export

GDPR Data Export Example

Export every event belonging to a data subject — the GDPR right to access (Art. 15) and data portability (Art. 20).

When a user invokes their GDPR rights, you must produce a complete, portable copy of the data you hold on them. In an event-sourced system that data is spread across many streams, so go-mink's DataExporter gathers it by explicit stream ID or by scanning with filters, decrypts protected fields, and returns a structured result. It also handles the awkward case where a key was already crypto-shredded: those events come back marked Redacted rather than crashing the export.

What this demonstrates

• Stream-based export — pass known stream IDs in ExportRequest.Streams for an efficient, scan-free export.
• Scan-based export with filters — FilterByTenantID, FilterByEventTypes, FilterByStreamPrefix, FilterByUserID, and CombineFilters select a subject's events when you don't know the stream IDs.
• Streaming export — ExportStream invokes a callback per event so large exports never materialize the whole result in memory.
• Time-range filtering — FromTime / ToTime scope an export to a date window.
• Crypto-shredding + export — after provider.RevokeKey, encrypted events export as Redacted (nil Data, RedactedCount incremented) instead of failing.

Running

go run ./examples/export

No infrastructure required — encryption keys and the event store are in-memory (local.New + memory.NewAdapter).

What happens

The example seeds five events across two tenants (Alice under tenant A, Bob under tenant B), with email/phone encrypted, then runs five demos:

1. Stream-based export — Alice's three known streams are exported by ID. It prints the subject, total event count, stream list, export timestamp, and each decrypted event.
2. Scan-based export — Export scans all events with filters and prints counts for: all tenant-A events, tenant-A OrderPlaced events (via CombineFilters), all Customer- streams (via FilterByStreamPrefix), and everything for user alice-1 (via FilterByUserID).
3. Streaming export — ExportStream walks two of Alice's streams, printing one line per streamed event and the final count.
4. Time-range export — one export limited to the last hour (FromTime) and one limited to events before a future cutoff (ToTime), printing each matching count.
5. Crypto-shredding + export — Bob's data exports cleanly first; then provider.RevokeKey("tenant-B") is called and the re-export reports the encrypted CustomerCreated as [REDACTED] while the unencrypted OrderPlaced still exports as [OK].

Key APIs

• mink.NewDataExporter(store, opts...) — construct an exporter over an event store.
• mink.WithExportBatchSize(n) — events loaded per batch during scan-based export (default 1000).
• mink.ExportRequest{...} — describes the export: SubjectID, Streams, Filter, FromTime, ToTime.
• exporter.Export(ctx, req) — returns an *ExportResult with Events, Streams, TotalEvents, RedactedCount, and ExportedAt.
• exporter.ExportStream(ctx, req, handler) — memory-efficient export invoking handler per ExportedEvent.
• mink.FilterByTenantID(id) / mink.FilterByEventTypes(types...) / mink.FilterByStreamPrefix(prefix) / mink.FilterByUserID(id) — built-in export filters.
• mink.CombineFilters(filters...) — AND-compose multiple filters.
• provider.RevokeKey(keyID) — crypto-shred a tenant's key; subsequent exports redact its encrypted events.

Related

• Examples: encryption · full-ecommerce
• Docs: Security · API reference