X-Git-Url: https://git.decadent.org.uk/gitweb/?p=dak.git;a=blobdiff_plain;f=dak%2Fstats.py;h=6f69e1a7f4239429fac841b78cdb47c92605a278;hp=a11b902eaa59e287a3d39405d80fe7b7d5932aa1;hb=519c1dbf89c13557afc15a429164616ac563d379;hpb=59fd5aa2a8be3b76dbc968429c457f096adfa472 diff --git a/dak/stats.py b/dak/stats.py index a11b902e..6f69e1a7 100755 --- a/dak/stats.py +++ b/dak/stats.py @@ -1,8 +1,8 @@ #!/usr/bin/env python -# Various statistical pr0nography fun and games -# Copyright (C) 2000, 2001, 2002, 2003 James Troup -# $Id: saffron,v 1.3 2005-11-15 09:50:32 ajt Exp $ +""" Various statistical pr0nography fun and games """ +# Copyright (C) 2000, 2001, 2002, 2003, 2006 James Troup +# Copyright (C) 2013 Luca Falavigna # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by @@ -31,223 +31,400 @@ ################################################################################ -import pg, sys; -import utils; -import apt_pkg; +import sys +import apt_pkg + +from datetime import datetime +from email.utils import mktime_tz, parsedate_tz +from mailbox import mbox +from os import listdir, system, unlink +from os.path import isfile, join, splitext +from re import findall, DOTALL, MULTILINE +from sys import stderr +from yaml import safe_load, safe_dump + +from daklib import utils +from daklib.dbconn import DBConn, get_suite_architectures, Suite, Architecture ################################################################################ -Cnf = None; -projectB = None; +Cnf = None + +stats = {} +users = {} +buffer = 0 +FORMAT_SWITCH = '2009-08' +blacklisted = ('dak', 'katie') + +NEW = ('^(\d{14})\|(?:jennifer|process-unchecked|.*?\|dak)' + '\|(Moving to new|ACCEPT-TO-NEW)') +new_ACTIONS = '^(\d{14})\|[^\|]*\|(\S+)\|NEW (\S+)[:\|]' +old_ACTIONS = ('(?:lisa|process-new)\|program start\|(.*?)\|' + '(?:lisa|process-new)\|program end') +old_ACTION = '^(\d{14})\|(?:lisa|process-new)\|(Accepting changes|rejected)\|' ################################################################################ def usage(exit_code=0): - print """Usage: saffron STAT + print """Usage: dak stats MODE Print various stats. -h, --help show this help and exit. -The following STAT modes are available: +The following MODEs are available: arch-space - displays space used by each architecture pkg-nums - displays the number of packages by suite/architecture daily-install - displays daily install stats suitable for graphing + new - stores stats about the NEW queue """ sys.exit(exit_code) ################################################################################ def per_arch_space_use(): - q = projectB.query(""" -SELECT a.arch_string as Architecture, sum(f.size) + session = DBConn().session() + q = session.execute(""" +SELECT a.arch_string as Architecture, sum(f.size) AS sum FROM files f, binaries b, architecture a WHERE a.id=b.architecture AND f.id=b.file - GROUP BY a.arch_string"""); - print q; - q = projectB.query("SELECT sum(size) FROM files WHERE filename ~ '.(diff.gz|tar.gz|dsc)$'"); - print q; + GROUP BY a.arch_string ORDER BY sum""").fetchall() + for j in q: + print "%-15.15s %s" % (j[0], j[1]) + print + q = session.execute("SELECT sum(size) FROM files WHERE filename ~ '.(diff.gz|tar.gz|dsc)$'").fetchall() + print "%-15.15s %s" % ("Source", q[0][0]) ################################################################################ def daily_install_stats(): - stats = {}; - file = utils.open_file("2001-11"); - for line in file.readlines(): - split = line.strip().split('~'); - program = split[1]; - if program != "katie": - continue; - action = split[2]; + stats = {} + f = utils.open_file("2001-11") + for line in f.readlines(): + split = line.strip().split('|') + program = split[1] + if program != "katie" and program != "process-accepted": + continue + action = split[2] if action != "installing changes" and action != "installed": - continue; - date = split[0][:8]; + continue + date = split[0][:8] if not stats.has_key(date): - stats[date] = {}; - stats[date]["packages"] = 0; - stats[date]["size"] = 0.0; + stats[date] = {} + stats[date]["packages"] = 0 + stats[date]["size"] = 0.0 if action == "installing changes": - stats[date]["packages"] += 1; + stats[date]["packages"] += 1 elif action == "installed": - stats[date]["size"] += float(split[5]); + stats[date]["size"] += float(split[5]) - dates = stats.keys(); - dates.sort(); + dates = stats.keys() + dates.sort() for date in dates: packages = stats[date]["packages"] size = int(stats[date]["size"] / 1024.0 / 1024.0) - print "%s %s %s" % (date, packages, size); + print "%s %s %s" % (date, packages, size) ################################################################################ def longest(list): - longest = 0; + longest = 0 for i in list: - l = len(i); + l = len(i) if l > longest: - longest = l; - return longest; - -def suite_sort(a, b): - if Cnf.has_key("Suite::%s::Priority" % (a)): - a_priority = int(Cnf["Suite::%s::Priority" % (a)]); - else: - a_priority = 0; - if Cnf.has_key("Suite::%s::Priority" % (b)): - b_priority = int(Cnf["Suite::%s::Priority" % (b)]); - else: - b_priority = 0; - return cmp(a_priority, b_priority); + longest = l + return longest def output_format(suite): - output_suite = []; + output_suite = [] for word in suite.split("-"): - output_suite.append(word[0]); - return "-".join(output_suite); - -# Obvious query with GROUP BY and mapped names -> 50 seconds -# GROUP BY but ids instead of suite/architecture names -> 28 seconds -# Simple query -> 14 seconds -# Simple query into large dictionary + processing -> 21 seconds -# Simple query into large pre-created dictionary + processing -> 18 seconds + output_suite.append(word[0]) + return "-".join(output_suite) def number_of_packages(): - arches = {}; - arch_ids = {}; - suites = {}; - suite_ids = {}; - d = {}; + arches = {} + arch_ids = {} + suites = {} + suite_ids = {} + d = {} + session = DBConn().session() # Build up suite mapping - q = projectB.query("SELECT id, suite_name FROM suite"); - suite_ql = q.getresult(); - for i in suite_ql: - (id, name) = i; - suites[id] = name; - suite_ids[name] = id; + for i in session.query(Suite).all(): + suites[i.suite_id] = i.suite_name + suite_ids[i.suite_name] = i.suite_id # Build up architecture mapping - q = projectB.query("SELECT id, arch_string FROM architecture"); - for i in q.getresult(): - (id, name) = i; - arches[id] = name; - arch_ids[name] = id; + for i in session.query(Architecture).all(): + arches[i.arch_id] = i.arch_string + arch_ids[i.arch_string] = i.arch_id # Pre-create the dictionary for suite_id in suites.keys(): - d[suite_id] = {}; + d[suite_id] = {} for arch_id in arches.keys(): - d[suite_id][arch_id] = 0; + d[suite_id][arch_id] = 0 # Get the raw data for binaries - q = projectB.query(""" -SELECT ba.suite, b.architecture - FROM binaries b, bin_associations ba - WHERE b.id = ba.bin"""); # Simultate 'GROUP by suite, architecture' with a dictionary - for i in q.getresult(): - (suite_id, arch_id) = i; - d[suite_id][arch_id] = d[suite_id][arch_id] + 1; + # XXX: Why don't we just get the DB to do this? + for i in session.execute("""SELECT suite, architecture, COUNT(suite) + FROM bin_associations + LEFT JOIN binaries ON bin = binaries.id + GROUP BY suite, architecture""").fetchall(): + d[ i[0] ][ i[1] ] = i[2] # Get the raw data for source - arch_id = arch_ids["source"]; - q = projectB.query(""" -SELECT suite, count(suite) FROM src_associations GROUP BY suite;"""); - for i in q.getresult(): - (suite_id, count) = i; - d[suite_id][arch_id] = d[suite_id][arch_id] + count; + arch_id = arch_ids["source"] + for i in session.execute('SELECT suite, COUNT(suite) FROM src_associations GROUP BY suite').fetchall(): + (suite_id, count) = i + d[suite_id][arch_id] = d[suite_id][arch_id] + count ## Print the results # Setup - suite_list = suites.values(); - suite_list.sort(suite_sort); - suite_id_list = []; - suite_arches = {}; + suite_list = suites.values() + suite_id_list = [] + suite_arches = {} for suite in suite_list: - suite_id = suite_ids[suite]; - suite_arches[suite_id] = {}; - for arch in Cnf.ValueList("Suite::%s::Architectures" % (suite)): - suite_arches[suite_id][arch] = ""; - suite_id_list.append(suite_id); - output_list = map(lambda x: output_format(x), suite_list); - longest_suite = longest(output_list); - arch_list = arches.values(); - arch_list.sort(); - longest_arch = longest(arch_list); + suite_id = suite_ids[suite] + suite_arches[suite_id] = {} + for arch in