From 823fa82d0092ac1baeadbe92186135903217aad6 Mon Sep 17 00:00:00 2001
From: liuxhit <23427350+liuxhit@users.noreply.github.com>
Date: Sun, 25 Jun 2023 12:32:32 +0800
Subject: [PATCH] docker: support retaining at least a few files in backup
 retention

---
 docker/runtime/backup-init.sh    | 64 ++++++++++++++++++++++++++++++--
 docker/runtime/backup-rotator.sh |  8 ++--
 2 files changed, 65 insertions(+), 7 deletions(-)

diff --git a/docker/runtime/backup-init.sh b/docker/runtime/backup-init.sh
index f4e21a1b5..7fb99fb87 100644
--- a/docker/runtime/backup-init.sh
+++ b/docker/runtime/backup-init.sh
@@ -9,8 +9,9 @@ mkdir -p "/etc/crontabs"
 chown git:git /backup
 chmod 2770 /backup
 
-# [string] BACKUP_INTERVAL   Period expression
-# [string] BACKUP_RETENTION  Period expression
+# [string] BACKUP_INTERVAL       Period expression
+# [string] BACKUP_RETENTION      Period expression(Deprecated for forward compatibility)
+# [int]    BACKUP_RETENTION_NUM  Number of surviving backup items
 if [ -z "${BACKUP_INTERVAL}" ]; then
 	echo "Backup disabled: BACKUP_INTERVAL has not been found" 1>&2
 	exit 1
@@ -21,6 +22,12 @@ if [ -z "${BACKUP_RETENTION}" ]; then
 	BACKUP_RETENTION='7d'
 fi
 
+if [ -z "${BACKUP_RETENTION_NUM}" ]; then
+	echo "Backup retention number is not defined, will calculate its value later" 1>&2
+else
+    echo "Backup retention number is given: ${BACKUP_RETENTION_NUM}, will check its value later" 1>&2
+fi
+
 # Parse BACKUP_INTERVAL environment variable and generate appropriate cron expression. Backup cron task will be run as scheduled.
 # Expected format: nu (n - number, u - unit) (eg. 3d means 3 days)
 # Supported units: h - hours, d - days, M - months
@@ -104,6 +111,43 @@ parse_generate_retention_expression() {
 	echo "${FIND_TIME_EXPR} +${TIME_INTERVAL:-7}"
 }
 
+# util function: convert expression to minutes, ignore input check
+# Expected format: nu (n - number, u - unit) (eg. 3d means 3 days)
+# Supported units: m - minutes, h - hours, d - days, M - months
+parse_expression_minutes() {
+	TIME_EXPRESSION="${1:-7d}"
+	# shellcheck disable=SC2001
+	TIME_INTERVAL=$(echo "${TIME_EXPRESSION}" | sed -e 's/[mhdM]$//')
+	# shellcheck disable=SC2001
+	TIME_UNIT=$(echo "${TIME_EXPRESSION}" | sed -e 's/^[0-9]\+//')
+
+	MINUTE_MULTIPLE="1"
+	if   [ "${TIME_UNIT}" = "m" ]; then
+		MINUTE_MULTIPLE="1"
+	elif [ "${TIME_UNIT}" = "h" ]; then
+		MINUTE_MULTIPLE="60"
+	elif [ "${TIME_UNIT}" = "d" ]; then
+		MINUTE_MULTIPLE="1440"
+	elif [ "${TIME_UNIT}" = "M" ]; then
+		MINUTE_MULTIPLE="43200"
+	else
+		echo "Parse error: expression is invalid: ${TIME_EXPRESSION}" 1>&2
+		exit 1
+	fi
+
+	echo "$(( TIME_INTERVAL * MINUTE_MULTIPLE ))"
+
+}
+
+# Giving BACKUP_INTERVAL and BACKUP_RETENTION, calculate the value of BACKUP_RETENTION_NUM we expected.
+# Using BACKUP_RETENTION_NUM directly if BACKUP_RETENTION_NUM is given, ignoring other inputs.
+calc_surviving_backups_number() {
+	# shellcheck disable=SC2086  # Since the input of these two variables has been verified as valid above
+	CALC_BACKUP_RETENTION_NUM="$(( $(parse_expression_minutes ${BACKUP_RETENTION}) / $(parse_expression_minutes ${BACKUP_INTERVAL}) ))"
+	
+	echo "${BACKUP_RETENTION_NUM:-${CALC_BACKUP_RETENTION_NUM}}"
+}
+
 add_backup_cronjob() {
 	CRONTAB_USER="${1:-git}"
 	CRONTAB_FILE="/etc/crontabs/${CRONTAB_USER}"
@@ -129,6 +173,9 @@ add_backup_cronjob() {
 
