#!/bin/bash

# Delta data generator quick test script

echo "=== Delta Data Generator Test ==="

# Check Python environment
python3 -c "import pandas, pyarrow, deltalake; print('✅ Dependency check passed')" || {
    echo "❌ Dependency check failed, please run: pip install -r requirements_generator.txt"
    exit 1
}

echo "Starting to generate test data..."

# Generate small batch test data
python3 generate_delta_data.py \
    --num-hosts 10 \
    --records-per-host 5 \
    --interval-seconds 1 \
    --duplicate-rate 0.2 \
    --single-batch \
    --output-dir ./test_delta

echo ""
echo "=== Test Completed ==="
echo ""
echo "Generated data saved in: ./test_delta/statement_logs"
echo ""
echo "View first few records:"
python3 -c "
from deltalake import DeltaTable
import pandas as pd

try:
    dt = DeltaTable('./test_delta/statement_logs')
    df = dt.to_pandas()
    print(f'Total records: {len(df)}')
    print('First 5 records preview:')
    print(df[['schema_name', 'stmt_type', 'exec_count', 'sum_latency']].head())
except Exception as e:
    print(f'Failed to read data: {e}')
    print('Please check the generated data files')
"
