hassio-addons/scrutiny/rootfs/usr/local/bin/scrutiny-ha-influxdb-preflight

#!/usr/bin/env bash
# shellcheck shell=bash
set -euo pipefail

addon_slug="${SCRUTINY_HA_ADDON_SLUG:-scrutiny}"
data_dir="${SCRUTINY_INFLUXDB_DATA_DIR:-/data/influxdb}"
backup_dir="${SCRUTINY_INFLUXDB_BACKUP_DIR:-/share/${addon_slug}}"
marker_file="${data_dir}/.scrutiny-influxdb-2.9-preflight-complete"
marker_tmp="${marker_file}.tmp"
backup_file="${backup_dir}/influxdb-pre-2.9.tar.gz"
temporary_backup="${backup_file}.tmp"
archive_root="$(basename "$data_dir")"
validation_method="content-sha256-v1"
source_manifest=""
archive_manifest=""

log() {
    local level="$1"
    local message="$2"
    printf 'time="%s" level=%s msg="%s" type=ha-influxdb-upgrade-preflight\n' \
        "$(date -u +"%Y-%m-%dT%H:%M:%SZ")" "$level" "$message"
}

cleanup() {
    rm -f "$temporary_backup" "$marker_tmp"
    if [[ -n "$source_manifest" ]]; then rm -f "$source_manifest"; fi
    if [[ -n "$archive_manifest" ]]; then rm -f "$archive_manifest"; fi
}
trap cleanup EXIT

has_existing_data() {
    if [[ -e "${data_dir}/influxd.bolt" || -e "${data_dir}/influxd.sqlite" ]]; then
        return 0
    fi

    if [[ -d "${data_dir}/engine" ]] && find "${data_dir}/engine" -mindepth 1 -print -quit | grep -q .; then
        return 0
    fi

    return 1
}

validate_archive() {
    local archive="$1"
    [[ -s "$archive" ]] || return 1
    gzip -t "$archive" >/dev/null 2>&1 || return 1
    tar -tzf "$archive" >/dev/null 2>&1
}

reset_manifests() {
    rm -f "$source_manifest" "$archive_manifest"
    source_manifest=""
    archive_manifest=""
}

validate_backup_against_source() {
    local archive="$1"
    local relative_path
    local source_path
    local archive_member
    local listing
    local source_checksum
    local archive_checksum
    local source_target
    local archive_target

    validate_archive "$archive" || return 1
    source_manifest="$(mktemp)"
    archive_manifest="$(mktemp)"

    (
        cd "$data_dir"
        find . -mindepth 1 -print | sed -e 's#^\./##' -e 's#/$##' | LC_ALL=C sort
    ) > "$source_manifest"

    tar -tzf "$archive" | \
        sed -e "s#^${archive_root}/##" -e '/^$/d' -e 's#/$##' | \
        LC_ALL=C sort > "$archive_manifest"

    if ! cmp -s "$source_manifest" "$archive_manifest"; then
        reset_manifests
        return 1
    fi

    while IFS= read -r relative_path; do
        source_path="${data_dir}/${relative_path}"
        archive_member="${archive_root}/${relative_path}"
        listing="$(tar -tvzf "$archive" -- "$archive_member" 2>/dev/null)" || {
            reset_manifests
            return 1
        }

        if [[ -L "$source_path" ]]; then
            [[ "${listing:0:1}" == "l" ]] || {
                reset_manifests
                return 1
            }
            source_target="$(readlink "$source_path")"
            archive_target="${listing##* -> }"
            [[ "$source_target" == "$archive_target" ]] || {
                reset_manifests
                return 1
            }
        elif [[ -f "$source_path" ]]; then
            [[ "${listing:0:1}" == "-" || "${listing:0:1}" == "h" ]] || {
                reset_manifests
                return 1
            }
            source_checksum="$(sha256sum "$source_path" | awk '{print $1}')"
            archive_checksum="$(tar -xOzf "$archive" -- "$archive_member" | sha256sum | awk '{print $1}')"
            [[ "$source_checksum" == "$archive_checksum" ]] || {
                reset_manifests
                return 1
            }
        elif [[ -d "$source_path" ]]; then
            [[ "${listing:0:1}" == "d" ]] || {
                reset_manifests
                return 1
            }
        else
            reset_manifests
            return 1
        fi
    done < "$source_manifest"

    reset_manifests
}

marker_value() {
    local key="$1"
    awk -F= -v key="$key" '$1 == key {sub(/^[^=]*=/, ""); print; exit}' "$marker_file"
}

write_marker() {
    local state="$1"
    local checksum="${2:-}"
    local validation="${3:-}"

    {
        printf 'state=%s\n' "$state"
        if [[ -n "$validation" ]]; then printf 'validation=%s\n' "$validation"; fi
        if [[ -n "$checksum" ]]; then printf 'backup_sha256=%s\n' "$checksum"; fi
    } > "$marker_tmp"
    mv "$marker_tmp" "$marker_file"
}

validate_existing_marker() {
    local state
    local validation
    local expected_checksum
    local actual_checksum

    state="$(marker_value state)"
    validation="$(marker_value validation)"

    case "$state" in
        no-legacy-data)
            if [[ -n "$validation" && "$validation" != "no-data-v1" ]]; then
                log error "no-data migration marker uses an unknown validation method"
                return 1
            fi
            if [[ -z "$validation" ]]; then
                write_marker no-legacy-data "" no-data-v1
                log warning "upgraded no-data migration marker with explicit validation state"
            fi
            return 0
            ;;
        legacy-backup)
            if [[ "$validation" != "$validation_method" ]]; then
                log error "legacy backup marker lacks content-level validation proof"
                return 1
            fi
            expected_checksum="$(marker_value backup_sha256)"
            if [[ -z "$expected_checksum" ]] || ! validate_archive "$backup_file"; then
                log error "legacy backup marker exists but its archive cannot be verified"
                return 1
            fi
            actual_checksum="$(sha256sum "$backup_file" | awk '{print $1}')"
            if [[ "$actual_checksum" != "$expected_checksum" ]]; then
                log error "legacy backup checksum does not match the migration marker"
                return 1
            fi
            return 0
            ;;
        *)
            log error "legacy migration marker lacks content-level validation proof"
            return 1
            ;;
    esac
}

mkdir -p "$data_dir"

if [[ -f "$marker_file" ]]; then
    if validate_existing_marker; then exit 0; fi
    exit 1
fi

if ! has_existing_data; then
    write_marker no-legacy-data "" no-data-v1
    log info "no existing InfluxDB data detected; recorded no-data migration state"
    exit 0
fi

mkdir -p "$backup_dir"

if [[ -e "$backup_file" ]] && validate_backup_against_source "$backup_file"; then
    log info "using existing content-verified InfluxDB backup at ${backup_file}"
else
    if [[ -e "$backup_file" ]]; then
        log warning "existing InfluxDB backup is stale, incomplete, or invalid and will be atomically replaced"
    fi

    log warning "creating offline InfluxDB backup at ${backup_file}"
    tar -C "$(dirname "$data_dir")" -czf "$temporary_backup" "$archive_root"

    if ! validate_backup_against_source "$temporary_backup"; then
        log error "created InfluxDB backup failed content-level validation"
        exit 1
    fi

    mv -f "$temporary_backup" "$backup_file"
    log info "InfluxDB backup created and content-verified at ${backup_file}"
fi

backup_checksum="$(sha256sum "$backup_file" | awk '{print $1}')"
write_marker legacy-backup "$backup_checksum" "$validation_method"
log info "InfluxDB 2.9 migration marker created with content-validation method and backup checksum"