkname)
- if app:
- appid, _ = app
- appscount[appid] += 1
- # Strip the '.apk' from apkname
- appver = apkname[:-4]
- appsvercount[appver] += 1
- else:
- if not apkname in unknownapks:
- unknownapks.append(apkname)
+ if not uri.endswith('.apk'):
+ continue
+ _, apkname = os.path.split(uri)
+ app = knownapks.getapp(apkname)
+ if app:
+ appid, _ = app
+ today['apps'][appid] += 1
+ # Strip the '.apk' from apkname
+ appver = apkname[:-4]
+ today['appsver'][appver] += 1
+ else:
+ if apkname not in today['unknown']:
+ today['unknown'].append(apkname)
+
+ # Save calculated aggregate data for today to cache
+ with open(agg_path, 'w') as f:
+ json.dump(today, f)
+
+ # Add today's stats (whether cached or recalculated) to the total
+ for appid in today['apps']:
+ appscount[appid] += today['apps'][appid]
+ for appid in today['appsver']:
+ appsvercount[appid] += today['appsver'][appid]
+ for uk in today['unknown']:
+ if uk not in unknownapks:
+ unknownapks.append(uk)
# Calculate and write stats for total downloads...
lst = []
alldownloads = 0
for appid in appscount:
count = appscount[appid]
- lst.append(app + " " + str(count))
+ lst.append(appid + " " + str(count))
if config['stats_to_carbon']:
- carbon_send('fdroid.download.' + app.replace('.', '_'), count)
+ carbon_send('fdroid.download.' + appid.replace('.', '_'),
+ count)
alldownloads += count
lst.append("ALL " + str(alldownloads))
f = open('stats/total_downloads_app.txt', 'w')
f.close()
f = open('stats/total_downloads_app_version.txt', 'w')
- f.write('# Total downloads by application and version, since October 2011\n')
+ f.write('# Total downloads by application and version, '
+ 'since October 2011\n')
lst = []
for appver in appsvercount:
count = appsvercount[appver]
logging.info("Processing repo types...")
repotypes = Counter()
for app in metaapps:
- if len(app['Repo Type']) == 0:
- rtype = 'none'
- else:
- if app['Repo Type'] == 'srclib':
- rtype = common.getsrclibvcs(app['Repo'])
- else:
- rtype = app['Repo Type']
+ rtype = app['Repo Type'] or 'none'
+ if rtype == 'srclib':
+ rtype = common.getsrclibvcs(app['Repo'])
repotypes[rtype] += 1
f = open('stats/repotypes.txt', 'w')
for rtype in repotypes:
checkmode = checkmode[:12]
if checkmode.startswith('Tags '):
checkmode = checkmode[:4]
- ucms[checkmode] += 1;
+ ucms[checkmode] += 1
f = open('stats/update_check_modes.txt', 'w')
for checkmode in ucms:
count = ucms[checkmode]
ctgs = Counter()
for app in metaapps:
for category in app['Categories']:
- ctgs[category] += 1;
+ ctgs[category] += 1
f = open('stats/categories.txt', 'w')
for category in ctgs:
count = ctgs[category]
continue
antifeatures = [a.strip() for a in app['AntiFeatures'].split(',')]
for antifeature in antifeatures:
- afs[antifeature] += 1;
+ afs[antifeature] += 1
f = open('stats/antifeatures.txt', 'w')
for antifeature in afs:
count = afs[antifeature]
licenses = Counter()
for app in metaapps:
license = app['License']
- licenses[license] += 1;
+ licenses[license] += 1
f = open('stats/licenses.txt', 'w')
for license in licenses:
count = licenses[license]
if __name__ == "__main__":
main()
-
~ianmdlvl / fdroidserver.git / blobdiff