1use super::*;
2use crate::storage::unified::metadata::{MetadataFilter, MetadataValue};
3
4impl RedDB {
5 pub fn is_replica_role(&self) -> bool {
6 matches!(
7 self.options.replication.role,
8 crate::replication::ReplicationRole::Replica { .. }
9 )
10 }
11
12 pub fn enforce_retention_policy(&self) -> Result<(), Box<dyn std::error::Error>> {
13 if self.options.read_only || self.is_replica_role() {
14 return Ok(());
15 }
16
17 if self.options.mode == StorageMode::Persistent {
20 let Some(path) = self.path() else {
21 return Ok(());
22 };
23
24 let Ok(mut metadata) = self.load_or_bootstrap_physical_metadata(true) else {
25 return Ok(());
26 };
27
28 self.prune_export_registry(&mut metadata.exports);
29 metadata.save_for_data_path(path)?;
30 }
31
32 let _ = self.sweep_ttl_expired_entities()?;
33
34 Ok(())
35 }
36
37 pub(crate) fn ttl_expired_entities_now(
38 &self,
39 ) -> Result<Vec<(String, EntityId)>, Box<dyn std::error::Error>> {
40 self.ttl_expired_entities_at(current_unix_ms())
41 }
42
43 pub fn replica_allows_entity_at_read(
44 &self,
45 collection: &str,
46 entity: &crate::storage::UnifiedEntity,
47 ) -> bool {
48 if !self.is_replica_role() {
49 return true;
50 }
51 !self.entity_expired_at(collection, entity, current_unix_ms())
52 }
53
54 fn sweep_ttl_expired_entities(&self) -> Result<usize, Box<dyn std::error::Error>> {
55 let to_delete = self.ttl_expired_entities_now()?;
56
57 let mut deleted = 0usize;
58 for (collection, id) in to_delete {
59 match self.store.delete(&collection, id) {
60 Ok(true) => deleted = deleted.saturating_add(1),
61 Ok(false) => {}
62 Err(err) => {
63 return Err(format!(
64 "failed deleting expired entity {id} from collection '{collection}': {err:?}"
65 )
66 .into());
67 }
68 }
69 }
70
71 Ok(deleted)
72 }
73
74 fn ttl_expired_entities_at(
75 &self,
76 now_ms: u64,
77 ) -> Result<Vec<(String, EntityId)>, Box<dyn std::error::Error>> {
78 let mut to_delete = Vec::<(String, EntityId)>::new();
79
80 let mut absolute_expired = self.expired_entities_by_expires_at(now_ms)?;
81 to_delete.append(&mut absolute_expired);
82
83 let mut relative_expired = self.expired_entities_by_ttl(now_ms)?;
84 to_delete.append(&mut relative_expired);
85
86 to_delete.sort_unstable();
87 to_delete.dedup();
88
89 Ok(to_delete)
90 }
91
92 fn entity_expired_at(
93 &self,
94 collection: &str,
95 entity: &crate::storage::UnifiedEntity,
96 now_ms: u64,
97 ) -> bool {
98 let Some(metadata) = self.store.get_metadata(collection, entity.id) else {
99 return false;
100 };
101
102 if metadata
103 .get("_expires_at")
104 .and_then(Self::metadata_u64)
105 .is_some_and(|expires_at_ms| expires_at_ms <= now_ms)
106 {
107 return true;
108 }
109
110 let ttl_ms = metadata.get("_ttl_ms").and_then(Self::metadata_u64);
111 let ttl_secs = if ttl_ms.is_none() {
112 metadata.get("_ttl").and_then(|value| {
113 Self::metadata_u64(value).and_then(|value_secs| value_secs.checked_mul(1000))
114 })
115 } else {
116 None
117 };
118
119 let Some(ttl_ms) = ttl_ms.or(ttl_secs) else {
120 return false;
121 };
122 entity
123 .created_at
124 .saturating_mul(1000)
125 .saturating_add(ttl_ms)
126 <= now_ms
127 }
128
129 fn expired_entities_by_expires_at(
130 &self,
131 now_ms: u64,
132 ) -> Result<Vec<(String, EntityId)>, Box<dyn std::error::Error>> {
133 let mut ids = self.store.filter_metadata_all(&[(
134 "_expires_at".to_string(),
135 MetadataFilter::Le(MetadataValue::Timestamp(now_ms)),
136 )]);
137
138 if let Ok(now_ms_i64) = i64::try_from(now_ms) {
139 ids.extend(self.store.filter_metadata_all(&[(
140 "_expires_at".to_string(),
141 MetadataFilter::Le(MetadataValue::Int(now_ms_i64)),
142 )]));
143 }
144
145 let now_ms_f64 = now_ms as f64;
146 if now_ms_f64.is_finite() {
147 ids.extend(self.store.filter_metadata_all(&[(
148 "_expires_at".to_string(),
149 MetadataFilter::Le(MetadataValue::Float(now_ms_f64)),
150 )]));
151 }
152
153 Ok(ids)
154 }
155
156 fn expired_entities_by_ttl(
157 &self,
158 now_ms: u64,
159 ) -> Result<Vec<(String, EntityId)>, Box<dyn std::error::Error>> {
160 let mut candidates = Vec::<(String, EntityId)>::new();
161
162 let ttl_ms_candidates = self
163 .store
164 .filter_metadata_all(&[("_ttl_ms".to_string(), MetadataFilter::IsNotNull)]);
165 candidates.extend(ttl_ms_candidates);
166
167 let ttl_candidates = self
168 .store
169 .filter_metadata_all(&[("_ttl".to_string(), MetadataFilter::IsNotNull)]);
170 candidates.extend(ttl_candidates);
171
172 if candidates.is_empty() {
173 return Ok(Vec::new());
174 }
175
176 candidates.sort_unstable();
177 candidates.dedup();
178
179 let mut expired = Vec::<(String, EntityId)>::new();
180 for (collection, entity_id) in candidates {
181 let Some(entity) = self.store.get(&collection, entity_id) else {
182 continue;
183 };
184
185 let Some(metadata) = self.store.get_metadata(&collection, entity_id) else {
186 continue;
187 };
188
189 let ttl_ms = metadata.get("_ttl_ms").and_then(Self::metadata_u64);
190 let ttl_secs = if ttl_ms.is_none() {
