Spreadsheet Output Diff Checker

Compare a candidate table produced by an AI or automation against an approved expected table.

Scenario and prerequisites

• - Use after formula linting when a reproducible expected output exists.
• Input tables are JSON arrays of redacted row objects with stable row_id values.
• The comparison is deterministic and does not infer acceptable business differences.

Input and output

Input contains expected and candidate arrays. Output is PASS or FAIL, with missing rows, extra rows, missing columns and cell-level mismatches.

Installation and execution

Run python scripts/compare_spreadsheet_outputs.py --input examples/input.json --output examples/output.json. Review each mismatch before changing a fixture or approving a new expected result.

Recovery and boundaries

Duplicate row IDs, missing row IDs and type/value changes are visible findings. The script does not open Excel, evaluate formulas or decide whether a difference is business-safe.

Acceptance

The sample must report one cell mismatch and FAIL. A corrected candidate must produce PASS.