#!/usr/bin/env bash
#
# refresh libgen databases from dump files

version="0.6.1"
release="20200805"

trap "trap_error" TERM
trap "trap_exit" EXIT
export TOP_PID=$$

main () {

	# PREFERENCES
	config=${XDG_CONFIG_HOME:-$HOME/.config}/books.conf

	# maximum age (in days) of database dump file to use
	max_age=5

	# database server to use
	dbhost="localhost"
	dbport="3306"
	dbuser="libgen"

	# where to get updates. A change here probably necessitates a change in the urls array
	# as dump file names can be site-specific. 
	base="http://gen.lib.rus.ec/dbdumps/"
	#base="https://lgdumps.xyz/dumps/"

	# database names
	declare -A databases=(
		[libgen]=libgen
		[compact]=libgen_compact
		[fiction]=libgen_fiction
	)

	# source config file if it exists
	[[ -f ${config} ]] && source ${config}

	# (mostly) END OF PREFERENCES

	# urls for dump files (minus datestamp and extension)
	declare -A urls=(
		[libgen]="${base}/libgen"
		[compact]="${base}/libgen_compact"
		[fiction]="${base}/fiction"
	)

	# sql to get time last modified for database
	declare -A lastmodified=(
		[libgen]="select max(timelastmodified) from updated;"
		[compact]="select max(timelastmodified) from updated;"
		[fiction]="select max(timelastmodified) from fiction;"
	)

	declare -A filter=(
		[libgen]='s/DEFINER[ ]*=[ ]*[^*]*\*/\*/;s/DEFINER[ ]*=[ ]*[^*]*PROCEDURE/PROCEDURE/;s/DEFINER[ ]*=[ ]*[^*]*FUNCTION/FUNCTION/'
		[compact]='s/DEFINER[ ]*=[ ]*[^*]*\*/\*/;s/DEFINER[ ]*=[ ]*[^*]*PROCEDURE/PROCEDURE/;s/DEFINER[ ]*=[ ]*[^*]*FUNCTION/FUNCTION/'
	)

	# sql to run BEFORE update
	declare -A before_update=(
	)

	# sql to run AFTER update
	declare -A after_update=(
		[compact]="drop trigger updated_edited;create table description (id int(11) not null auto_increment, md5 varchar(32) not null default '', descr varchar(20000) not null default '', toc mediumtext not null, TimeLastModified timestamp not null default current_timestamp on update current_timestamp, primary key (id), unique key md5_unique (md5) using btree, key time (timelastmodified) using btree, key md5_hash (md5) using hash);"
	)

	declare -A options=(
		[wget]="-nv"
		[wget_verbose]=""
		[unrar]="-inul"
		[unrar_verbose]=""
	)


	tmpdir=$(mktemp -d /var/tmp/libgen.XXXXXX)

	while getopts "cd:efhH:knp:P:R:u:U:v@" OPTION
	do
	case $OPTION in
		n)
			no_action=1
			;;
		f)
			force_refresh=1
			;;
		d)
			max_age=${OPTARG}
			;;
		u)
			if [[ -v "databases[${OPTARG}]" ]]; then
				dbs+=" ${OPTARG}"
			else
				exit_with_error "-u ${OPTARG}: no such database"
			fi
			;;
		v)
			verbose="_verbose"
			;;
		H)
			dbhost="${OPTARG}"
			;;
		P)
			dbport="${OPTARG}"
			;;
		U)
			dbuser="${OPTARG}"
			;;
		p)
			password="${OPTARG}"
			if [[ -z $password ]]; then
				password=$(read_password)
				echo
			fi
			;;
		c)
			if [[ ! -f "${config}" ]]; then
				cat <<-EOT > "${config}"
				dbhost=${dbhost}
				dbport=${dbport}
				dbuser=${dbuser}
				base=${base}
				EOT
			else
				exit_with_error "-c: config file ${config} exists, either remove it or edit it directly"
			fi
			exit
			;;
		e)
			if [[ -f "$config" ]]; then
				if [[ "$VISUAL" ]]; then "$VISUAL" "$config";
				elif [[ "$EDITOR" ]]; then "$EDITOR" "$config";
				else exit_with_error "-e: no editor configured, can not edit $config"
				fi
			else
				exit_with_error "-e: config file does not exist, create is first (see -c)"
			fi
			exit
			;;
		a)
			if url_available "${OPTARG}"; then
				base="${OPTARG}"
			else
				exit_with_error "-a ${OPTARG}: repository not available"
			fi
			;;
		@)
			use_torsocks=1
			source $(which torsocks) on
			;;
		k)
			keep_downloaded_files=1
			;;
		h)
			help
			exit
			;;
	esac
	done

	check_sanity

	[[ -n ${password} ]] && dbpass="-p${password}"

	# skip credential check when using password prompt option - password will be asked often enough as it is...
	[[ ${dbpass} != "-p" ]] && check_credentials

	[[ -z ${dbs} ]] && dbs="${!databases[@]}"

	pushd $tmpdir >/dev/null
	for db in ${dbs}; do
		database=${databases[$db]}
		if [[ $(db_exists "$database") ]]; then
			db_dump=$(is_available ${db} ${max_age})
			if [[ -n $db_dump ]]; then
				[[ -n $verbose ]] && echo "update available for ${db}: ${db_dump}"
				if [[ -z ${no_action} ]]; then
					wget ${options[wget${verbose}]} ${db_dump}
					unrar ${options[unrar${verbose}]} x $(basename ${db_dump})
					[[ -n "${filter[$db]}" ]] && run_filter $(unrar lb $(basename ${db_dump})) "${filter[$db]}"
					drop_tables=$(drop_table_sql "${database}")
					[[ -n $drop_tables ]] && dbx ${database} "${drop_tables}"
					[[ -n ${before_update[$db]} ]] && dbx ${database} "${before_update[$db]}"
					[[ -n ${filter[$db]} ]] && filter_command="|sed -e '${filter[$db]}'"
					if [[ -n $verbose ]]; then
						echo "importing $(basename ${db_dump}) into ${database}"
						pv $(unrar lb $(basename ${db_dump})) | dbx ${database}
					else
						dbx ${database} < $(unrar lb $(basename ${db_dump}))
					fi
					[[ -n ${after_update[$db]} ]] && dbx ${database} "${after_update[$db]}"
				fi
			else
				[[ -n $verbose ]] && echo "no update available for ${db}"
			fi
		else
			echo "database '$database' does not exist, please create it before attempting to refresh" >&2
		fi
	done
	popd >/dev/null
}