191 metadata.get("_ttl").and_then(|value| {
192 Self::metadata_u64(value).and_then(|value_secs| value_secs.checked_mul(1000))
193 })
194 } else {
195 None
196 };
197
198 let Some(ttl_ms) = ttl_ms.or(ttl_secs) else {
199 continue;
200 };
201
202 let created_at_ms = entity.created_at.saturating_mul(1000);
203 let expiry_ms = created_at_ms.saturating_add(ttl_ms);
204 if expiry_ms <= now_ms {
205 expired.push((collection, entity_id));
206 }
207 }
208
209 Ok(expired)
210 }
211
212 fn metadata_u64(value: &MetadataValue) -> Option<u64> {
213 match value {
214 MetadataValue::Int(v) if *v >= 0 => Some(*v as u64),
215 MetadataValue::Timestamp(v) => Some(*v),
216 MetadataValue::Float(v) => {
217 if !v.is_finite() || !v.is_sign_positive() || v.fract().abs() >= f64::EPSILON {
218 return None;
219 }
220 if *v > u64::MAX as f64 {
221 return None;
222 }
223 Some(v.trunc() as u64)
224 }
225 MetadataValue::String(v) => v.parse::<u64>().ok(),
226 _ => None,
227 }
228 }
229
230 pub fn node(&self, collection: impl Into<String>, label: impl Into<String>) -> NodeBuilder {
243 NodeBuilder::new(
244 self.store.clone(),
245 self.preprocessors.clone(),
246 collection,
247 label,
248 )
249 }
250
251 pub fn edge(&self, collection: impl Into<String>, label: impl Into<String>) -> EdgeBuilder {
263 EdgeBuilder::new(
264 self.store.clone(),
265 self.preprocessors.clone(),
266 collection,
267 label,
268 )
269 }
270
271 pub fn vector(&self, collection: impl Into<String>) -> VectorBuilder {
282 VectorBuilder::new(self.store.clone(), self.preprocessors.clone(), collection)
283 }
284
285 pub fn row(&self, table: impl Into<String>, columns: Vec<(&str, Value)>) -> RowBuilder {
296 RowBuilder::new(
297 self.store.clone(),
298 self.preprocessors.clone(),
299 table,
300 columns,
301 )
302 }
303
304 pub fn doc(&self, collection: impl Into<String>) -> DocumentBuilder {
318 DocumentBuilder::new(self.store.clone(), self.preprocessors.clone(), collection)
319 }
320
321 pub fn kv(
332 &self,
333 collection: impl Into<String>,
334 key: impl Into<String>,
335 value: Value,
336 ) -> KvBuilder {
337 KvBuilder::new(
338 self.store.clone(),
339 self.preprocessors.clone(),
340 collection,
341 key,
342 value,
343 )
344 }
345
346 pub fn get_kv(&self, collection: &str, key: &str) -> Option<(Value, EntityId)> {
350 let manager = self.store.get_collection(collection)?;
351 let entities = manager.query_all(|_| true);
352 for entity in entities {
353 if let EntityData::Row(ref row) = entity.data {
354 if let Some(ref named) = row.named {
355 if let Some(Value::Text(ref k)) = named.get("key") {
356 if &**k == key {
357 let value = named.get("value").cloned().unwrap_or(Value::Null);
358 return Some((value, entity.id));
359 }
360 }
361 }
362 }
363 }
364 None
365 }
366
367 pub fn delete_kv(
369 &self,
370 collection: &str,
371 key: &str,
372 ) -> Result<bool, super::super::error::DevXError> {
373 let Some((_, id)) = self.get_kv(collection, key) else {
374 return Ok(false);
375 };
376 self.store
377 .delete(collection, id)
378 .map_err(|err| super::super::error::DevXError::Storage(format!("{err:?}")))?;
379 Ok(true)
380 }
381
382 pub(crate) fn with_initialized_metadata(self) -> Result<Self, Box<dyn std::error::Error>> {
383 if self.options.mode == StorageMode::Persistent {
384 if let Ok(metadata) = self.load_or_bootstrap_physical_metadata(false) {
386 crate::reserved_fields::validate_physical_metadata_contracts(&metadata)
387 .map_err(|err| err.to_string())?;
388 }
389 }
393 self.load_collection_ttl_defaults_from_metadata();
394 self.load_hypertables_from_metadata();
398 self.recover_queue_pending_state();
399 Ok(self)
400 }
401
402 pub(crate) fn persist_metadata(&self) -> Result<(), Box<dyn std::error::Error>> {
403 if self.options.mode != StorageMode::Persistent || self.options.read_only {
404 return Ok(());
405 }
406 let Some(path) = self.path() else {
407 return Ok(());
408 };
409
410 let previous = self.load_or_bootstrap_physical_metadata(false).ok();
411 let collection_roots = self.physical_collection_roots();
412 let indexes = self
413 .native_physical_state()
414 .map(|state| self.physical_index_state_from_native_state(&state, previous.as_ref()))
415 .unwrap_or_else(|| self.physical_index_state());
416 let mut metadata = PhysicalMetadataFile::from_state(
417 self.options.clone(),
418 self.catalog_snapshot(),
419 collection_roots,
420 indexes,
421 previous.as_ref(),
422 );
423 metadata.collection_ttl_defaults_ms = self.collection_ttl_defaults_snapshot();
424 metadata.hypertables = self.hypertable_registry_snapshot();
428 metadata.save_for_data_path(path)?;
429 self.persist_native_physical_header(&metadata)?;
430 Ok(())
431 }
432
433 fn bootstrap_metadata_from_native_state(&self) -> Result<bool, Box<dyn std::error::Error>> {
434 if self.options.mode != StorageMode::Persistent || self.options.read_only {
435 return Ok(false);
436 }
437 let Some(path) = self.path() else {
438 return Ok(false);
439 };
440 let Some(native_state) = self.native_physical_state() else {
