#!/bin/bash
#set -v
# Removed because wget errors with error 8 (Server issued an error response.)
#set -e

## This script will scan www.taler.net for broken links and e-mail a copy of the log if any are found.

logfile="$HOME/linkchecker.log"
wait_time="1"
recurse_level="1"
ignore_list="(.*)demo.taler.net(.*)\/orders\/(.*)" # appears to do *nothing*

# Remove old log

if [ -f "$logfile" ]
	then
		echo $logfile " exists.  Moving to" $logfile".old."
		mv $logfile $logfile.old
	else
		echo "Info: existing log file '$logfile' not found."
fi

# Use wget to scan hosts and save output
for url in "https://www.taler.net/" "https://docs.taler.net/" "https://taler-systems.net/" "https://demo.taler.net/" "https://bank.demo.taler.net/" "https://shop.demo.taler.net/" "https://donations.demo.taler.net/" "https://survey.demo.taler.net/" ; do
	echo "Starting check on $url"
	wget --spider \
		--recursive \
		--no-directories \
		--no-verbose \
		--span-hosts \
		--level="$recurse_level" \
		--wait="$wait_time" \
		--reject-regex "$ignore_list" \
		"$url" 2>&1 | tee --append "$logfile"
done

# Search the log for the phrase "broken link" as this is what wget will report
if grep -iRl 'broken link!!' $logfile
	then
		echo "Found broken links.  Build should fail (exit 1), triggering e-mail notification."
		exit 1
	else
		echo "No broken links found.  Nothing more to do."
		exit 0
fi