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