441 return Ok(false);
442 };
443 if !Self::native_state_is_bootstrap_complete(&native_state) {
444 return Ok(false);
445 }
446
447 let previous = PhysicalMetadataFile::load_for_data_path(path).ok();
448 let metadata = self.metadata_from_native_state(&native_state, previous.as_ref());
449 metadata.save_for_data_path(path)?;
450 self.persist_native_physical_header(&metadata)?;
451 Ok(true)
452 }
453
454 pub fn rebuild_physical_metadata_from_native_state(
457 &self,
458 ) -> Result<bool, Box<dyn std::error::Error>> {
459 self.bootstrap_metadata_from_native_state()
460 }
461
462 pub(crate) fn native_state_is_bootstrap_complete(native_state: &NativePhysicalState) -> bool {
463 let registry_complete = native_state.registry.as_ref().map(|registry| {
464 registry.collections_complete
465 && registry.indexes_complete
466 && registry.graph_projections_complete
467 && registry.analytics_jobs_complete
468 && registry.vector_artifacts_complete
469 });
470 let recovery_complete = native_state
471 .recovery
472 .as_ref()
473 .map(|recovery| recovery.snapshots_complete && recovery.exports_complete);
474 let catalog_complete = native_state
475 .catalog
476 .as_ref()
477 .map(|catalog| catalog.collections_complete);
478
479 registry_complete == Some(true)
480 && recovery_complete == Some(true)
481 && catalog_complete == Some(true)
482 }
483
484 pub(crate) fn load_or_bootstrap_physical_metadata(
485 &self,
486 persist_bootstrapped: bool,
487 ) -> Result<PhysicalMetadataFile, Box<dyn std::error::Error>> {
488 if self.options.mode != StorageMode::Persistent {
489 return Err("physical metadata requires persistent mode".into());
490 }
491 let Some(path) = self.path() else {
492 return Err("database path is not available".into());
493 };
494 let native_state = self.native_physical_state();
495
496 match PhysicalMetadataFile::load_for_data_path(path) {
497 Ok(metadata) => {
498 if let Some(native_state) = native_state.as_ref() {
499 let inspection = Self::inspect_native_header_against_metadata(
500 native_state.header,
501 &metadata,
502 );
503 if Self::repair_policy_for_inspection(&inspection)
504 == NativeHeaderRepairPolicy::NativeAheadOfMetadata
505 {
506 let bootstrapped =
507 self.metadata_from_native_state(native_state, Some(&metadata));
508 if persist_bootstrapped && !self.options.read_only {
509 bootstrapped.save_for_data_path(path)?;
510 self.persist_native_physical_header(&bootstrapped)?;
511 }
512 return Ok(bootstrapped);
513 }
514 }
515 Ok(metadata)
516 }
517 Err(err) => {
518 let Some(native_state) = native_state else {
519 return Err(err.into());
520 };
521 let is_fresh_empty = native_state.header.sequence == 0
537 && native_state.registry.is_none()
538 && native_state.catalog.is_none()
539 && native_state.recovery.is_none();
540 if !is_fresh_empty && !Self::native_state_is_bootstrap_complete(&native_state) {
541 return Err(err.into());
542 }
543 let metadata = self.metadata_from_native_state(&native_state, None);
544 if persist_bootstrapped && !self.options.read_only {
545 metadata.save_for_data_path(path)?;
546 self.persist_native_physical_header(&metadata)?;
547 }
548 Ok(metadata)
549 }
550 }
551 }
552
553 pub(crate) fn physical_metadata_preference(&self) -> Option<&'static str> {
554 let path = self.path()?;
555 let native_state = self.native_physical_state();
556 let metadata = PhysicalMetadataFile::load_for_data_path(path).ok();
557
558 match (metadata, native_state) {
559 (Some(metadata), Some(native_state)) => {
560 let inspection =
561 Self::inspect_native_header_against_metadata(native_state.header, &metadata);
562 match Self::repair_policy_for_inspection(&inspection) {
563 NativeHeaderRepairPolicy::InSync => Some("sidecar_current"),
564 NativeHeaderRepairPolicy::RepairNativeFromMetadata => Some("sidecar_current"),
565 NativeHeaderRepairPolicy::NativeAheadOfMetadata => Some("native_ahead"),
566 }
567 }
568 (Some(_), None) => Some("sidecar_only"),
569 (None, Some(_)) => Some("sidecar_missing_native_available"),
570 (None, None) => Some("sidecar_missing_no_native"),
571 }
572 }
573
574 fn metadata_from_native_state(
575 &self,
576 native_state: &NativePhysicalState,
577 previous: Option<&PhysicalMetadataFile>,
578 ) -> PhysicalMetadataFile {
579 let now = SystemTime::now()
580 .duration_since(UNIX_EPOCH)
581 .unwrap_or_default()
582 .as_millis();
583 let catalog = self.catalog_snapshot();
584 let catalog_name = catalog.name.clone();
585 let catalog_total_entities = catalog.total_entities;
586 let catalog_total_collections = catalog.total_collections;
587 let indexes = self.physical_index_state();
588
589 let mut manifest =
590 crate::api::SchemaManifest::now(self.options.clone(), catalog.total_collections);
591 manifest.updated_at_unix_ms = now;
592
593 let manifest_events = native_state
594 .manifest
595 .as_ref()
596 .map(|summary| {
597 summary
598 .recent_events
599 .iter()
600 .map(|event| crate::physical::ManifestEvent {
601 collection: event.collection.clone(),
