#!/bin/sh
#
######################################################
#
# Script: linkSort.sh
# Description: Count and sort Google links
#              output by Google Webmaster Tools
#
# Copyright 2007 Apogee Web Consulting LLC
# http://www.apogee-web-consulting.com
#
# This is open source code.
# Use for personal or commercial use.
# If you break it, fix it.
# If it's already broken, fix it.
# Enjoy!
#
######################################################

# Check args
case $# in
1) LINK_FILE="$1" ;;
*) echo "usage: $0 Filename.csv"; exit ;;
esac

# Parse links file
awk -F, '{print $1}' $LINK_FILE |grep http > /tmp/pagesAll.txt
cat /tmp/pagesAll.txt | sort -u | \
while read page
do
  echo $page `grep -c "$page" /tmp/pagesAll.txt` >> /tmp/pageCounts.txt
done

# Output results
sed 's/\"//g' /tmp/pageCounts.txt |sort -n -k2
echo "========================"
awk '{n+=$NF} END {print "total pages:",n}' /tmp/pageCounts.txt
echo "unique pages: `wc -l /tmp/pageCounts.txt |awk '{print $1}'`"

# Clean up
rm /tmp/pageCounts.txt
rm /tmp/pagesAll.txt