DataRescue
Repair messy CSV/JSON without losing track of a single row.
- Type
- Open-source tool
- Interface
- Workbench + CLI
- Data
- CSV / JSON
- Repo
- vanovri/datarescue
Problem
Messy CSV and JSON files need validating and repairing before they are useful, and every change should be traceable.
Approach
InputCSV / JSON files
RulesRule-based normalization
QuarantineRejected rows set aside
AuditReport exports
Build gallery
Results
- Fewer manual validation errors: rule-based validation and normalization replace hand-checking of CSV/JSON rows.
- Nothing silently lost: rejected rows are quarantined instead of dropped, so every repair can be reviewed and reversed.
- Traceable changes: exported audit reports record what was changed and what was rejected.
- Reproducible pipeline: the same rules run from the local workbench or the CLI, and behaviour is covered by automated tests (AI-assisted).
Lessons
- Keeping rejected rows instead of dropping them makes repairs auditable.