602 object_key: event.object_key.clone(),
603 kind: match event.kind.as_str() {
604 "insert" => crate::physical::ManifestEventKind::Insert,
605 "update" => crate::physical::ManifestEventKind::Update,
606 "remove" => crate::physical::ManifestEventKind::Remove,
607 _ => crate::physical::ManifestEventKind::Checkpoint,
608 },
609 block: crate::physical::BlockReference {
610 index: event.block_index,
611 checksum: event.block_checksum,
612 },
613 snapshot_min: event.snapshot_min,
614 snapshot_max: event.snapshot_max,
615 })
616 .collect()
617 })
618 .unwrap_or_default();
619
620 let graph_projections = native_state
621 .registry
622 .as_ref()
623 .and_then(|registry| {
624 registry.graph_projections_complete.then(|| {
625 registry
626 .graph_projections
627 .iter()
628 .map(|projection| crate::physical::PhysicalGraphProjection {
629 name: projection.name.clone(),
630 created_at_unix_ms: projection.created_at_unix_ms,
631 updated_at_unix_ms: projection.updated_at_unix_ms,
632 state: "materialized".to_string(),
633 source: projection.source.clone(),
634 node_labels: projection.node_labels.clone(),
635 node_types: projection.node_types.clone(),
636 edge_labels: projection.edge_labels.clone(),
637 last_materialized_sequence: projection.last_materialized_sequence,
638 })
639 .collect()
640 })
641 })
642 .or_else(|| previous.map(|metadata| metadata.graph_projections.clone()))
643 .unwrap_or_default();
644
645 let analytics_jobs = native_state
646 .registry
647 .as_ref()
648 .and_then(|registry| {
649 registry.analytics_jobs_complete.then(|| {
650 registry
651 .analytics_jobs
652 .iter()
653 .map(|job| crate::physical::PhysicalAnalyticsJob {
654 id: job.id.clone(),
655 kind: job.kind.clone(),
656 state: job.state.clone(),
657 projection: job.projection.clone(),
658 created_at_unix_ms: job.created_at_unix_ms,
659 updated_at_unix_ms: job.updated_at_unix_ms,
660 last_run_sequence: job.last_run_sequence,
661 metadata: job.metadata.clone(),
662 })
663 .collect()
664 })
665 })
666 .or_else(|| previous.map(|metadata| metadata.analytics_jobs.clone()))
667 .unwrap_or_default();
668
669 let exports = native_state
670 .recovery
671 .as_ref()
672 .and_then(|recovery| {
673 recovery.exports_complete.then(|| {
674 recovery
675 .exports
676 .iter()
677 .map(|export| crate::physical::ExportDescriptor {
678 name: export.name.clone(),
679 created_at_unix_ms: export.created_at_unix_ms,
680 snapshot_id: export.snapshot_id,
681 superblock_sequence: export.superblock_sequence,
682 data_path: self
683 .path()
684 .map(|path| {
685 crate::physical::PhysicalMetadataFile::export_data_path_for(
686 path,
687 &export.name,
688 )
689 .display()
690 .to_string()
691 })
692 .unwrap_or_default(),
693 metadata_path: self
694 .path()
695 .map(|path| {
696 let export_data_path =
697 crate::physical::PhysicalMetadataFile::export_data_path_for(
698 path,
699 &export.name,
700 );
701 crate::physical::PhysicalMetadataFile::metadata_path_for(
702 &export_data_path,
703 )
704 .display()
705 .to_string()
706 })
707 .unwrap_or_default(),
708 collection_count: export.collection_count as usize,
709 total_entities: export.total_entities as usize,
710 })
711 .collect()
712 })
713 })
714 .or_else(|| previous.map(|metadata| metadata.exports.clone()))
715 .unwrap_or_default();
716
717 let snapshots = native_state
718 .recovery
719 .as_ref()
720 .and_then(|recovery| {
721 recovery.snapshots_complete.then(|| {
722 recovery
723 .snapshots
724 .iter()
725 .map(|snapshot| crate::physical::SnapshotDescriptor {
726 snapshot_id: snapshot.snapshot_id,
727 created_at_unix_ms: snapshot.created_at_unix_ms,
728 superblock_sequence: snapshot.superblock_sequence,
729 collection_count: snapshot.collection_count as usize,
730 total_entities: snapshot.total_entities as usize,
731 })
732 .collect()
733 })
734 })
735 .or_else(|| previous.map(|metadata| metadata.snapshots.clone()))
736 .unwrap_or_else(|| {
737 vec![crate::physical::SnapshotDescriptor {
738 snapshot_id: native_state.header.sequence,
739 created_at_unix_ms: now,
740 superblock_sequence: native_state.header.sequence,
741 collection_count: catalog_total_collections,
742 total_entities: catalog_total_entities,
743 }]
744 });
745
746 let catalog_stats = native_state
747 .catalog
748 .as_ref()
749 .and_then(|native_catalog| {
750 native_catalog.collections_complete.then(|| {
751 native_catalog
752 .collections
753 .iter()
754 .map(|collection| {
755 (
756 collection.name.clone(),
757 crate::api::CollectionStats {
758 entities: collection.entities as usize,
759 cross_refs: collection.cross_refs as usize,
760 segments: collection.segments as usize,
761 },
762 )
763 })
764 .collect::<BTreeMap<_, _>>()
765 })
766 })
767 .or_else(|| previous.map(|metadata| metadata.catalog.stats_by_collection.clone()))
768 .unwrap_or_else(|| catalog.stats_by_collection.clone());
769
770 PhysicalMetadataFile {
771 protocol_version: crate::physical::PHYSICAL_METADATA_PROTOCOL_VERSION.to_string(),
