CSV converter: verification companion

CSV converter: edge cases and repeatable checks

This companion to the CSV conversion example on my profile focuses on its test suite. Both entries describe the same self-directed project, developed and reviewed with AI assistance. All sample records are fictional.

The tests run the actual Python command that a user would run. They check leading-zero identifiers, quoted commas and newlines, Unicode, optional UTF-8 BOMs and empty datasets. Invalid headers, wrong field counts, unfinished quotes and invalid UTF-8 must return a clear error before creating an output file.

The suite also checks that input bytes remain unchanged and existing output files are refused. The example supports its stated schema up to 20,000 rows and 10 MiB; different schemas need an agreed mapping and changes to the script.

The published September 8 Ubuntu verification records 13 passing tests, plus a 20,000-row CRLF check and an output comparison. The verification record includes file hashes so the result can be matched to the source. These are dated results, with source and run instructions available for review.

Tests: https://github.com/SovNodeAI/csv-json-cleanup-example/blob/main/test_convert.py

Verification record: https://github.com/SovNodeAI/csv-json-cleanup-example/blob/main/VERIFICATION.json