Projects & Experiments

# Validate a PDF is a real GST invoice, not a plain itinerary
def _is_valid_invoice(text):
    invoice_no = _first_match(r"(?:Invoice No)[:\s]*([A-Z0-9]{8,})", text, re.I)
    if not invoice_no or re.match(r"^[0-9]{2}[A-Z]{5}", invoice_no):
        return False  # reject GSTIN-shaped strings
    if not re.search(r"(?:GSTIN|GST)[:\s]*[A-Z0-9]{15}", text, re.I):
        return False
    has_customer_gstin = bool(re.search(r"GSTIN of Customer", text, re.I))
    has_pnr = bool(re.search(r"PNR[:\s]*[A-Z0-9]{6}", text, re.I))
    return has_customer_gstin or has_pnr