772 generated_at_unix_ms: now,
773 last_loaded_from: Some("native_bootstrap".to_string()),
774 last_healed_at_unix_ms: Some(now),
775 manifest,
776 catalog: crate::api::CatalogSnapshot {
777 name: catalog_name,
778 total_entities: native_state
779 .catalog
780 .as_ref()
781 .map(|summary| summary.total_entities as usize)
782 .unwrap_or(catalog_total_entities),
783 total_collections: native_state
784 .catalog
785 .as_ref()
786 .map(|summary| summary.collection_count as usize)
787 .unwrap_or(catalog_total_collections),
788 stats_by_collection: catalog_stats,
789 updated_at: SystemTime::now(),
790 },
791 manifest_events,
792 collection_ttl_defaults_ms: previous
793 .map(|metadata| metadata.collection_ttl_defaults_ms.clone())
794 .unwrap_or_default(),
795 collection_contracts: previous
796 .map(|metadata| metadata.collection_contracts.clone())
797 .unwrap_or_default(),
798 hypertables: previous
799 .map(|metadata| metadata.hypertables.clone())
800 .unwrap_or_default(),
801 tree_definitions: previous
802 .map(|metadata| metadata.tree_definitions.clone())
803 .unwrap_or_default(),
804 indexes,
805 graph_projections,
806 analytics_jobs,
807 exports,
808 superblock: crate::physical::SuperblockHeader {
809 format_version: native_state.header.format_version,
810 sequence: native_state.header.sequence,
811 copies: crate::physical::DEFAULT_SUPERBLOCK_COPIES,
812 manifest: crate::physical::ManifestPointers {
813 oldest: crate::physical::BlockReference {
814 index: native_state.header.manifest_oldest_root,
815 checksum: 0,
816 },
817 newest: crate::physical::BlockReference {
818 index: native_state.header.manifest_root,
819 checksum: 0,
820 },
821 },
822 free_set: crate::physical::BlockReference {
823 index: native_state.header.free_set_root,
824 checksum: 0,
825 },
826 collection_roots: native_state.collection_roots.clone(),
827 },
828 snapshots,
829 }
830 }
831
832 pub(crate) fn reconcile_index_states_with_native_artifacts(
833 &self,
834 mut indexes: Vec<PhysicalIndexState>,
835 ) -> Vec<PhysicalIndexState> {
836 let native_artifacts = self
837 .native_physical_state()
838 .and_then(|state| state.registry)
839 .map(|registry| registry.vector_artifacts)
840 .unwrap_or_default();
841 for index in &mut indexes {
842 let Some(collection) = index.collection.as_deref() else {
843 continue;
844 };
845 let Some(artifact_kind) = Self::native_artifact_kind_for_index(index.kind) else {
846 continue;
847 };
848 let Some(artifact) = native_artifacts.iter().find(|artifact| {
849 artifact.collection == collection && artifact.artifact_kind == artifact_kind
850 }) else {
851 index.build_state = "metadata-only".to_string();
852 continue;
853 };
854 index.entries = artifact.vector_count as usize;
855 index.estimated_memory_bytes = artifact.serialized_bytes;
856 index.backend = format!("{}+native-artifact", index_backend_name(index.kind));
857 index.artifact_kind = Some(artifact.artifact_kind.clone());
858 index.artifact_checksum = Some(artifact.checksum);
859 index.build_state = "artifact-published".to_string();
860 if let Some(pages) = self.native_vector_artifact_pages() {
861 index.artifact_root_page = pages
862 .into_iter()
863 .find(|page| {
864 page.collection == artifact.collection
865 && page.artifact_kind == artifact.artifact_kind
866 })
867 .map(|page| page.root_page);
868 }
869 }
870 indexes
871 }
872
873 pub(crate) fn warmup_native_vector_artifact_for_index(
874 &self,
875 index: &PhysicalIndexState,
876 ) -> Result<(), String> {
877 let Some(collection) = index.collection.as_deref() else {
878 return Ok(());
879 };
880 let Some(artifact_kind) = Self::native_artifact_kind_for_index(index.kind) else {
881 return Ok(());
882 };
883 self.warmup_native_vector_artifact(collection, Some(artifact_kind))?;
884 Ok(())
885 }
886
887 pub(crate) fn apply_runtime_native_artifact_to_index_state(
888 &self,
889 index: &mut PhysicalIndexState,
890 ) -> Result<(), String> {
891 let Some(collection) = index.collection.as_deref() else {
892 return Ok(());
893 };
894 let Some(artifact_kind) = Self::native_artifact_kind_for_index(index.kind) else {
895 return Ok(());
896 };
897 let artifact = self.inspect_native_vector_artifact(collection, Some(artifact_kind))?;
898 index.entries = artifact
899 .graph_edge_count
900 .or(artifact.text_posting_count)
901 .unwrap_or(artifact.node_count) as usize;
902 index.estimated_memory_bytes = artifact.byte_len;
903 index.backend = format!("{}+native-artifact", index_backend_name(index.kind));
904 index.artifact_kind = Some(artifact.artifact_kind.clone());
905 index.artifact_checksum = Some(artifact.checksum);
906 index.build_state = "ready".to_string();
907 index.artifact_root_page = self
908 .native_vector_artifact_pages()
909 .and_then(|pages| {
910 pages.into_iter().find(|page| {
911 page.collection == artifact.collection
912 && page.artifact_kind == artifact.artifact_kind
913 })
914 })
915 .map(|page| page.root_page);
916 Ok(())
917 }
918
919 pub(crate) fn physical_index_state_from_native_state(
