frontierqueries

#!/bin/bash

#
# Copyright (c) 2009, FERMI NATIONAL ACCELERATOR LABORATORY
# All rights reserved. 
#
# For details of the Fermitools (BSD) license see Fermilab-2009.txt or
#  http://fermitools.fnal.gov/about/terms.html
#

# Parse a log file created by setting FRONTIER_LOG_LEVEL=debug and
#  FRONTIER_LOG_FILE=filename and print to stdout one query per line.
# The query result size is also printed in the form (NNN/XX%) where NNN
#  is the full size in bytes and optionally /XX% is the compression ratio
#  if it was compressed.  The size is by default at the end of the
#  line but with a -sizefirst option it is at the beginning.
# Before the query will be a hyphen (-) if the query is satisfied from
#  the frontier internal cache, an asterisk (*) if it is a "short"
#  time-to-live query that is sent out, a plus (+) if it is a "forever"
#  time-to-live query, or a blank if it is a "long" time-to-live query.
# The -sort option sorts the output by full size.  Implies -sizefirst.
# The -secsfirst option prints the elapsed seconds first.  Otherwise the
#  elapsed seconds always follows the size
# Written by Dave Dykstra
# $Id$
#  
SORT=cat
if [ "$1" = "-sizefirst" ]; then
    PRINTORDER="size, elapsedseconds, mark, query"
    shift
elif [ "$1" = "-sort" ]; then
    # sort by size
    PRINTORDER="size, elapsedseconds, mark, query"
    SORT="sort -t( -k 2 -n"
    shift
elif [ "$1" = "-secsfirst" ]; then
    PRINTORDER="elapsedseconds, size, mark, query"
    shift
else
    PRINTORDER="mark, query, size, elapsedseconds"
fi
if [ $# != 1 ]; then
	echo "Usage: $0 [-sizefirst|-sort|-secsfirst] frontier_client.log" >&2
	exit 1
fi
awk '
BEGIN{
    inquery=0
    query=""
}
function parsetime(mon,day,hhmmss,year) {
    split(hhmmss,times,":")
    if (mon == "Jan") {month="01"}
    else if (mon == "Feb") {month="02"}
    else if (mon == "Mar") {month="03"}
    else if (mon == "Apr") {month="04"}
    else if (mon == "May") {month="05"}
    else if (mon == "Jun") {month="06"}
    else if (mon == "Jul") {month="07"}
    else if (mon == "Aug") {month="08"}
    else if (mon == "Sep") {month="09"}
    else if (mon == "Oct") {month="10"}
    else if (mon == "Nov") {month="11"}
    else if (mon == "Dec") {month="12"}
    return mktime(year " " month " " day " " times[1] " " times[2] " " times[3])
}
/querying on chan/{
    starttime=parsetime($9,$10,$11,$12)
}
/chan .* response .* finished at/{
    elapsedseconds=parsetime($10,$11,$12,$13)-starttime
}
/encoding request/{
    if (query != "") {
	print '"$PRINTORDER"'
    }
    mark=" "
    start=index($0, "t [")
    query=substr($0,start+3)
    end=index(query, "]")
    if (end==0) {
	end=length(query)+1
	inquery=1
    }
    query=substr(query,1,end-1)
    if (inquery==1)
	next
}
(inquery==1){
   end=index($0, "]") 
    if (end==0) {
        end=length($0)+1
    }
    else {
	inquery=0
    }
    query=query substr($0,1,end-1)
}
/<GET .*&ttl=forever/{
    mark="+"
}
/<GET .*&ttl=short/{
    mark="*"
}
/HIT .* client cache/{
    mark="-"
}
/full size/{
    start=index($0,"full size")
    fullsize=substr($0,start+10)
    end=index(fullsize,")")
    fullsize=substr(fullsize,1,end-1)
    start=index($0,"uncompressing")
    if (start != 0) {
	compressedsize=substr($0,start+14)
	end=index(compressedsize," ")
	compressedsize=substr(compressedsize,1,end-1)
	compressedpercent=sprintf("/%.1f%%",(compressedsize*100/fullsize))
    }
    else {
        compressedpercent=""
    }
    size="(" fullsize compressedpercent ")"
}
END{
    if (query != "") {
	print '"$PRINTORDER"'
    }
}
' $1 | $SORT