#!/usr/bin/env python2.5
"""
Tool for accessing the search index
"""
# XXX: fix path
import sys; sys.path.insert(0, '.'); sys.path.insert(0, '..')
import os, os.path
import datetime, pytz
# configuration and the LogSearchIndex module
import config, utils, log_search, channels
def _open_index (options, open_mode) :
"""
Opens the LogSearchIndex
"""
return log_search.LogSearchIndex(config.LOG_CHANNELS, options.index_path, open_mode)
def _open_index_and_channel (options, channel_name, open_mode) :
"""
Opens+returns a LogSearchIndex and a LogChannel
"""
# open the LogSearchIndex
index = _open_index(options, open_mode)
# open the channel
channel = config.LOG_CHANNELS.lookup(channel_name)
# return
return index, channel
def _insert_lines (index, options, channel, lines) :
"""
Insert the given lines into the index.
Assumes the lines will be in time-order, and prints out as status messages the date and count for the inserted lines
"""
# last date
date = None
# count
count = 0
# iter lines
for line in lines :
# output new date header?
if not options.quiet and (not date or line.timestamp.date() != date) :
# previous date's line count?
if date :
print "OK: %d lines" % count
# reset count
count = 0
# timestamp's date
date = line.timestamp.date()
# status header
print "%s:" % (date.strftime('%Y-%m-%d'), ),
# insert
index.insert_line(channel, line)
# count
count += 1
# final count line
if not options.quiet and date :
print "OK: %d lines" % count
def _load_channel_date (index, options, channel, date) :
"""
Loads the logs for the given date from the channel's LogSource into the given LogSearchIndex
"""
if not options.quiet :
print "Loading date for channel %s" % channel.id
try :
# load lines for date
lines = channel.source.get_date(date)
except Exception, e :
if not options.skip_missing :
raise
if not options.quiet :
print "\tSkipped: %s" % (e, )
else :
# insert
_insert_lines(index, options, channel, lines)
def _parse_date (options, date_str, tz=None, fmt='%Y-%m-%d') :
"""
Parse the given datetime, using the given timezone(defaults to options.tz) and format
"""
# default tz
if not tz :
tz = options.tz
try :
# parse
return datetime.datetime.strptime(date_str, fmt).replace(tzinfo=tz)
except Exception, e :
raise CommandError("[ERROR] Invalid date: %s: %s" % (date_str, e))
def _output_lines (options, lines) :
"""
Display the formatted LogLines
"""
# display as plaintext
for line, txt_data in options.formatter.format_txt(lines, full_timestamps=True) :
print txt_data
class CommandError (Exception) :
"""
Error with command-line arguments
"""
pass
def cmd_create (options) :
"""
Creates a new index
"""
# open index
index = _open_index(options, 'c' if not options.force else '*')
# that's all
pass
def cmd_load (options, channel_name, *dates) :
"""
Loads the logs for a specific channel for the given dates (in terms of the channe logs' timezone) into the index
"""
# open index/channel
index, channel = _open_index_and_channel(options, channel_name, 'c' if options.create else 'a')
# handle each date
for date_str in dates :
# prase date
try :
date = _parse_date(options, date_str, channel.source.tz)
# handle errors
except CommandError, e :
if options.skip_missing :
print "[ERROR] %s" % (date_name, e)
else :
raise
# otherwise, load
else :
_load_channel_date(index, options, channel, date)
def cmd_load_month (options, channel_name, *months) :
"""
Loads the logs for a specific channel for the given months (in terms of the channel's timezone) into the index
"""
# open index/channel
index, channel = _open_index_and_channel(options, channel_name, 'c' if options.create else 'a')
# handle each date
for month_str in months :
# prase date
try :
month = _parse_date(options, month_str, channel.source.tz, '%Y-%m')
# handle errors
except CommandError, e :
# skip?
if options.skip_missing :
print "[ERROR] %s" % (date_name, e)
continue
else :
raise
# get the set of days
days = list(channel.source.get_month_days(month))
print "Loading %d days of logs:" % (len(days))
# load each day
for date in days :
# convert to datetime
dt = datetime.datetime.combine(date, datetime.time(0)).replace(tzinfo=channel.source.tz)
# load
_load_channel_date(index, options, channel, dt)
def cmd_search (options, channel_name, query) :
"""
Search the index for events on a specific channel with the given query
"""
# sanity-check
if options.create :
raise Exception("--create doesn't make sense for 'search'")
# open index/channel
index, channel = _open_index_and_channel(options, channel_name, 'r')
# search
lines = index.search_simple(channel, query)
# display
_output_lines(options, lines)
def cmd_list (options, channel_name, *dates) :
"""
List the indexed events for a specific date
"""
# sanity-check
if options.create :
raise Exception("--create doesn't make sense for 'search'")
# open index/channel
index, channel = _open_index_and_channel(options, channel_name, 'r')
# ...for each date
for date_str in dates :
# parse date
date = _parse_date(options, date_str)
# list
lines = index.list(channel, date)
# display
_output_lines(options, lines)
def cmd_autoload (options, *channel_names) :
"""
Automatically loads all channel logs that have not been indexed yet (by logfile mtime)
"""
# open index
index = _open_index(options, 'c' if options.create else 'a')
# default to all channels
if not channel_names :
channels = config.LOG_CHANNELS
else :
channels = [config.LOG_CHANNELS.lookup(channel_name) for channel_name in channel_names]
# iterate channels
for channel in channels :
if not options.quiet :
print "Channel %s:" % channel.id,
# path to our state file
statefile_path = os.path.join(options.autoload_state_path, 'chan-%s' % channel.id)