920 &self,
921 native_state: &NativePhysicalState,
922 previous: Option<&PhysicalMetadataFile>,
923 ) -> Vec<PhysicalIndexState> {
924 let mut fresh = self.physical_index_state();
925 let Some(registry) = native_state.registry.as_ref() else {
926 if let Some(previous) = previous {
927 for index in &previous.indexes {
928 if !fresh.iter().any(|candidate| candidate.name == index.name) {
929 fresh.push(index.clone());
930 }
931 }
932 }
933 return fresh;
934 };
935
936 for index in &mut fresh {
937 if let Some(native) = registry
938 .indexes
939 .iter()
940 .find(|candidate| candidate.name == index.name)
941 {
942 index.enabled = native.enabled;
943 index.last_refresh_ms = native.last_refresh_ms;
944 index.backend = native.backend.clone();
945 index.entries = native.entries as usize;
946 index.estimated_memory_bytes = native.estimated_memory_bytes;
947 if index.artifact_kind.is_none() {
948 index.artifact_kind = Self::native_artifact_kind_for_index(index.kind)
949 .map(|value| value.to_string());
950 }
951 if index.build_state == "catalog-derived" {
952 index.build_state = "registry-loaded".to_string();
953 }
954 }
955 }
956
957 for native in ®istry.indexes {
958 if fresh.iter().any(|index| index.name == native.name) {
959 continue;
960 }
961 let Some(kind) = Self::index_kind_from_str(&native.kind) else {
962 continue;
963 };
964 fresh.push(PhysicalIndexState {
965 name: native.name.clone(),
966 kind,
967 collection: native.collection.clone(),
968 enabled: native.enabled,
969 entries: native.entries as usize,
970 estimated_memory_bytes: native.estimated_memory_bytes,
971 last_refresh_ms: native.last_refresh_ms,
972 backend: native.backend.clone(),
973 artifact_kind: Self::native_artifact_kind_for_index(kind)
974 .map(|value| value.to_string()),
975 artifact_root_page: None,
976 artifact_checksum: None,
977 build_state: "registry-loaded".to_string(),
978 });
979 }
980
981 if !registry.indexes_complete {
982 if let Some(previous) = previous {
983 for index in &previous.indexes {
984 if !fresh.iter().any(|candidate| candidate.name == index.name) {
985 fresh.push(index.clone());
986 }
987 }
988 }
989 }
990
991 fresh
992 }
993
994 pub(crate) fn graph_projections_from_native_state(
995 &self,
996 native_state: &NativePhysicalState,
997 ) -> Vec<PhysicalGraphProjection> {
998 native_state
999 .registry
1000 .as_ref()
1001 .map(|registry| {
1002 registry
1003 .graph_projections
1004 .iter()
1005 .map(|projection| PhysicalGraphProjection {
1006 name: projection.name.clone(),
1007 created_at_unix_ms: projection.created_at_unix_ms,
1008 updated_at_unix_ms: projection.updated_at_unix_ms,
1009 state: "materialized".to_string(),
1010 source: projection.source.clone(),
1011 node_labels: projection.node_labels.clone(),
1012 node_types: projection.node_types.clone(),
1013 edge_labels: projection.edge_labels.clone(),
1014 last_materialized_sequence: projection.last_materialized_sequence,
1015 })
1016 .collect()
1017 })
1018 .unwrap_or_default()
1019 }
1020
1021 pub(crate) fn analytics_jobs_from_native_state(
1022 &self,
1023 native_state: &NativePhysicalState,
1024 ) -> Vec<PhysicalAnalyticsJob> {
1025 native_state
1026 .registry
1027 .as_ref()
1028 .map(|registry| {
1029 registry
1030 .analytics_jobs
1031 .iter()
1032 .map(|job| PhysicalAnalyticsJob {
1033 id: job.id.clone(),
1034 kind: job.kind.clone(),
1035 state: job.state.clone(),
1036 projection: job.projection.clone(),
1037 created_at_unix_ms: job.created_at_unix_ms,
1038 updated_at_unix_ms: job.updated_at_unix_ms,
1039 last_run_sequence: job.last_run_sequence,
1040 metadata: job.metadata.clone(),
1041 })
1042 .collect()
1043 })
1044 .unwrap_or_default()
1045 }
1046
1047 pub(crate) fn exports_from_native_state(
1048 &self,
1049 native_state: &NativePhysicalState,
1050 ) -> Vec<ExportDescriptor> {
1051 native_state
1052 .recovery
1053 .as_ref()
1054 .map(|recovery| {
1055 recovery
1056 .exports
1057 .iter()
1058 .map(|export| ExportDescriptor {
1059 name: export.name.clone(),
1060 created_at_unix_ms: export.created_at_unix_ms,
1061 snapshot_id: export.snapshot_id,
1062 superblock_sequence: export.superblock_sequence,
1063 data_path: self
1064 .path()
1065 .map(|path| {
1066 crate::physical::PhysicalMetadataFile::export_data_path_for(
1067 path,
1068 &export.name,
1069 )
1070 .display()
1071 .to_string()
1072 })
1073 .unwrap_or_default(),
1074 metadata_path: self
1075 .path()
1076 .map(|path| {
1077 let export_data_path =
1078 crate::physical::PhysicalMetadataFile::export_data_path_for(
1079 path,
1080 &export.name,
1081 );
1082 crate::physical::PhysicalMetadataFile::metadata_path_for(
1083 &export_data_path,
1084 )
1085 .display()
1086 .to_string()
1087 })
1088 .unwrap_or_default(),
1089 collection_count: export.collection_count as usize,
1090 total_entities: export.total_entities as usize,
1091 })
1092 .collect()
1093 })
1094 .unwrap_or_default()
1095 }
1096
1097 pub(crate) fn snapshots_from_native_state(
1098 &self,
