Go data mapping review

Explicit Mapping

Import, extraction, and mapping code should make user intent explicit. Prefer canonical domain keys when source data already has them. When a source uses non-canonical headers, aliases, prior-year inputs, or external workspace data, require a versioned profile, schema metadata, flag, or other configured mapping with documented override order.

Unknown, missing, or ambiguous mappings should fail with deterministic diagnostics rather than silently guessing. A review should flag substring matching, fuzzy aliases, and fallback rules that can silently import the wrong field.

Bad:

func mapColumn(name string) string {
	if strings.Contains(strings.ToLower(name), "date") {
		return "posting_date"
	}
	return name
}

Better:

func mapColumn(profile Profile, name string) (FieldID, error) {
	field, ok := profile.Columns[name]
	if !ok {
		return "", fmt.Errorf("profile %q: unknown column %q", profile.Name, name)
	}
	return field, nil
}

Guessing can silently import the wrong data. A configured mapping makes the user’s intent reviewable.

← Source transforms and evaluators Guide index Control flow and mutation →

Explicit Mapping

Sources