Initial commit: Open sourcing all of the Maple Open Technologies code.

cloud/maplefile-backend/internal/repo/collection/anonymize_old_ips.go

@@ -0,0 +1,76 @@
+// monorepo/cloud/maplefile-backend/internal/repo/collection/anonymize_old_ips.go
+package collection
+
+import (
+	"context"
+	"time"
+
+	"github.com/gocql/gocql"
+	"go.uber.org/zap"
+)
+
+// AnonymizeOldIPs anonymizes IP addresses in collection tables older than the cutoff date
+func (impl *collectionRepositoryImpl) AnonymizeOldIPs(ctx context.Context, cutoffDate time.Time) (int, error) {
+	totalAnonymized := 0
+
+	// Anonymize collections_by_id table (primary table)
+	count, err := impl.anonymizeCollectionsById(ctx, cutoffDate)
+	if err != nil {
+		impl.Logger.Error("Failed to anonymize collections_by_id",
+			zap.Error(err),
+			zap.Time("cutoff_date", cutoffDate))
+		return totalAnonymized, err
+	}
+	totalAnonymized += count
+
+	impl.Logger.Info("IP anonymization completed for collection tables",
+		zap.Int("total_anonymized", totalAnonymized),
+		zap.Time("cutoff_date", cutoffDate))
+
+	return totalAnonymized, nil
+}
+
+// anonymizeCollectionsById processes the collections_by_id table
+func (impl *collectionRepositoryImpl) anonymizeCollectionsById(ctx context.Context, cutoffDate time.Time) (int, error) {
+	count := 0
+
+	// Query all collections (efficient primary key scan, no ALLOW FILTERING)
+	query := `SELECT id, created_at, ip_anonymized_at FROM maplefile.collections_by_id`
+	iter := impl.Session.Query(query).WithContext(ctx).Iter()
+
+	var id gocql.UUID
+	var createdAt time.Time
+	var ipAnonymizedAt *time.Time
+
+	for iter.Scan(&id, &createdAt, &ipAnonymizedAt) {
+		// Filter in application code: older than cutoff AND not yet anonymized
+		if createdAt.Before(cutoffDate) && ipAnonymizedAt == nil {
+			// Update the record to anonymize IPs
+			updateQuery := `
+				UPDATE maplefile.collections_by_id
+				SET created_from_ip_address = '',
+				    modified_from_ip_address = '',
+				    ip_anonymized_at = ?
+				WHERE id = ?
+			`
+			if err := impl.Session.Query(updateQuery, time.Now(), id).WithContext(ctx).Exec(); err != nil {
+				impl.Logger.Error("Failed to anonymize collection record",
+					zap.String("collection_id", id.String()),
+					zap.Error(err))
+				continue
+			}
+			count++
+		}
+	}
+
+	if err := iter.Close(); err != nil {
+		impl.Logger.Error("Error during collections_by_id iteration", zap.Error(err))
+		return count, err
+	}
+
+	impl.Logger.Debug("Anonymized collections_by_id table",
+		zap.Int("count", count),
+		zap.Time("cutoff_date", cutoffDate))
+
+	return count, nil
+}