refresh_data.sh

#!/bin/bash
# ============================================================
# VI Portal Dashboard — Data Refresh Script
# Run this when new quarterly PDFs are added
# ============================================================

echo "🔄 VI Portal Data Refresh"
echo "========================="

# Configuration
PDF_DIR="${1:-.}"           # Directory containing VI Portal PDFs
DATA_DIR="data"             # Output data directory
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"

echo "📂 PDF Directory: $PDF_DIR"
echo "📊 Output: $DATA_DIR/"

# Count PDFs
PDF_COUNT=$(ls -1 "$PDF_DIR"/VI_Portal*.pdf 2>/dev/null | wc -l)
echo "📄 Found $PDF_COUNT PDF files"

if [ "$PDF_COUNT" -eq 0 ]; then
    echo "❌ No VI Portal PDFs found in $PDF_DIR"
    echo "   Place files like VI_Portal31032025.pdf in the directory"
    exit 1
fi

# Run extraction
echo ""
echo "⚙️  Extracting data..."
python3 "$SCRIPT_DIR/extract_data.py" "$PDF_DIR" "$DATA_DIR"

if [ $? -eq 0 ]; then
    echo ""
    echo "✅ Data refresh complete!"
    echo "   JSON: $DATA_DIR/vi_portal_data.json"
    echo "   CSV:  $DATA_DIR/vi_portal_kpis.csv"
    echo ""
    echo "🚀 To start the dashboard:"
    echo "   streamlit run app.py"
else
    echo "❌ Extraction failed. Check the error messages above."
    exit 1
fi