1099 native_state: &NativePhysicalState,
1100 ) -> Vec<crate::physical::SnapshotDescriptor> {
1101 native_state
1102 .recovery
1103 .as_ref()
1104 .map(|recovery| {
1105 recovery
1106 .snapshots
1107 .iter()
1108 .map(|snapshot| crate::physical::SnapshotDescriptor {
1109 snapshot_id: snapshot.snapshot_id,
1110 created_at_unix_ms: snapshot.created_at_unix_ms,
1111 superblock_sequence: snapshot.superblock_sequence,
1112 collection_count: snapshot.collection_count as usize,
1113 total_entities: snapshot.total_entities as usize,
1114 })
1115 .collect()
1116 })
1117 .unwrap_or_default()
1118 }
1119
1120 fn index_kind_from_str(value: &str) -> Option<crate::index::IndexKind> {
1121 match value {
1122 "btree" => Some(crate::index::IndexKind::BTree),
1123 "vector.hnsw" => Some(crate::index::IndexKind::VectorHnsw),
1124 "vector.inverted" => Some(crate::index::IndexKind::VectorInverted),
1125 "vector.turbo" => Some(crate::index::IndexKind::VectorTurbo),
1126 "graph.adjacency" => Some(crate::index::IndexKind::GraphAdjacency),
1127 "text.fulltext" => Some(crate::index::IndexKind::FullText),
1128 "document.pathvalue" => Some(crate::index::IndexKind::DocumentPathValue),
1129 "search.hybrid" => Some(crate::index::IndexKind::HybridSearch),
1130 _ => None,
1131 }
1132 }
1133
1134 pub(crate) fn native_artifact_kind_for_index(kind: IndexKind) -> Option<&'static str> {
1135 match kind {
1136 IndexKind::VectorHnsw => Some("hnsw"),
1137 IndexKind::VectorInverted => Some("ivf"),
1138 IndexKind::VectorTurbo => Some("turboquant"),
1139 IndexKind::GraphAdjacency => Some("graph.adjacency"),
1140 IndexKind::FullText => Some("text.fulltext"),
1141 IndexKind::DocumentPathValue => Some("document.pathvalue"),
1142 _ => None,
1143 }
1144 }
1145
1146 fn index_is_declared(&self, name: &str) -> bool {
1147 self.physical_metadata()
1148 .map(|metadata| metadata.indexes.iter().any(|index| index.name == name))
1149 .unwrap_or(false)
1150 }
1151
1152 pub(crate) fn graph_projection_is_declared(&self, name: &str) -> bool {
1153 self.physical_metadata()
1154 .map(|metadata| {
1155 metadata
1156 .graph_projections
1157 .iter()
1158 .any(|projection| projection.name == name)
1159 })
1160 .unwrap_or(false)
1161 }
1162
1163 pub(crate) fn graph_projection_is_operational(&self, name: &str) -> bool {
1164 self.operational_graph_projections()
1165 .into_iter()
1166 .any(|projection| projection.name == name && projection.state == "materialized")
1167 }
1168
1169 pub(crate) fn analytics_job_id(kind: &str, projection: Option<&str>) -> String {
1170 match projection {
1171 Some(projection) => format!("{kind}::{projection}"),
1172 None => format!("{kind}::global"),
1173 }
1174 }
1175
1176 pub(crate) fn update_physical_metadata<T, F>(
1177 &self,
1178 mutator: F,
1179 ) -> Result<T, Box<dyn std::error::Error>>
1180 where
1181 F: FnOnce(&mut PhysicalMetadataFile) -> T,
1182 {
1183 if self.options.mode != StorageMode::Persistent {
1184 return Err("physical metadata operations require persistent mode".into());
1185 }
1186 if self.options.read_only {
1187 return Err("physical metadata operations are not allowed in read-only mode".into());
1188 }
1189 let Some(path) = self.path() else {
1190 return Err("database path is not available".into());
1191 };
1192
1193 let mut metadata = self.load_or_bootstrap_physical_metadata(true)?;
1194
1195 if metadata.indexes.is_empty() {
1196 metadata.indexes = self.physical_index_state();
1197 }
1198 metadata.superblock.collection_roots = self.physical_collection_roots();
1199
1200 let result = mutator(&mut metadata);
1201 metadata.save_for_data_path(path)?;
1202 self.persist_native_physical_header(&metadata)?;
1203 Ok(result)
1204 }
1205
1206 pub(crate) fn persist_native_physical_header(
1207 &self,
1208 metadata: &PhysicalMetadataFile,
1209 ) -> Result<(), Box<dyn std::error::Error>> {
1210 if !self.paged_mode {
1211 return Ok(());
1212 }
1213
1214 let existing_page = self
1215 .store
1216 .physical_file_header()
1217 .map(|header| header.collection_roots_page)
1218 .filter(|page| *page != 0);
1219 let existing_registry_page = self
1220 .store
1221 .physical_file_header()
1222 .map(|header| header.registry_page)
1223 .filter(|page| *page != 0);
1224 let existing_recovery_page = self
1225 .store
1226 .physical_file_header()
1227 .map(|header| header.recovery_page)
1228 .filter(|page| *page != 0);
1229 let existing_catalog_page = self
1230 .store
1231 .physical_file_header()
1232 .map(|header| header.catalog_page)
1233 .filter(|page| *page != 0);
1234 let existing_metadata_state_page = self
1235 .store
1236 .physical_file_header()
1237 .map(|header| header.metadata_state_page)
1238 .filter(|page| *page != 0);
1239 let existing_vector_artifact_page = self
1240 .store
1241 .physical_file_header()
1242 .map(|header| header.vector_artifact_page)
1243 .filter(|page| *page != 0);
1244 let existing_manifest_page = self
1245 .store
1246 .physical_file_header()
1247 .map(|header| header.manifest_page)
