Dry-run validation before load

Two tools scan existing data for violations without changing anything. Run them as a pair after every bulk mutation, and to vet data before you rely on it.

The two validators

eddytor validate table eddytor sales orders   # checks for constraint violations via SQL

validate_constraints(table)                          # PK, NOT NULL, CHECK expressions
validate_domain_values(table, columns=["status"])   # domain mismatches + typo suggestions

• validate_constraints - checks primary keys, NOT NULL, and CHECK expressions. Returns violations with sample rows.
• validate_domain_values - scans constrained columns for values outside their domain, and returns similarity suggestions (e.g. "actve" → "active", "Electroncis" → "Electronics").

The fix loop

1. validate_constraints + validate_domain_values.
2. Fix flagged rows with merge_rows (use the typo suggestions).
3. Re-run both.
4. profile_table for a final sanity check.

Catch orphans from reference domains

Reference domains validate live against the source table - if someone deletes a source value, dependent rows are orphaned. validate_domain_values finds them.

Next

• Bulk corrections safely · Profile a table.