#!/usr/bin/env bash

# Use rclone to sync the last two good Wikimedia XML data dumps.

set -euxo pipefail

if [ -z "$1" ]; then
    echo "Please supply an rclone remote"
    exit 1
fi

RCLONE_REMOTE="$1"
MIRROR_URL=$(rclone config show "$RCLONE_REMOTE" | grep "url" | awk '{print $3}')
NTFY_IP=$(sudo tailscale status | grep "dietpi" | awk '{print $1}')

cleanup() {
    echo "Cleaning up"
    rm -fv "${HOME}/rsync-filelist-last-2-good.txt"
    rm -fv "${HOME}/rsync-filelist-last-2-good-en.txt"
    exit
}

trap cleanup 0 1 2 3 6

wget "${MIRROR_URL}/rsync-filelist-last-2-good.txt" \
    -O "${HOME}/rsync-filelist-last-2-good.txt"

grep "enwiki" "${HOME}/rsync-filelist-last-2-good.txt" |
    grep -v "tenwiki" |
    tee "${HOME}/rsync-filelist-last-2-good-en.txt"

rclone sync \
    --http-no-head \
    -P \
    --transfers 16 \
    --include-from "${HOME}/rsync-filelist-last-2-good-en.txt" "${RCLONE_REMOTE}:" \
    /naspool/archives/wikimedia-xmldatadumps-en

curl \
    -H prio:default \
    -H tags:incoming_envelope \
    -d "Syncing of wikimedia xml datadumps succeeded" \
    "http://${NTFY_IP}:8080/wikimedia_xmldatadumps_en"

exit 0