Release v1.0.12

Release 1.0.12: chunked processing, transaction per chunk, supportsBatch/anonymizeBatch

Changelog

Added

• Chunked processing and bounded memory: AnonymizeService::anonymizeEntity() now loads records in chunks of batch_size (LIMIT/OFFSET with ORDER BY primary key) instead of loading all rows at once. Memory usage is bounded by the configured batch size. A single COUNT(*) query is run once for progress reporting.
• Transaction per chunk: All updates for a chunk are committed in one transaction (one commit per chunk instead of per row), improving throughput when using a custom anonymizer or property-based anonymization.
• EntityAnonymizerServiceInterface: Two new methods for custom anonymizer services:
  • supportsBatch(): bool — Return true to have the bundle call anonymizeBatch() once per chunk; false to call anonymize() for each record.
  • anonymizeBatch(EntityManagerInterface $em, ClassMetadata $metadata, array $records, bool $dryRun): array — Receives the full chunk of records and returns a map record index => (column => new value). Only called when supportsBatch() returns true.

Changed

• EntityAnonymizerServiceInterface (breaking): The interface now requires three methods: supportsBatch(), anonymize(), and anonymizeBatch(). Existing implementations must add supportsBatch() and anonymizeBatch(). For record-by-record behavior, return false from supportsBatch() and implement anonymizeBatch() e.g. by delegating to anonymize() per record (see demos and USAGE.md).

Fixed

(none)

Documentation

• CONFIGURATION.md: batch_size description updated to state that it limits rows loaded per query (memory) and that one transaction is committed per chunk (performance).
• USAGE.md: Section "Anonymizing via a custom service" updated with the three interface methods and the record-by-record vs batch behaviour.

---

Commits

• Release 1.0.12 - Chunked processing, EntityAnonymizerServiceInterface batch support (d858b0b)