get_suite_architectures(suite): + suite_arches[suite_id][arch.arch_string] = "" + suite_id_list.append(suite_id) + output_list = [ output_format(i) for i in suite_list ] + longest_suite = longest(output_list) + arch_list = arches.values() + arch_list.sort() + longest_arch = longest(arch_list) # Header output = (" "*longest_arch) + " |" for suite in output_list: - output = output + suite.center(longest_suite)+" |"; - output = output + "\n"+(len(output)*"-")+"\n"; + output = output + suite.center(longest_suite)+" |" + output = output + "\n"+(len(output)*"-")+"\n" # per-arch data - arch_list = arches.values(); - arch_list.sort(); - longest_arch = longest(arch_list); + arch_list = arches.values() + arch_list.sort() + longest_arch = longest(arch_list) for arch in arch_list: - arch_id = arch_ids[arch]; - output = output + arch.center(longest_arch)+" |"; + arch_id = arch_ids[arch] + output = output + arch.center(longest_arch)+" |" for suite_id in suite_id_list: if suite_arches[suite_id].has_key(arch): - count = repr(d[suite_id][arch_id]); + count = "%d" % d[suite_id][arch_id] else: - count = "-"; - output = output + count.rjust(longest_suite)+" |"; - output = output + "\n"; - print output; + count = "-" + output = output + count.rjust(longest_suite)+" |" + output = output + "\n" + print output + +################################################################################ + +def parse_new_uploads(data): + global stats + latest_timestamp = stats['timestamp'] + for entry in findall(NEW, data, MULTILINE): + timestamp = entry[0] + if stats['timestamp'] >= timestamp: + continue + date = parse_timestamp(timestamp) + if date not in stats: + stats[date] = {'stats': {'NEW': 0, 'ACCEPT': 0, + 'REJECT': 0, 'PROD': 0}, 'members': {}} + stats[date]['stats']['NEW'] += 1 + stats['history']['stats']['NEW'] += 1 + latest_timestamp = timestamp + return latest_timestamp + + +def parse_actions(data, logdate): + global stats + latest_timestamp = stats['timestamp'] + if logdate <= FORMAT_SWITCH: + for batch in findall(old_ACTIONS, data, DOTALL): + who = batch.split()[0] + if who in blacklisted: + continue + for entry in findall(old_ACTION, batch, MULTILINE): + action = entry[1] + if action.startswith('Accepting'): + action = 'ACCEPT' + elif action.startswith('rejected'): + action = 'REJECT' + timestamp = entry[0] + if stats['timestamp'] >= timestamp: + continue + date = parse_timestamp(entry[0]) + if date not in stats: + stats[date] = {'stats': {'NEW': 0, 'ACCEPT': 0, + 'REJECT': 0, 'PROD': 0}, 'members': {}} + stats[date]['stats'][action] += 1 + stats['history']['stats'][action] += 1 + if who not in stats[date]['members']: + stats[date]['members'][who] = {'ACCEPT': 0, 'REJECT': 0, + 'PROD': 0} + stats[date]['members'][who][action] += 1 + if who not in stats['history']['members']: + stats['history']['members'][who] = {'ACCEPT': 0, 'REJECT': 0, + 'PROD': 0} + stats['history']['members'][who][action] += 1 + latest_timestamp = timestamp + parse_prod(logdate) + if logdate >= FORMAT_SWITCH: + for entry in findall(new_ACTIONS, data, MULTILINE): + action = entry[2] + timestamp = entry[0] + if stats['timestamp'] >= timestamp: + continue + date = parse_timestamp(timestamp) + if date not in stats: + stats[date] = {'stats': {'NEW': 0, 'ACCEPT': 0, + 'REJECT': 0, 'PROD': 0}, 'members': {}} + member = entry[1] + if member in blacklisted: + continue + if date not in stats: + stats[date] = {'stats': {'NEW': 0, 'ACCEPT': 0, + 'REJECT': 0, 'PROD': 0}, 'members': {}} + if member not in stats[date]['members']: + stats[date]['members'][member] = {'ACCEPT': 0, 'REJECT': 0, + 'PROD': 0} + if member not in stats['history']['members']: + stats['history']['members'][member] = {'ACCEPT': 0, + 'REJECT': 0, 'PROD': 0} + stats[date]['stats'][action] += 1 + stats[date]['members'][member][action] += 1 + stats['history']['stats'][action] += 1 + stats['history']['members'][member][action] += 1 + latest_timestamp = timestamp + return latest_timestamp + + +def parse_prod(logdate): + global stats + global users + maildate = ''.join([x[-2:] for x in logdate.split('-')]) + mailarchive = join(utils.get_conf()['Dir::Base'], 'mail/archive', + 'mail-%s.xz' % maildate) + if not isfile(mailarchive): + return + (fd, tmpfile) = utils.temp_filename(utils.get_conf()['Dir::TempPath']) + system('xzcat %s > %s' % (mailarchive, tmpfile)) + for message in mbox(tmpfile): + if (message['subject'] and + message['subject'].startswith('Comments regarding')): + try: + member = users[' '.join(message['From'].split()[:-1])] + except KeyError: + continue + ts = mktime_tz(parsedate_tz(message['date'])) + timestamp = datetime.fromtimestamp(ts).strftime("%Y%m%d%H%M%S") + date = parse_timestamp(timestamp) + if date not in stats: + stats[date] = {'stats': {'NEW': 0, 'ACCEPT': 0, + 'REJECT': 0, 'PROD': 0}, 'members': {}} + if member not in stats[date]['members']: + stats[date]['members'][member] = {'ACCEPT': 0, 'REJECT': 0, + 'PROD': 0} + if member not in stats['history']['members']: + stats['history']['members'][member] = {'ACCEPT': 0, + 'REJECT': 0, 'PROD': 0} + stats[date]['stats']['PROD'] += 1 + stats[date]['members'][member]['PROD'] += 1 + stats['history']['stats']['PROD'] += 1 + stats['history']['members'][member]['PROD'] += 1 + unlink(tmpfile) + + +def parse_timestamp(timestamp): + y = int(timestamp[:4]) + m = int(timestamp[4:6]) + return '%d-%02d' % (y, m) + + +def new_stats(logdir, yaml): + global Cnf + global stats + try: + with open(yaml, 'r') as fd: + stats = safe_load(fd) + except IOError: + pass + if not stats: + stats = {'history': {'stats': {'NEW': 0, 'ACCEPT': 0, + 'REJECT': 0, 'PROD': 0}, 'members': {}}, + 'timestamp': '19700101000000'} + latest_timestamp = stats['timestamp'] + for fn in sorted(listdir(logdir)): + if fn == 'current': + continue + log = splitext(fn)[0] + if log < parse_timestamp(stats['timestamp']): + continue + logfile = join(logdir, fn) + if isfile(logfile): + if fn.endswith('.bz2'): + # This hack is required becaue python2 does not support + # multi-stream files (http://bugs.python.org/issue1625) + (fd, tmpfile) = utils.temp_filename(Cnf['Dir::TempPath']) + system('bzcat %s > %s' % (logfile, tmpfile)) + with open(tmpfile, 'r') as fd: + data = fd.read() + unlink(tmpfile) + else: + with open(logfile, 'r') as fd: + data = fd.read() + ts = parse_new_uploads(data) + if ts > latest_timestamp: + latest_timestamp = ts + ts = parse_actions(data, log) + if ts > latest_timestamp: + latest_timestamp = ts + stderr.write('.') + stderr.flush() + stderr.write('\n') + stderr.flush() + stats['timestamp'] = latest_timestamp + with open(yaml, 'w') as fd: + safe_dump(stats, fd) ################################################################################ def main (): - global Cnf, projectB; + global Cnf + global users - Cnf = utils.get_conf(); - Arguments = [('h',"help","Saffron::Options::Help")]; + Cnf = utils.get_conf() + Arguments = [('h',"help","Stats::Options::Help")] for i in [ "help" ]: - if not Cnf.has_key("Saffron::Options::%s" % (i)): - Cnf["Saffron::Options::%s" % (i)] = ""; + if not Cnf.has_key("Stats::Options::%s" % (i)): + Cnf["Stats::Options::%s" % (i)] = "" - args = apt_pkg.ParseCommandLine(Cnf, Arguments, sys.argv); + args = apt_pkg.parse_commandline(Cnf, Arguments, sys.argv) - Options = Cnf.SubTree("Saffron::Options") + Options = Cnf.subtree("Stats::Options") if Options["Help"]: - usage(); + usage() if len(args) < 1: - utils.warn("saffron requires at least one argument"); - usage(1); + utils.warn("dak stats requires a MODE argument") + usage(1) elif len(args) > 1: - utils.warn("saffron accepts only one argument"); - usage(1); - mode = args[0].lower(); - - projectB = pg.connect(Cnf["DB::Name"], Cnf["DB::Host"], int(Cnf["DB::Port"])); + if args[0].lower() != "new": + utils.warn("dak stats accepts only one MODE argument") + usage(1) + elif args[0].lower() == "new": + utils.warn("new MODE requires an output file") + usage(1) + mode = args[0].lower() if mode == "arch-space": - per_arch_space_use(); + per_arch_space_use() elif mode == "pkg-nums": - number_of_packages(); + number_of_packages() elif mode == "daily-install": - daily_install_stats(); + daily_install_stats() + elif mode == "new": + users = utils.get_users_from_ldap() + new_stats(Cnf["Dir::Log"], args[1]) else: - utils.warn("unknown mode '%s'" % (mode)); - usage(1); + utils.warn("unknown mode '%s'" % (mode)) + usage(1) ################################################################################ if __name__ == '__main__': main() -