#!/usr/bin/ksh
#
# process_weblog4carisma.ksh -  Process the THEMIS webserver logs
# for CARISMA statistics.
# Processes weblog files in /disks/themisdata/weblogs, created by grab_weblog.ksh
#
# File Locations:
# original WEBLOGS at danaid: /var/log/httpd/access_log.1.gz
# archived WEBLOGS: /disks/themisdata/weblogs/archived
# CARISMA WEBLOGS: /disks/themisdata/weblogs/carisma
#	

#
## Set the environment variables
#
	if [[ -z $THMSOC ]]
	then
		THMSOC=/disks/socware/thmsoc_dp_current
		export THMSOC
	fi
	. ${THMSOC}/src/config/soc_it_to_me.config

#
## Function Definitions
#
function nmonth {
	# Input name of month (Jan, Feb, ...)
	# Return number of the month
	nmon="00"
	if [[ $1 = "Dec" ]]
	then
		nmon="12"
	elif [[ $1 = "Nov" ]]
	then
		nmon="11"
	elif [[ $1 = "Oct" ]]
	then
		nmon="10"
	elif [[ $1 = "Sep" ]]
	then
		nmon="09"
	elif [[ $1 = "Aug" ]]
	then	
		nmon="08"
	elif [[ $1 = "Jul" ]]
	then
		nmon="07"
	elif [[ $1 = "Jun" ]]
	then
		nmon="06"
	elif [[ $1 = "May" ]]
	then
		nmon="05"
	elif [[ $1 = "Apr" ]]
	then
		nmon="04"
	elif [[ $1 = "Mar" ]]
	then
		nmon="03"
	elif [[ $1 = "Feb" ]]
	then
		nmon="02"
	elif [[ $1 = "Jan" ]]
	then
		nmon="01"
	fi
	echo $nmon
}

function ndate {
	# Input a line of a weblog
	# Return the date in line as YYYYMMDD
	# If weblog file format changes, this function will have to change
	day1=$(echo $4| awk -F/ '{print substr($1,2)}')
	month1=$(echo $4| awk -F/ '{print $2}')
	year1=$(echo $4| awk -F/ '{print substr($3,0,4)}')
	month1n=$(nmonth $month1)
	echo ${year1}${month1n}${day1}
}

#
## Handle input arguments 
#
##NONE

#
## Check for lock file
#
	if [[ -a /var/tmp/process_weblog4carisma_lock ]] ; then
		echo Process in progress, quitting........
		exit 0
	fi

	echo $$ > /var/tmp/process_weblog4carisma_lock

#
## Process weblogs, for CARISMA entries only
#
	for file in $(ls -1 ${WEBLOGS}/thmweb*)
	do
		echo "Script: process_weblog4carisma.ksh. File: $file"
		topline=$(head -1 $file)
		bottomline=$(tail -1 $file)
		startDate=$(ndate $topline)
		stopDate=$(ndate $bottomline) 

		# New filenames for carisma
		carismafile=carisma_accesslog_${startDate}_${stopDate}

		# Save all CARISMA entries in a separate file
		egrep -i "^.*(mag_atha|mag_fsim|mag_fsmi|mag_gill|mag_pina|mag_rank|mag_snkq|ype=gmag).*$" $file |egrep -v "128.32.18" |egrep -v " 404 " >> $carismafile			        
        
		cp $carismafile ${WEBLOGS}/carisma
		rm $carismafile 
		echo "${carismafile} was processed and moved to ${WEBLOGS}/carisma"
		mv $file ${WEBLOGS}/archived
		echo "${WEBLOGS}/$file was processed and moved to ${WEBLOGS}/archived"
		echo "====================="
	done

#
## Cleanup and exit
#
	rm /var/tmp/process_weblog4carisma_lock
	echo "End of process_weblog4carisma.ksh"
	exit 0