Strip Null Fields for GDPR Data Minimization

A signup captured email only; phone, DOB, and marketing consent were never asked. Default strip (null, Deep on) leaves the minimal collected record.

Input (record.json):
{
  "email": "sam@example.com",
  "phone": null,
  "dob": null,
  "marketingConsent": null
}

Strip Values (targets: null · Deep on):
{
  "email": "sam@example.com"
}

3 values stripped.

A user explicitly declined marketing. That false is real consent data and is preserved, while the unasked phone field is removed.

Input:
{
  "email": "jo@example.com",
  "marketingConsent": false,
  "phone": null
}

Strip Values (targets: null · Deep on):
{
  "email": "jo@example.com",
  "marketingConsent": false
}

false survives; only the null phone is stripped.

An address object exists but every sub-field is null. Ticking null + empty objects removes the children, then the empty address cascades away.

Input:
{
  "email": "li@example.com",
  "address": { "line1": null, "postcode": null }
}

Strip Values (targets: null + empty objects · Deep on):
{
  "email": "li@example.com"
}

3 values stripped (line1, postcode, then empty address).

An empty purchases array means the user was checked and has none — analytically different from never-checked. Leave empty arrays unticked.

Input:
{ "email": "ash@example.com", "purchases": [], "phone": null }

Strip Values (targets: null only · empty arrays UNCHECKED):
{ "email": "ash@example.com", "purchases": [] }

The meaningful empty list is preserved.

For audit samples, mask identifiers first, then strip empties so the shared file is both pseudonymised and minimal.

# 1. Mask emails/names with the anonymiser tool
#    -> see /tool/json-anonymizer
# 2. Drop the masked file here
# 3. Strip Values (null, Deep on) -> Download
# Result: minimal + pseudonymised record set for the DPIA appendix

Stripping operates on the file you drop, not on your database. It models the minimised shape for new writes; it does not satisfy a right-to-erasure request for records already persisted. Use your database delete plus an audit log for that.

If null is a meaningful 'this was cleared' marker rather than 'never collected', stripping it loses that signal. Leave the null checkbox off for those records, or use a sentinel value before stripping.

false is never stripped — there is no falsy mode. A recorded decline ("marketingConsent": false) survives, which is what you want for demonstrating you honoured the choice.

[] can mean the data was gathered and the result is empty, which is itself a fact. Don't tick empty arrays when that distinction matters to a controller or processor record.

Health, biometric, or other Article 9 data in the dropped file is processed entirely in your browser and never uploaded. Still follow your own handling policy for the local file (encryption at rest, etc.).

A line-delimited export with one record per line is not a single JSON document and will not parse. Wrap records into a JSON array first, then strip the array.

A trailing comma or BOM throws on JSON.parse and the run fails. Repair with json-format-fixer before stripping.

A field set to " " is not matched by empty strings (only zero-length "" is). Trim such values upstream if they represent 'not collected'.

Output is 2-space pretty JSON; there is no indent control. If your storage layer wants compact JSON, minify the result afterward with json-minifier.