log_sum.sh

#!/bin/bash


if [ "$#" -le 0 ]; then
    echo "Usage $0 [-n N] (-c|-2|-r|-F|-t) <filename>"
    exit 2
fi

FILE="${@: -1}" #File is the last argument
nrOfResult=0




#-r FILE exist and read permission.
if [ ! -r $FILE ]; then
    echo -e "File does not exist, using stdin\n"
	FILE="temp"
	cat <&0 > "$FILE"
fi


function mostConnectionAttemps() {
    # 1. Get all ips from file
    # 2. Sort
    # 3. Uniq --count (prefix lines by the number of occurrences)
    # 4. Sort numeric and reverse the result
    # 5. Change place of the arguments of the result of sort.
    local VAR=`awk '{ print $1 }' $FILE | sort | uniq --count | sort -nr | awk '{ print $2, "\011", $1}'`
    myPrint "$VAR" "<IP><NrOfResult>" "$1"

}


function mostSuccessfulAttemps() {
    local VAR=`awk {'print $9, $1 '} $FILE | grep -E '^2..|^3..' | sort | uniq --count | sort -rn | awk {'print $3, "\011", $1'}`
    myPrint "$VAR" "<ip><nrOfRequests>" "$1"
}


function mostBadRequest() {
    local VAR=`awk {'print $9, $1 '} $FILE | grep -E '^4..|^5..' | sort | uniq --count | sort -rn | awk {'print $2, "\011", $3'}`
    myPrint "$VAR" "<Code><Ip>" "$1"
}

function mostCommonResultCode() {
    # 1. Get all status codes
    # 2. Sort
    # 3. Uniq --count
    # 4. Sort numeric and reverse
    # 5. Change order of print by using awk.

    local VAR=`awk {'print $9, $1'} $FILE | sort | uniq --count | sort -rn | awk {'print $2, "\11", $3'}`   
    myPrint "$VAR" "<StatusCode><ip>" "$1"  

}



function countBytesByIp() {
   
    # local VAR=`awk {'if ($10 !="-")print $10, $1'} $FILE | sort -k2,2 | awk '{arr[$2]+=$1} END { for(i in arr) { print i, "\011", arr[i]}}' | sort -rnk2,2`
    # myPrint "$VAR" "<ip><bytes>" "$1"
    
    i=0
    sum=0
    ip=""
    
    # Get valid data, sort it and save to file.
    awk {'if ($10 !="-")print $10, $1'} $FILE | sort -k2 > tempFile

    # Append a new line to the file because we are off by one. 
    echo -e "\n" >> tempFile
    # For each line in tempFile.
    while read -r line 
    do
      
        #Fetch the columns
        col1=$(echo $line | awk {'print $2'})
        col2=$(echo $line | awk {'print $1'})
        
        #if it is the first time running the loop, set IP to first column. 
        if [ "$i" -eq 0 ]; then
           ip=$col1
           i=1
        fi
        
        
        #If it's the same ip, add to sum. 
        if [ "$col1" = "$ip" ]; then
            sum=$((sum + col2))
            ip=$col1
        else
             #print previous sum and ip to a list.
             echo -e "$ip\t$sum" >> list
             # reset variables.
             sum=$col2
             ip=$col1
        fi

    done < tempFile 

    #read result from the list.
	result=`cat list | sort -rnk2,2`
	#print result
	myPrint "$result" "<ip><byte>" "$1" 
	#cleanup
    rm list tempFile
}


# $1 $VAR 
# $2 Description
# $3 N - Number of results.
function myPrint() {
    # If we have -N set, use head to limit the results
    if [ "$3" -gt 0 ]; then
        echo "$2"
        printf "$1" | head -n "$3"
    else
        printf "$2\n$1"
    fi
}



while getopts :n:c2rFt option
do
    case $option in
        n)
            nrOfResult=$OPTARG
            ;; 
        c)
            # call function with argument (N)
            mostConnectionAttemps $nrOfResult
            ;;
        2)
            mostSuccessfulAttemps $nrOfResult
            ;;
        r)
            mostCommonResultCode $nrOfResult
            ;;
        F)
            mostBadRequest $nrOfResult
            ;;

        t)
            countBytesByIp $nrOfResult
            ;;

                #*)
#            echo "Not a valid argument"
 #           ;;
    esac
done