dbx () {
	database=$1
	shift

        if [ $# -gt 0 ]; then
                mysql -Bsssss -h ${dbhost} -P ${dbport} -u ${dbuser} ${dbpass} ${database} -e "$*"
        else
                mysql -Bsssss -h ${dbhost} -P ${dbport} -u ${dbuser} ${dbpass} ${database}
        fi
}


# check whether there is a dump file which is more recent than the current database and no older
# than $max_age
is_available () {
        db="$1"
        max_age="$2"

	db_age=$(db_age $db)

	age=0

	while [[ $age -lt $db_age && $age -lt $max_age ]]; do
		timestamp=$(date -d "@$(($(date +%s) - $((60*60*24*$age))))" +%Y-%m-%d)
        	result=$(w3m -dump ${base} | awk '{ print $1 }'|grep $(basename "${urls[$db]}_${timestamp}.rar"))
		[[ -n $result ]] && break
		let age+=1
	done

    [[ -n $result ]] && echo $(dirname ${urls[$db]})/${result}
}

# drop tables to prepare database for refresh
drop_table_sql () {
        database="$1"
	dbx "$database" "SELECT concat('DROP TABLE IF EXISTS ', table_name, ';') FROM information_schema.tables WHERE table_schema = '$dbname';"
}

# returns database name if it exists, nothing otherwise
db_exists () {
	database="$1"
	dbx $database "select schema_name from information_schema.schemata where schema_name='$database';" 2>/dev/null
}

# return database age in days
db_age () {
	db="$1"
	now=$(date +%s)
	age=0
	if [[ "$force_refresh" -gt 0 ]]; then
		age=$max_age
	else
		db_last_modified=$(date -d "$(dbx $database ${lastmodified[$db]})" +%s)
		age=$(((${now}-${db_last_modified})/60/60/24))
	fi
	echo -n $age
}

# run filter on dump
run_filter () {
	dump_file="$1"
	filter="$2"
	if [[ -n $verbose ]]; then
		echo "running '$filter' on '$dump_file'"
	fi
	sed -i -e "$filter" "$dump_file"
}

# find tool, returns the first|one|found, exit with error message if none found
find_tool () {
        IFS='|' read -ra tools <<< "$*"

        found=0

        for tool in "${tools[@]}"; do
                if [[ -n $(which "$tool") ]]; then
                        found=1
                        break
                fi
        done

        if [[ $found -eq 0 ]]; then
                if [[ ${#tools[@]} -gt 1 ]]; then
                        exit_with_error "missing programs: $*; install at least one of these: ${tools[*]} and try again"
                else
                        exit_with_error "missing program: $1; please install and try again"
                fi
        fi

        echo "$tool"
}

# read password from command line
read_password () {
	password=""
	prompt="Enter database password:"
	while IFS= read -p "$prompt" -r -s -n 1 char
	do
	    if [[ $char == $'\0' ]]
	    then
		break
	    fi
	    prompt='*'
	    password+="$char"
	done
	echo "$password"
}

check_credentials () {
	if [[ ! $(dbx "" "select true;" 2>/dev/null) ]]; then
		exit_with_error "database connection error, bad username or password?"
	fi
}

url_available () {
	url="$1"
	wget -q --spider "$url"
}

check_sanity () {
	find_tool "w3m" > /dev/null
	find_tool "wget" > /dev/null
	find_tool "unrar" > /dev/null
	[[ -n $verbose ]] && find_tool "pv" > /dev/null
}

cleanup () {
        if [[ ! -v keep_downloaded_files ]]; then
            rm -rf ${tmpdir}
        else
            echo "-k option active, temporary directory ${tmpdir} not removed"
        fi
}

trap_error () {
        cleanup
        exit 1
}

trap_exit () {
        cleanup
        exit
}

# echo error message to stdout and terminate main
exit_with_error () {
	echo "$(basename $0): $*" >&2

        kill -s TERM $TOP_PID
}

help () {
        echo $(basename $(readlink -f $0)) "version $version"
        cat <<- EOT

	Usage: refresh_libgen OPTIONS

	Performs a refresh from a database dump file for the chosen libgen databases.

	    -n		do not refresh database
	 		use together with '-v' to check if recent dumps are available
	    -f		force refresh, use this on first install
	    -v		be verbose about what is being updated
	    -d DAYS	only use database dump files no older than DAYS days (default: ${max_age})
	    -u DBS	refresh DBS databases (default: ${!databases[@]})

	    -H DBHOST	database host (${dbhost})
	    -P DBPORT	database port (${dbport})
	    -U DBUSER	database user (${dbuser})
	    -R REPO	dump repository (${base})
	    -c 		create a config file using current settings (see -H, -P, -U, -R)
	    -e		edit config file

	    -p DBPASS	database password (cache password for this session)
	 		use empty string ("") to get password prompt

	    -@		use tor (through torsocks) to connect to libgen server
	    -k		keep downloaded files after exit
	    -h		this help message

	EOT
}

main "$@"