scripts/start.sh

#!/bin/bash
set -x # Print commands for debugging

echo "=================================================="
echo "   RAG-KB-SYSTEM STARTUP DIAGNOSTICS"
echo "=================================================="
echo "Date: $(date)"
echo "User: $(whoami)"
echo "PWD: $(pwd)"
echo "LS /app:"
ls -la /app
echo "LS /app/scripts:"
ls -la /app/scripts
echo "Python version:"
python3 --version

echo "Checking environment variables..."
if [ -z "$HF_DATASET_REPO" ]; then
    echo "CRITICAL WARNING: HF_DATASET_REPO is NOT set!"
else
    echo "HF_DATASET_REPO is set to: $HF_DATASET_REPO"
fi

if [ -z "$HF_TOKEN" ]; then
    echo "CRITICAL WARNING: HF_TOKEN is NOT set!"
else
    echo "HF_TOKEN is set (length: ${#HF_TOKEN})"
fi

echo "=================================================="
echo "   STARTING DATA DOWNLOAD"
echo "=================================================="

# Download persistent data if configured
if [ -n "$HF_DATASET_REPO" ]; then
    echo "Running download_data.py..."
    # Use -u for unbuffered output to ensure logs appear immediately
    python3 -u scripts/download_data.py
    
    echo "Download script finished (exit code: $?)"
    
    echo "Checking downloaded files:"
    ls -lh rag-kb.db || echo "rag-kb.db MISSING"
    ls -lh vector_store/ || echo "vector_store/ MISSING"
else
    echo "Skipping download because HF_DATASET_REPO is unset."
fi

echo "=================================================="
echo "   STARTING NEXT.JS SERVER"
echo "=================================================="
exec node server.js