#!/usr/bin/env bash
#
# Refresh the per-country CIDR lists that functions_geo.php reads.
#
#   geo-update.sh pt          # one or more ISO country codes
#   geo-update.sh             # falls back to $GEO_ALLOW_COUNTRIES
#
# Source is ipdeny.com, which publishes one aggregated zone file per country
# from the RIR allocation data. Allocations move slowly, so monthly is plenty:
#
#   17 4 3 * *  /srv/sites/<site>/shared/platform-tools/bin/geo-update.sh pt
#
# The files are committed to the repo. This script only refreshes them, so a
# deploy on a box that cannot reach ipdeny still has working data — it is just
# as current as the last commit. That is the whole reason for committing them.
#
# Nothing is overwritten unless the download passed every check: a truncated
# file, an HTML error page served with 200, or a suspiciously short list would
# otherwise turn into "no visitor is in Portugal" — which, with GEO_FAIL_CLOSED,
# locks the country out of its own site.

set -euo pipefail

DATA_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)/data/geo"
V4_URL="https://www.ipdeny.com/ipblocks/data/countries"
V6_URL="https://www.ipdeny.com/ipv6/ipaddresses/blocks"

# A real country zone has hundreds of ranges. Anything under this is a bad
# download, not a small country -- the smallest national allocations still run
# to dozens of blocks, and we would rather keep yesterday's file.
MIN_V4_LINES=20
MIN_V6_LINES=1

countries=("$@")
if [ ${#countries[@]} -eq 0 ]; then
    read -r -a countries <<< "${GEO_ALLOW_COUNTRIES//[,;]/ }"
fi
if [ ${#countries[@]} -eq 0 ]; then
    echo "uso: $(basename "$0") <cc> [cc...]   (ou defina GEO_ALLOW_COUNTRIES)" >&2
    exit 2
fi

mkdir -p "$DATA_DIR"
tmp="$(mktemp -d)"
trap 'rm -rf "$tmp"' EXIT

status=0

fetch_zone() {
    local cc="$1" family="$2" url="$3" min="$4"
    local out="$DATA_DIR/${cc}.${family}"
    local work="$tmp/${cc}.${family}"

    if ! curl -fsS --max-time 60 -o "$work" "$url"; then
        echo "  ${cc}.${family}: download falhou -- fica o ficheiro anterior" >&2
        return 1
    fi

    # Every line must be a bare CIDR. An HTML error page fails here, which is
    # the check that actually matters: those arrive with a 200.
    if grep -qvE '^[0-9a-fA-F:.]+/[0-9]{1,3}$' "$work"; then
        echo "  ${cc}.${family}: conteúdo não é uma lista de CIDR -- ignorado" >&2
        return 1
    fi

    local n
    n=$(wc -l < "$work")
    if [ "$n" -lt "$min" ]; then
        echo "  ${cc}.${family}: só $n linhas (mínimo $min) -- ignorado" >&2
        return 1
    fi

    sort -V -o "$work" "$work"

    if [ -f "$out" ] && cmp -s "$work" "$out"; then
        echo "  ${cc}.${family}: sem alterações ($n)"
        return 0
    fi

    mv "$work" "$out"
    echo "  ${cc}.${family}: actualizado ($n ranges)"
}

for cc in "${countries[@]}"; do
    cc="$(echo "$cc" | tr '[:upper:]' '[:lower:]')"
    [[ "$cc" =~ ^[a-z]{2}$ ]] || { echo "  '$cc' não é um código ISO -- ignorado" >&2; status=1; continue; }

    echo "$cc:"
    fetch_zone "$cc" v4 "$V4_URL/${cc}.zone"  "$MIN_V4_LINES" || status=1
    # IPv6 is published separately and some countries have no file at all; a
    # missing v6 list is not an error, the v4 list still gates those visitors.
    fetch_zone "$cc" v6 "$V6_URL/${cc}.zone"  "$MIN_V6_LINES" || true
done

exit $status