#!/bin/bash
egrep -o '("WARC-Target-URI":"https?:|"Last-Modified":"[^"]*")'|\
 egrep -o '(https?:|:".*"$)' |\
 tr '\012' \# | sed 's/:#:/ /g'|tr \# '\012' | tr -d \"|\
 sed 's/ [[:digit:]][[:digit:]]\?:[[:digit:]][[:digit:]]:[[:digit:]][[:digit:]] / /;s/\(https\? \)\(: \)\?[MTWFSa-z]..\.\?, \?/\1/;s/ \([-+][[:digit:]]\{4\}\|[[:upper:]]\{2,3\}\)$//;s/ [[:digit:]]\{1,2\} / /;s/\/[[:digit:]]\{1,2\}\/\([[:digit:]]\{4\}\)$/ \1/'|\
awk '{c[$0]+=1} END {for (k in c) {print k, c[k]}}'