#!/usr/bin/env python
# waccess - Copyright (c) 2001,2002, TundraWare Inc., All Rights Reserved
# $Id: waccess,v 1.5 2002/08/30 22:32:01 tundra Exp $

#
# Look for selected strings passed on the command line in the http access log.
# If found, dump the address, name, item retrieved, and access date for the
# matching record.


import commands
import getopt
import socket
import sys

##########
# Booleans
##########

FALSE = 0==1
TRUE = not FALSE

##########
# Constants & Tables
##########

# List of IP addesses to ignore.  Records with IP addresses found
# in this list will be ignored entirely.  The addresses here may
# be partial IP quads.

IGNORED = ["127.0", "192.168.0."]


##########
# Function Defintions
##########

def usage():
    print "usage: waccess [-rs -f logfile]"
    sys.exit(2)
    


##########
# Command Line Processing
##########

LOG = "/var/log/httpd-access.log"
REVERSE = FALSE
SHOW    = TRUE


try:
    opts, args = getopt.getopt(sys.argv[1:], '-f:rs')
except getopt.GetoptError:
    usage()
    
for opt, val in opts:
    if opt == "-f":
        LOG = val
    if opt == "-r":
        REVERSE = TRUE
        SHOW = TRUE
    if opt == "-s":
        SHOW = FALSE
        REVERSE = FALSE

##########
# Process the log
##########

f = open(LOG)

matched = {}
for a in args:
    matched[a] = 0

total = 0

# Read in the whole log file
for record in f.read().splitlines():

    total += 1
    fields = record.split()

    # See if this is an IP address to ignore

    PROCESS = TRUE
    for ignoreIP in IGNORED:
        if fields[0].startswith(ignoreIP):
            PROCESS = FALSE
       
    if PROCESS:

        # Check each log record for a match with any command line argument

        MATCHED = FALSE
        for a in args:
            if record.count(a):
                i = 0
                revname = ""
                matched[a] += 1
                MATCHED = TRUE


        # But only display the matching record once, regardless of how many
        # matching substrings are found.

        if MATCHED:
            if REVERSE:
                try:
                    revname = socket.gethostbyaddr(fields[0])[0]
                except:
                    revname = "NO REVERSE RESOLUTION"

            if SHOW:
                print fields[3][1:], " " * (19 - len(fields[3][1:])), \
                      fields[0], " " * (15 - len(fields[0])), \
                      revname[-(35+1):], " " * (35 - len(revname)), \
                      fields[5], " " * (8 - len(fields[5])), fields[6]

        

f.close()
print "\nProcessed %d Total Records.\n" % (total,)
for a in args:
    print "Found %d Matching Records Containing: %s" % (matched[a], a)