from PyPDF2 import PdfMerger, PdfReader
import os

# Directory containing the invoice PDFs
invoices_dir = "invoices"

# Get all PDF files and sort them
pdf_files = [f for f in os.listdir(invoices_dir) if f.endswith('.pdf')]
pdf_files.sort()

# Create a PDF merger object
merger = PdfMerger()

# Add each PDF to the merger
for pdf_file in pdf_files:
    pdf_path = os.path.join(invoices_dir, pdf_file)

    # For Audible PDFs, only include the first page
    if "Audible" in pdf_file:
        merger.append(pdf_path, pages=(0, 1))
        print(f"Added (first page only): {pdf_file}")
    else:
        merger.append(pdf_path)
        print(f"Added: {pdf_file}")

# Write the combined PDF
output_path = "receipts.pdf"
merger.write(output_path)
merger.close()

print(f"\nCombined {len(pdf_files)} PDFs into {output_path}")