Get a clean file in the layout your next import needs, with rejected rows separated and explained. You also receive a reusable Python command for future exports using the same agreed rules. A suitable job might be mapping a supplier stock export to your product-import columns, or turning a daily CSV / JSON export into one consistent report file. The fixed scope covers one input format, one output format, up to 50,000 rows / 20 MB, 10 fields and five agreed mapping or validation rules. Original files stay intact. Text identifiers such as leading-zero product codes are preserved. Duplicate records follow the rule we agree, and invalid rows are reported with reasons. Delivery includes Python 3 source, a CLI command, an example configuration, an error / summary report, checks for the agreed normal, malformed-input, duplicate and repeat-run cases, and concise setup / run notes. One consolidated in-scope revision is included. Send a small synthetic or redacted sample and describe the result you need. I will propose a bounded mapping / validation plan before we start. Initial delivery is two working days after we agree the scope, expected output, access and payment arrangements. Website scraping, live API access, OCR, production changes and ongoing hosting are outside this service. I work through a Serbian sole proprietorship and use AI tools for implementation, tests and written communication. The linked example is a personal demonstration using synthetic data, with 12 checked Python tests. Verification commands and actual results are included with the handover.