# override?
if options.reload :
# load all
mtime = None
if not options.quiet :
print "reloading all:",
# stat for mtime
else :
# stat for mtime, None if unknown
mtime = utils.mtime(statefile_path, ignore_missing=True)
if mtime and not options.quiet :
print "last load=%s:" % mtime,
elif not options.quiet :
print "no previous load state:",
# only after some specific date?
if options.after :
after = options.after
print "after=%s:" % after,
else :
after = None
# get lines
lines = channel.source.get_modified(mtime, after)
# insert
if not options.quiet :
print "inserting..."
_insert_lines(index, options, channel, lines)
# write autoload state
open(statefile_path, 'w').close()
def cmd_help (options, *args) :
"""
Help about commands
"""
import inspect
# general help stuff
options._parser.print_help()
# specific command?
if args :
# the command name
command, = args
# XXX: display info about specific command
xxx
# general
else :
print
print "Available commands:"
# build list of all cmd_* objects
cmd_objects = [(name, obj) for name, obj in globals().iteritems() if name.startswith('cmd_') and inspect.isfunction(obj)]
# sort alphabetically
cmd_objects.sort()
# iterate through all cmd_* objects
for cmd_func_name, cmd_func in cmd_objects :
# remove cmd_ prefix
cmd_name = cmd_func_name[4:]
# inspect
cmd_args, cmd_varargs, cmd_varkw, cmd_default = inspect.getargspec(cmd_func)
cmd_doc = inspect.getdoc(cmd_func)
# remove the "options" arg
cmd_args = cmd_args[1:]
# display
print "\t%10s %-30s : %s" % (cmd_name, inspect.formatargspec(cmd_args, cmd_varargs, None, cmd_default), cmd_doc)
def main (argv) :
"""
Command-line main, with given argv
"""
from optparse import OptionParser
# define parser
parser = OptionParser(
usage = "%prog [options] <command> [ ... ]",
add_help_option = False,
)
# define command-line arguments
parser.add_option('-h', "--help", dest="help", help="Show this help message and exit", action="store_true" )
parser.add_option('-F', "--formatter", dest="formatter_name", help="LogFormatter to use", metavar="FMT", type="choice", default="irssi",
choices=[fmt_name for fmt_name in config.LOG_FORMATTERS.iterkeys()])
parser.add_option('-I', "--index", dest="index_path", help="Index database path", metavar="PATH", default="logs/index" )
parser.add_option( "--autoload-state", dest="autoload_state_path", help="Path to autoload state dir", metavar="PATH", default="logs/autoload-state" )
parser.add_option( "--after", dest="after", help="Only autoload logfiles after the given date", metavar="DATE", default=None )
parser.add_option('-Z', "--timezone", dest="tz_name", help="Timezone for output", metavar="TZ", default="UTC" )
parser.add_option('-f', "--force", dest="force", help="Force dangerous operation", action="store_true" )
parser.add_option( "--create", dest="create", help="Create index database", action="store_true" )
parser.add_option( "--skip-missing", dest="skip_missing", help="Skip missing logfiles", action="store_true" )
parser.add_option( "--reload", dest="reload", help="Force reload lines", action="store_true" )
parser.add_option( "--quiet", dest="quiet", help="Supress status messages", action="store_true" )
# parse
options, args = parser.parse_args(argv[1:])
# postprocess stuff
options._parser = parser
options.tz = pytz.timezone(options.tz_name)
options.formatter = config.LOG_FORMATTERS[options.formatter_name](options.tz, "%H:%M:%S", None, None)
if options.after :
options.after = _parse_date(options, options.after)
# special-case --help
if options.help :
return cmd_help(options, *args)
# must have at least the command argument
if not args :
raise CommandError("Missing command")
# pop command
command = args.pop(0)
# get func
func = globals().get('cmd_%s' % command)
# unknown command?
if not func :
raise CommandError("Unknown command: %s" % command)
# call
func(options, *args)
if __name__ == '__main__' :
try :
main(sys.argv)
sys.exit(0)
except CommandError, e :
print e
sys.exit(1)