#!/bin/bash
set -e

DATA_DIR="/home/coder/private/sanjay-data"
PROJECT="chalk-develop"
DATASET="tpch_test"
PROJECT_DATASET="${PROJECT}:${DATASET}"
SCALES=("0.1" "1.0" "10.0" "100.0")
TABLES=("customer" "lineitem" "nation" "orders" "part" "partsupp" "region" "supplier")

# Create dataset if it doesn't exist.
bq mk --dataset "${PROJECT_DATASET}" 2>/dev/null || echo "Dataset ${DATASET} already exists."

for scale in "${SCALES[@]}"; do
  scale_underscore="${scale//./_}"
  for table in "${TABLES[@]}"; do
    bq_table="${PROJECT_DATASET}.tpch-${scale_underscore}-${table}"
    file="${DATA_DIR}/tpch-${scale}/${table}.parquet"
    if bq show "${bq_table}" &>/dev/null; then
      echo "Skipping ${bq_table} (already exists)."
    else
      echo "Loading ${file} -> ${bq_table}"
      bq load --source_format=PARQUET "${bq_table}" "${file}"
    fi
  done
done

echo "Done. Uploaded all tables to ${PROJECT_DATASET}."