#!/usr/bin/env bash
# Downloads BIRD financial SQLite into ctrl/seed/data/financial.sqlite and
# the golden eval set into api/evals/data/. Runs on the HOST (the api image
# deliberately doesn't include the data). Override BIRD_URL if upstream
# moves.
#
# Layout inside the outer dev.zip (as of the 2024-06-27 dump):
#   dev_<date>/
#     dev.sql, dev.json, dev_tables.json, dev_tied_append.json
#     dev_databases.zip  → financial/financial.sqlite (and 10 other DBs)
set -euo pipefail
cd "$(dirname "$0")"

DEST="data/financial.sqlite"
EVAL_DIR="../../api/evals/data"
mkdir -p data "$EVAL_DIR"

if [ -f "$DEST" ]; then
    echo "already have $DEST ($(du -h "$DEST" | cut -f1))"
    exit 0
fi

URL="${BIRD_URL:-https://bird-bench.oss-cn-beijing.aliyuncs.com/dev.zip}"
TMP="$(mktemp -d)"
trap 'rm -rf "$TMP"' EXIT
OUTER="$TMP/dev.zip"

echo "fetching $URL (~330MB compressed; full dev set is ~3GB extracted)..."
curl -fL --progress-bar -o "$OUTER" "$URL"

echo "extracting outer archive..."
unzip -q "$OUTER" -d "$TMP"
ROOT=$(find "$TMP" -maxdepth 1 -type d -name 'dev_*' | head -1)
test -n "$ROOT" || { echo "could not locate dev_* root inside outer zip" >&2; exit 1; }

echo "extracting financial.sqlite from inner archive..."
unzip -q -j "$ROOT/dev_databases.zip" '*/financial/financial.sqlite' -d data/

echo "copying golden eval set to $EVAL_DIR/..."
cp "$ROOT/dev.json" "$ROOT/dev.sql" "$ROOT/dev_tables.json" "$EVAL_DIR/"

echo "ready:"
echo "  $DEST ($(du -h "$DEST" | cut -f1))"
echo "  $(ls -1 "$EVAL_DIR/" | wc -l) files in $EVAL_DIR/"