#!/usr/bin/env bash
# oxpulse-partner-edge-ru-subnets-update — Refresh the RU-subnet list used by split-routing.
#
# Sources:
#   1. RIPE delegated-ripencc-extended-latest (primary)
#   2. ipdeny.com RU zone file (fallback)
#
# On success: atomically replaces OUTFILE and reloads the live nft set.
# On failure: keeps the last-good OUTFILE intact (graceful degradation).
#
# NOTE: this script runs in system.slice (unmarked = direct egress) under systemd.
# On an RU edge, direct foreign egress may be ТСПУ-throttled.  That is why:
#   - Both RIPE and ipdeny fallback are tried (resilience to single-source throttle).
#   - Persistent=true in the timer means a missed daily run retries on next boot.
#   - keep-last-good ensures a throttled fetch leaves the routing table intact.
#
# RIPE count→CIDR: the delegated-stats `count` field is NOT always a power of 2
# (legacy ERX merges, aggregated sub-allocations).  Power-of-2 rounding would
# over-cover adjacent non-RU space and route it direct (un-tunneled) into ТСПУ.
# Fix: awk emits RANGE lines; python3 uses summarize_address_range to produce the
# exact minimal aligned CIDR set covering [start, start+count-1] with no over-cover.
#
# Variables overridable for testing:
#   OUTFILE        destination path (default /etc/oxpulse-partner-edge/ru-subnets.txt)
#   MIN_ENTRIES    minimum aggregate count before accepting update (default 1000)
#   RIPE_FIXTURE   when set, used as the curl-piped content for RIPE (test shim)
#   IPDENY_FIXTURE when set, used as the curl-piped content for ipdeny (test shim)
set -euo pipefail

OUTFILE="${OUTFILE:-/etc/oxpulse-partner-edge/ru-subnets.txt}"
MIN_ENTRIES="${MIN_ENTRIES:-1000}"
RIPE_URL="https://ftp.ripe.net/pub/stats/ripencc/delegated-ripencc-extended-latest"
IPDENY_URL="https://www.ipdeny.com/ipblocks/data/countries/ru.zone"

# Tmp file in same dir as OUTFILE for atomic mv (same filesystem → rename, not copy).
OUTDIR="$(dirname "$OUTFILE")"
TMPFILE="${OUTDIR}/.ru-subnets.tmp.$$"

cleanup() {
    rm -f "$TMPFILE"
}
trap cleanup EXIT

# ── fetch raw lines ─────────────────────────────────────────────────────────
RAW_FILE="$(mktemp /tmp/ru-ripe-raw-XXXXXX)"
trap 'rm -f "$RAW_FILE"; cleanup' EXIT

SOURCE=""

# Primary: RIPE delegated stats → awk emits "RANGE <start_ip> <count>" per RU ipv4 row.
# count is NOT assumed to be a power of 2 — python3 does exact range decomposition via
# summarize_address_range(start, start+count-1) to avoid over-covering non-RU space.
if curl -sf --proto '=https' --tlsv1.2 --max-time 60 "$RIPE_URL" \
    | awk -F'|' '/^ripencc\|RU\|ipv4\|/ { print "RANGE " $4 " " $5 }' \
    > "$RAW_FILE" && [ -s "$RAW_FILE" ]; then
    SOURCE="RIPE"
elif curl -sf --proto '=https' --tlsv1.2 --max-time 60 "$IPDENY_URL" \
    > "$RAW_FILE" && [ -s "$RAW_FILE" ]; then
    SOURCE="ipdeny"
else
    echo "ERROR: Both RIPE and ipdeny downloads failed" >&2
    exit 1
fi

# ── python3 collapse_addresses aggregation ──────────────────────────────────
# Handles two line formats:
#   RANGE <start_ip> <count>  — from RIPE; exact range decomposition
#   <cidr>                    — from ipdeny; fed directly to ip_network
PYAGG="$(mktemp /tmp/ru-aggregate-XXXXXX.py)"
trap 'rm -f "$RAW_FILE" "$PYAGG"; cleanup' EXIT

cat > "$PYAGG" << 'PYEOF'
import sys
import ipaddress

networks = []
for line in sys.stdin:
    line = line.strip()
    if not line or line.startswith("#"):
        continue
    if line.startswith("RANGE "):
        # RIPE format: exact range decomposition via summarize_address_range.
        # Avoids power-of-2 rounding that would over-cover adjacent non-RU space.
        _, start_str, count_str = line.split()
        start = ipaddress.ip_address(start_str)
        count = int(count_str)
        end = ipaddress.ip_address(int(start) + count - 1)
        try:
            networks.extend(ipaddress.summarize_address_range(start, end))
        except (ValueError, TypeError):
            pass
    else:
        # ipdeny format: plain CIDR — fed directly.
        try:
            networks.append(ipaddress.ip_network(line, strict=False))
        except ValueError:
            pass

aggregated = list(ipaddress.collapse_addresses(networks))
for net in sorted(aggregated):
    print(net)
PYEOF

mkdir -p "$OUTDIR"
python3 "$PYAGG" < "$RAW_FILE" > "$TMPFILE"

# ── MIN_ENTRIES guard ────────────────────────────────────────────────────────
COUNT=$(wc -l < "$TMPFILE")

if [ "$COUNT" -lt "$MIN_ENTRIES" ]; then
    echo "ERROR: Only $COUNT entries after aggregation (minimum $MIN_ENTRIES). Aborting. Last-good file preserved." >&2
    exit 1
fi

# ── atomic mv to OUTFILE ─────────────────────────────────────────────────────
mv -f "$TMPFILE" "$OUTFILE"
chmod 0644 "$OUTFILE"

echo "OK: $COUNT aggregated RU subnets written to $OUTFILE (source: $SOURCE)"

# ── reload the live nft set ──────────────────────────────────────────────────
# try-restart is idempotent: no-op if the unit is not active; re-apply is idempotent.
# Do NOT hard-fail if the unit is absent (edge may not have split-routing active yet).
# stderr NOT suppressed: failures land in the journal for operator diagnostics.
systemctl try-restart oxpulse-partner-edge-split-routing.service || true