manabendu-ai/pdf-profit-parser-pro

Manual changes saved

The current implementation is only a static frontend demo and is incomplete. Extend this application into a fully functional end-to-end system by implementing and wiring a real Python backend that executes on file upload. When a user uploads a PDF, the application must immediately invoke Python processing logic (not mock data) to parse the uploaded UK bank statement, detect the issuing bank, extract real transaction data, and return results dynamically to the UI. Use Python libraries such as pdfplumber, pdf2image, pytesseract, regex, and pandas to process both digital and scanned PDFs. The backend must extract only true transactions and must strictly ignore balances, summaries, opening/closing balances, or running totals. Each transaction must use separate debit and credit fields, never an amount + direction model, ensuring that exactly one of debit or credit is populated per transaction. The backend must detect all UK banks appearing in the document (issuer and counterparties), determine the issuer bank, apply category rules, calculate VAT only on eligible debit transactions using HMRC-aligned logic, and generate detailed audit explanations for each transaction. The frontend must be updated to display the real extracted transactions in a table with explicit Debit and Credit columns, show detected bank information, display VAT and audit details, and allow users to download a real Excel file generated by the backend containing transaction data and VAT summaries. Remove all placeholder text, demo warnings, and static UI-only behavior, and ensure the upload button triggers actual backend computation and returns real results.

initial commit