#!/usr/bin/env bash
set -euo pipefail

# Export Concelier linksets/advisory_chunks from Postgres to a tar.zst bundle.
# Usage:
#   PGURI=postgres://user:pass@host:5432/db \
#   ./scripts/concelier/export-linksets-tarball.sh out/linksets/linksets-stage-backfill.tar.zst
#
# Optional env:
#   PGSCHEMA=public                # schema that owns linksets/advisory_chunks
#   LINKSETS_TABLE=linksets        # table name for linksets
#   CHUNKS_TABLE=advisory_chunks   # table name for advisory chunks
#   TMPDIR=/tmp/export-linksets    # working directory (defaults to mktemp)

TARGET="${1:-}"
if [[ -z "${TARGET}" ]]; then
  echo "Usage: PGURI=... $0 out/linksets/linksets-stage-backfill.tar.zst" >&2
  exit 1
fi

if [[ -z "${PGURI:-}" ]]; then
  echo "PGURI environment variable is required (postgres://...)" >&2
  exit 1
fi

PGSCHEMA="${PGSCHEMA:-public}"
LINKSETS_TABLE="${LINKSETS_TABLE:-linksets}"
CHUNKS_TABLE="${CHUNKS_TABLE:-advisory_chunks}"
WORKDIR="${TMPDIR:-$(mktemp -d)}"

mkdir -p "${WORKDIR}"
OUTDIR="$(dirname "${TARGET}")"
mkdir -p "${OUTDIR}"

echo "==> Exporting linksets from ${PGSCHEMA}.${LINKSETS_TABLE}"
psql "${PGURI}" -c "\copy (select row_to_json(t) from ${PGSCHEMA}.${LINKSETS_TABLE} t) to '${WORKDIR}/linksets.ndjson'"

echo "==> Exporting advisory_chunks from ${PGSCHEMA}.${CHUNKS_TABLE}"
psql "${PGURI}" -c "\copy (select row_to_json(t) from ${PGSCHEMA}.${CHUNKS_TABLE} t) to '${WORKDIR}/advisory_chunks.ndjson'"

LINKSETS_COUNT="$(wc -l < "${WORKDIR}/linksets.ndjson")"
CHUNKS_COUNT="$(wc -l < "${WORKDIR}/advisory_chunks.ndjson")"

echo "==> Writing manifest.json"
jq -n --argjson linksets "${LINKSETS_COUNT}" --argjson advisory_chunks "${CHUNKS_COUNT}" \
  '{linksets: $linksets, advisory_chunks: $advisory_chunks}' \
  > "${WORKDIR}/manifest.json"

echo "==> Building tarball ${TARGET}"
tar -I "zstd -19" -cf "${TARGET}" -C "${WORKDIR}" linksets.ndjson advisory_chunks.ndjson manifest.json

echo "==> SHA-256"
sha256sum "${TARGET}"

echo "Done. Workdir: ${WORKDIR}"