Ledger

import pandas as pd
import requests
from invoice_vision_model import DocumentParser
from fuzzywuzzy import fuzz

def process_unstructured_invoices(pdf_batch_path):
    parser = DocumentParser(model='layout-lmv3-finance')
    extracted_data = []
    
    for pdf_file in os.listdir(pdf_batch_path):
        # 1. Vision & NLP Extraction
        raw_json = parser.extract(os.path.join(pdf_batch_path, pdf_file))
        
        # 2. Math Validation & Cleansing
        net_val = float(raw_json.get('net_amount', 0))
        vat = float(raw_json.get('tax_amount', 0))
        total = float(raw_json.get('total_amount', 0))
        
        if abs((net_val + vat) - total) > 0.05:
            raise ValueError(f"Audit failed for {pdf_file}: Math incongruence.")

        # 3. Fuzzy match for Supplier ID on ERP Database
        supplier_id = match_erp_supplier(raw_json['vendor_name'])
        
        extracted_data.append({
            'erp_vendor_id': supplier_id,
            'invoice_num': raw_json['invoice_id'],
            'total_eur': total,
            'due_date': raw_json['due_date']
        })
        
    # 4. Bulk API Injection
    df = pd.DataFrame(extracted_data)
    response = requests.post('https://api.erp-system.com/v2/invoices/bulk',
                             headers={'Auth': 'Bearer XXXXX'},
                             json=df.to_dict(orient='records'))
    
    return response.status_code