#!/bin/bash

# give some time for log collector to do its job, 5 minutes didn't seem to be enough
clean_delay=20

# this is a better implementation, it will catch any file in the s3 share, whatever folder path there is

[[ ! -d /data/ngenix/extracting-s3fs/ ]] && echo error: could not find /data/ngenix/extracting-s3fs/ && exit 1

[[ -z `mount | grep s3fs` ]] && echo error: s3fs not mounted && exit 1

mkdir -p /data/ngenix/traces/
cd /data/ngenix/traces/

[[ ! -f last ]] && touch last
mv -f last last.old
find /data/ngenix/s3fs/logs/ -type f > last

oldmd5=`md5sum last.old | awk '{print $1}'`
newmd5=`md5sum last | awk '{print $1}'`

if [[ ! $oldmd5 = $newmd5 ]]; then
	date --rfc-email

	echo -n \ cleaning up $clean_delay minutes old files in /data/ngenix/extracting-s3fs/ ...
	find /data/ngenix/extracting-s3fs/ -maxdepth 1 -type f -mmin +$clean_delay -exec rm -f {} \; && echo done

	echo \ new log archives are:

	logs=`diff last.old last | sed -rn 's/^> (.*)/\1/p'`

	if [[ -z $logs ]]; then
		echo \ info: no _new_ log archive - but some got removed
		# echo /data/ngenix/s3fs/logs/s91051/s91051_20241219_121458.log.gz | sed -rn 's@.*/([^/]+)$@\1@p'
		#diff last.old last
		echo
		exit 1
	fi

	echo "$logs"

	for logpath in $logs; do
                logfile=${logpath##*/}
                logfile=`echo $logfile | sed 's/\.gz//'`

                if [[ -f /data/ngenix/extracting-s3fs/$logfile ]]; then
			# keep up with the loop even if that one exists
                        echo -e \ warn: /data/ngenix/extracting-s3fs/$logfile is already there\\n
                else
                        echo -n \ /data/ngenix/extracting-s3fs/$logfile ...
                        nice zcat $logpath > /data/ngenix/extracting-s3fs/$logfile && echo done || echo FAIL
                fi

                unset logfile
	done; unset log
	echo
fi