 CRONTAB_USER=$(awk -v val="${PUID}" -F ":" '$3==val{print $1}' /etc/passwd)
 
+# ================================================================================
+# Keep these lines for the purpose of checking input parameters
+# ================================================================================
 # Up to this point, it was desirable that interpreter handles the command errors and halts execution upon any error.
 # From now, we handle the errors our self.
 set +e
@@ -138,7 +185,18 @@ if [ -z "${RETENTION_EXPRESSION}" ]; then
 	echo "Couldn't generate backup retention expression. Aborting backup setup" 1>&2
 	exit 1
 fi
+# ================================================================================
+
+RETENTION_NUM="$(calc_surviving_backups_number)"
+
+if echo "${RETENTION_NUM}" | grep -qE '^[0-9]+$'; then
+	echo "use RETENTION_NUM: ${RETENTION_NUM}"
+else
+	echo "RETENTION_NUM(${RETENTION_NUM}) is not a valid number, please check the input:" 1>&2
+	echo "input vars: BACKUP_INTERVAL=${BACKUP_INTERVAL} BACKUP_RETENTION=${BACKUP_RETENTION} BACKUP_RETENTION_NUM=${BACKUP_RETENTION_NUM}" 1>&2
+	exit 1
+fi
 
 # Backup rotator cron will run every 5 minutes
-add_backup_cronjob "${CRONTAB_USER}" "*/5 * * * *" "/app/gogs/docker/runtime/backup-rotator.sh" "'${BACKUP_PATH}' '${RETENTION_EXPRESSION}'"
+add_backup_cronjob "${CRONTAB_USER}" "*/5 * * * *" "/app/gogs/docker/runtime/backup-rotator.sh" "'${BACKUP_PATH}' '${RETENTION_NUM}'"
 add_backup_cronjob "${CRONTAB_USER}" "$(parse_generate_cron_expression)" "/app/gogs/docker/runtime/backup-job.sh" "'${BACKUP_PATH}'"
diff --git a/docker/runtime/backup-rotator.sh b/docker/runtime/backup-rotator.sh
index 3b6d10abf..b9dad4830 100644
--- a/docker/runtime/backup-rotator.sh
+++ b/docker/runtime/backup-rotator.sh
@@ -1,10 +1,10 @@
 #!/usr/bin/env sh
 
 # This is very simple, yet effective backup rotation script.
-# Using find command, all files that are older than BACKUP_RETENTION_DAYS are accumulated and deleted using rm.
+# All files that are older than the latest RETENTION_NUM backup files are accumulated and deleted using rm.
 main() {
 	BACKUP_PATH="${1:-}"
-	FIND_EXPRESSION="${2:-mtime +7}"
+	RETENTION_NUM="${2:-1}"
 
 	if [ -z "${BACKUP_PATH}" ]; then
 		echo "Error: Required argument missing BACKUP_PATH" 1>&2
@@ -21,8 +21,8 @@ main() {
 		exit 1
 	fi
 
-	# shellcheck disable=SC2086
-	find "${BACKUP_PATH}/" -type f -name "gogs-backup-*.zip" -${FIND_EXPRESSION} -print -exec rm "{}" +
+	# shellcheck disable=SC2012  # File name is expected
+	ls -1t "${BACKUP_PATH}/"gogs-backup-*.zip | tail -n +"$(( RETENTION_NUM + 1 ))" | xargs -n10 --no-run-if-empty rm
 }
 
 main "$@"