Adding another column to awk output - linux

Question

I have a HAProxy log file with content similar to this:
Feb 28 11:16:10 localhost haproxy[20072]: 88.88.88.88:6152 [28/Feb/2017:11:16:01.220] frontend backend_srvs/srv1 9063/0/0/39/9102 200 694 - - --VN 9984/5492/191/44/0 0/0 {Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/56.0.2924.87 Safari/537.36|http://subdomain.domain.com/location1} "GET /location1 HTTP/1.1"
Feb 28 11:16:10 localhost haproxy[20072]: 88.88.88.88:6152 [28/Feb/2017:11:16:10.322] frontend backend_srvs/srv1 513/0/0/124/637 200 14381 - - --VN 9970/5491/223/55/0 0/0 {Mozilla/5.0 AppleWebKit/537.36 Chrome/56.0.2924.87 Safari/537.36|http://subdomain.domain.com/location2} "GET /location2 HTTP/1.1"
Feb 28 11:16:13 localhost haproxy[20072]: 88.88.88.88:6152 [28/Feb/2017:11:16:10.960] frontend backend_srvs/srv1 2245/0/0/3/2248 200 7448 - - --VN 9998/5522/263/54/0 0/0 {another user agent with fewer columns|http://subdomain.domain.com/location3} "GET /location3 HTTP/1.1"
Feb 28 11:16:13 localhost haproxy[20072]: 88.88.88.88:6152 [28/Feb/2017:11:16:10.960] frontend backend_srvs/srv1 2245/0/0/3/2248 200 7448 - - --VN 9998/5522/263/54/0 0/0 {Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/56.0.2924.87 Safari/537.36|} "GET /another_location HTTP/1.1"
I want to extract some of the fields in order to have the following output:
Field 1 Field 2 Field 3 Field 4 Field 5 Field 6
Date/time HTTP status code HTTP Method Request HTTP version Referer URL
Basically, in this particular case the output should be:
Feb 28 11:16:10 200 GET /location1 HTTP/1.1 http://subdomain.domain.com/location1
Feb 28 11:16:10 200 GET /location2 HTTP/1.1 http://subdomain.domain.com/location2
Feb 28 11:16:13 200 GET /location3 HTTP/1.1 http://subdomain.domain.com/location3
Feb 28 11:16:13 200 GET /another_location HTTP/1.1
The only problem here is extracting the Referer URL which is between curly brackets together with the user agent and they're separated by a pipe. Also, the user agent has a variable number of fields.
The only solution I could think of was extracting the referer url separately and then pasting the columns together:
requests_temp=`grep -F " 88.88.88.88:" /root/file.log | tr -d '"'`
requests=`echo "${requests_temp}" | awk '{print $1" "$2" "$3" "$11, $(NF-2), $(NF-1), $NF}' > /tmp/requests_tmp`
referer_url=`echo "${requests_temp}" | awk 'NR > 1 {print $1}' RS='{' FS='}' | awk -F'|' '{ print $2 }' > /tmp/referer_url_tmp`
paste /tmp/abuse_requests_tmp /tmp/referer_url_tmp
But I don't really like this method. Is there any other way in which I can do it using only one awk line? Maybe assign the referer url column to a variable inside awk and then using it to create the same output?

Answer 1

try below solution -
awk '/88.88.88.88/ {gsub(/"/,"",$0);split($(NF-3),a,"|"); {print $1,$2,$3,$11, $(NF-2), $(NF-1), $NF, substr(a[2],1,(length(a[2])-1))}}' a
Feb 28 11:16:10 200 GET /location1 HTTP/1.1 http://subdomain.domain.com/location1
Feb 28 11:16:10 200 GET /location2 HTTP/1.1 http://subdomain.domain.com/location2
Feb 28 11:16:13 200 GET /location3 HTTP/1.1 http://subdomain.domain.com/location3
Feb 28 11:16:13 200 GET /another_location HTTP/1.1

Answer 2

You can do all at once using awk:
awk '$6 ~ /88\.88\.88\.88:[0-9]+/{
split($0,a,/[{}]/)
$0=a[1] OFS a[3]
split(a[2],b,"|")
print $1,$2,$3,$11,substr($18,2),$19,substr($20,1,length($20)-1),b[2]
}' file.log
The first split is splitting the variable part of line (included in between the {...}) into the array a.
The line is rebuilt in order to have a fix number of fields $0=a[1] OFS a[3]
The second split allows extracting the URL from variable based on | characters.
At last the print shows all needed elements. Note the substr are here for removing the ".