1248 .filter(|page| *page != 0);
1249 let (manifest_page, manifest_checksum) = self.store.write_native_manifest_summary(
1250 metadata.superblock.sequence,
1251 &metadata.manifest_events,
1252 existing_manifest_page,
1253 )?;
1254 let (collection_roots_page, collection_roots_checksum) = self
1255 .store
1256 .write_native_collection_roots(&metadata.superblock.collection_roots, existing_page)?;
1257 let registry_summary = self.native_registry_summary_from_metadata(metadata);
1258 let (registry_page, registry_checksum) = self
1259 .store
1260 .write_native_registry_summary(®istry_summary, existing_registry_page)?;
1261 let recovery_summary = Self::native_recovery_summary_from_metadata(metadata);
1262 let (recovery_page, recovery_checksum) = self
1263 .store
1264 .write_native_recovery_summary(&recovery_summary, existing_recovery_page)?;
1265 let catalog_summary = Self::native_catalog_summary_from_metadata(metadata);
1266 let (catalog_page, catalog_checksum) = self
1267 .store
1268 .write_native_catalog_summary(&catalog_summary, existing_catalog_page)?;
1269 let metadata_state_summary = Self::native_metadata_state_summary_from_metadata(metadata);
1270 let (metadata_state_page, metadata_state_checksum) =
1271 self.store.write_native_metadata_state_summary(
1272 &metadata_state_summary,
1273 existing_metadata_state_page,
1274 )?;
1275 let vector_artifact_records = self.native_vector_artifact_records();
1276 let vector_artifact_payloads = vector_artifact_records
1277 .iter()
1278 .map(|(summary, bytes)| {
1279 (
1280 summary.collection.clone(),
1281 summary.artifact_kind.clone(),
1282 bytes.clone(),
1283 )
1284 })
1285 .collect::<Vec<_>>();
1286 let (vector_artifact_page, vector_artifact_checksum, _vector_artifact_pages) =
1287 self.store.write_native_vector_artifact_store(
1288 &vector_artifact_payloads,
1289 existing_vector_artifact_page,
1290 )?;
1291 let mut header = Self::native_header_from_metadata(metadata);
1292 header.manifest_page = manifest_page;
1293 header.manifest_checksum = manifest_checksum;
1294 header.collection_roots_page = collection_roots_page;
1295 header.collection_roots_checksum = collection_roots_checksum;
1296 header.registry_page = registry_page;
1297 header.registry_checksum = registry_checksum;
1298 header.recovery_page = recovery_page;
1299 header.recovery_checksum = recovery_checksum;
1300 header.catalog_page = catalog_page;
1301 header.catalog_checksum = catalog_checksum;
1302 header.metadata_state_page = metadata_state_page;
1303 header.metadata_state_checksum = metadata_state_checksum;
1304 header.vector_artifact_page = vector_artifact_page;
1305 header.vector_artifact_checksum = vector_artifact_checksum;
1306 self.store.update_physical_file_header(header)?;
1307 self.store.persist()?;
1308 Ok(())
1309 }
1310
1311 pub(crate) fn native_header_from_metadata(
1312 metadata: &PhysicalMetadataFile,
1313 ) -> PhysicalFileHeader {
1314 PhysicalFileHeader {
1315 format_version: metadata.superblock.format_version,
1316 sequence: metadata.superblock.sequence,
1317 manifest_oldest_root: metadata.superblock.manifest.oldest.index,
1318 manifest_root: metadata.superblock.manifest.newest.index,
1319 free_set_root: metadata.superblock.free_set.index,
1320 manifest_page: 0,
1321 manifest_checksum: 0,
1322 collection_roots_page: 0,
1323 collection_roots_checksum: 0,
1324 collection_root_count: metadata.superblock.collection_roots.len() as u32,
1325 snapshot_count: metadata.snapshots.len() as u32,
1326 index_count: metadata.indexes.len() as u32,
1327 catalog_collection_count: metadata.catalog.total_collections as u32,
1328 catalog_total_entities: metadata.catalog.total_entities as u64,
1329 export_count: metadata.exports.len() as u32,
1330 graph_projection_count: metadata.graph_projections.len() as u32,
1331 analytics_job_count: metadata.analytics_jobs.len() as u32,
1332 manifest_event_count: metadata.manifest_events.len() as u32,
1333 registry_page: 0,
1334 registry_checksum: 0,
1335 recovery_page: 0,
1336 recovery_checksum: 0,
1337 catalog_page: 0,
1338 catalog_checksum: 0,
1339 metadata_state_page: 0,
1340 metadata_state_checksum: 0,
1341 vector_artifact_page: 0,
1342 vector_artifact_checksum: 0,
1343 }
1344 }
1345
1346 fn recover_queue_pending_state(&self) {
1347 const QUEUE_META_COLLECTION: &str = "red_queue_meta";
1348
1349 let Some(manager) = self.store.get_collection(QUEUE_META_COLLECTION) else {
1350 return;
1351 };
1352
1353 let pending_rows = manager.query_all(|entity| {
1354 entity.data.as_row().is_some_and(|row| {
1355 matches!(
1356 row.get_field("kind"),
1357 Some(crate::storage::schema::Value::Text(kind)) if &**kind == "queue_pending"
1358 )
1359 })
1360 });
1361
1362 for row in pending_rows {
1363 let _ = self.store.delete(QUEUE_META_COLLECTION, row.id);
1364 }
1365 }
1366}
1367
1368fn current_unix_ms() -> u64 {
1369 SystemTime::now()
1370 .duration_since(UNIX_EPOCH)
1371 .unwrap_or_default()
1372 .as_millis()
1373 .min(u128::from(u64::MAX)) as u64
1374}