Null-Safe Inputs And Mapping

When a value is optional (or can differ by feature/region/constructor), never assume it’s present or non-empty. Apply consistent null/empty/whitespace handling, gate context-dependent logic, and validate after any transform (e.g., JSON/model conversion) so missing fields don’t silently change behavior.

Apply this standard:

1. Normalize inputs early: treat whitespace-only strings as null; use string.IsNullOrWhiteSpace for string parameters.
2. Guard required combinations: validate that when one input is provided (e.g., identity type/IDs), the other required fields are not null/empty; throw a clear PSInvalidOperationException (or equivalent) for invalid combinations.
3. Gate context usage: only dereference optional context objects when they’re guaranteed to exist; if not, either skip inference or throw a clear error.
4. Validate after serialization/type conversion: if you bridge models (JSON serialize/deserialize, dictionary casts, etc.), null-check the result and confirm critical fields (e.g., IDs, target details) survived.
5. Preserve existing values unless explicitly overridden: in update flows, don’t overwrite nullable/optional properties with defaults just because the new parameters are absent.

Pattern example (normalization + post-mapping validation):

string changeReference = boundParameters.TryGetValue("-ChangeReference", out var v)
    ? v as string
    : null;
changeReference = string.IsNullOrWhiteSpace(changeReference) ? null : changeReference;

// Example for mapping/bridge where fields can be lost
var json = JsonConvert.SerializeObject(restoreRequest);
var mapped = JsonConvert.DeserializeObject<MyCrrRestoreRequest>(json);
if (mapped == null || mapped.TargetDetails == null || string.IsNullOrEmpty(mapped.SourceResourceId))
{
    throw new InvalidOperationException("Mapping lost required fields (TargetDetails/SourceResourceId). Update the bridge models.");
}

This prevents NullReferenceExceptions, incorrect defaults, and silent behavior changes caused by missing/empty inputs or divergent model mappings.