rotate-www-logs

#!/bin/bash

# shellcheck source=scripts/lib/common.sh
source "$(dirname -- "$( readlink -f -- "$0"; )")/lib/common.sh"

usage(){
PrintHelp \
 "[command]$(basename "${BASH_SOURCE[0]}")[/] [path][parameter]<NEW_LOG_FILENAME>[/]" \
 "$(cat << EOF
Move Apache access log files into a by-month subdirectory in the format [path]<LOGS_ROOT>/apache/<YYYY-MM>[/], then gzip them.

Must be run as [user]root[/], and is generally run by the Apache [command]rotatelogs[/] subprocess as such.
EOF
)" \
 "$(cat << EOF
[path][parameter]<NEW_LOG_FILENAME>[/]

	The filename of the new Apache access log file created by [command]rotatelogs[/].
EOF
)"
}

if [ $# -eq 0 ] || { [ $# -eq 1 ] && { [ "$1" = "--help" ] || [ "$1" = "-h" ]; }; }; then
	usage
fi

# Apache has a habit of starting this script twice, which can stomp on its own files.
for pid in $(pidof -x rotate-www-logs); do
	if [ "${pid}" != $$ ]; then
		# We `echo()` and `exit()` instead of `ExitWithError()` because Apache prints stderr to the log, but not stdout. We don't need this logged.
		echo "rotate-www-logs is already running with PID ${pid}"
		exit 1
	fi
done

# Prevent the loop from entering if no matches are found for the pattern.
shopt -s nullglob

filenameBase=$(basename "$1" | sed --regexp-extended "s/\.[0-9]+$//")
directory=$(dirname "$1")

for filename in "${directory}/${filenameBase}".*; do
	# When Apache calls this script, it passes the filename of the new log file it created.
	# Thus, we check here to make sure we don't process and then delete the brand-new log file!
	if [ "${filename}" != "$1" ]; then
		# Apache log files can have data for more than one day. Pull out entries for different days into different files.
		dates=$(grep --extended-regexp --only-matching "\[[0-9]{1,2}\/[a-zA-Z]{3}\/20[0-9]{2}" "${filename}" | sort --unique)

		while read -r line; do
			logRawDate=$(echo "${line}" | sed "s/\[//g" | sed "s/\// /g")
			logDate=$(date --date="${logRawDate}" "+%Y-%m-%d")
			logMonth=$(date --date="${logRawDate}" "+%Y-%m")
			grepString=${line//\[/}
			logFilename="www-access-${logDate}.log"

			mkdir --parents "${directory}/${logMonth}"

			# Is the log file already existing and gzipped?
			if [ -f "${directory}/${logMonth}/${logFilename}.gz" ]; then
				gunzip "${directory}/${logMonth}/${logFilename}.gz"
			fi

			# `ipv6loganon` is provided by the `ipv6calc` package.
			grep --extended-regexp "\[${grepString}" "${filename}" | ipv6loganon --anonymize-paranoid >> "${directory}/${logMonth}/${logFilename}"

			gzip --force --best "${directory}/${logMonth}/${logFilename}"

			chown --preserve-root --recursive www-data:adm "${directory}/${logMonth}"
			chmod --preserve-root --recursive g+w "${directory}/${logMonth}"
		done <<< "${dates}"

		rm "${filename}"
	fi
done

# Set this for `fail2ban` to use.
ln --force --symbolic "$1" "${directory}"/current.log