datum-persistence 0.11.3

Event sourcing, durable state, and CQRS persistence for Datum
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
//! Public conformance suite for persistence storage backends.
//!
//! Out-of-tree backends should run every suite relevant to the capabilities
//! they implement against isolated backend instances. Backend errors are
//! returned; invariant violations panic with an assertion message so ordinary
//! Rust test runners report the failed obligation directly. The query suites
//! are the normative executable contract for invariant 12: current/live
//! boundaries, inclusive ordering, gapless resume, and durable coalescing.

use std::{
    collections::HashSet,
    sync::atomic::{AtomicU8, AtomicU64, Ordering},
    time::Duration,
};

use futures_util::StreamExt;
use tokio::time::timeout;

use crate::{
    AppendOutcome, AtomicWrite, DurableStateStore, EventStream, Journal, PersistenceError,
    PersistenceId, PersistenceResult, ReplayRange, Revision, SelectedSnapshot, SequenceNr,
    SerializedPayload, SerializerId, SnapshotCriteria, SnapshotMetadata, SnapshotStore,
    StoredEvent, Tag,
    query::{
        DurableStateChange, DurableStateChangesQuery, EventEnvelope, EventsByPersistenceIdQuery,
        EventsByTagQuery, Offset, PersistenceIdsQuery, QueryMode, QueryStream,
    },
};

static TEST_NAMESPACE: AtomicU64 = AtomicU64::new(1);
const QUERY_WAIT: Duration = Duration::from_secs(2);
const FAILING_JOURNAL_PASS: u8 = 0;
const FAILING_JOURNAL_FAIL_NEXT: u8 = 1;
const FAILING_JOURNAL_REJECT_NEXT: u8 = 2;

/// In-memory journal test double with one-shot append failure controls.
#[derive(Debug, Default)]
pub struct FailingJournal {
    inner: crate::MemoryJournal,
    next_append: AtomicU8,
}

impl FailingJournal {
    /// Creates an empty pass-through journal.
    #[must_use]
    pub fn new() -> Self {
        Self::default()
    }

    /// Makes the next append return a storage failure, then resumes pass-through behavior.
    pub fn fail_next_append(&self) {
        self.next_append
            .store(FAILING_JOURNAL_FAIL_NEXT, Ordering::Release);
    }

    /// Makes the next append deterministically reject every atomic write.
    pub fn reject_next_append(&self) {
        self.next_append
            .store(FAILING_JOURNAL_REJECT_NEXT, Ordering::Release);
    }
}

#[async_trait::async_trait]
impl Journal for FailingJournal {
    async fn append(&self, writes: Vec<AtomicWrite>) -> PersistenceResult<Vec<AppendOutcome>> {
        match self
            .next_append
            .swap(FAILING_JOURNAL_PASS, Ordering::AcqRel)
        {
            FAILING_JOURNAL_FAIL_NEXT => Err(PersistenceError::Failure(
                "injected journal append failure".into(),
            )),
            FAILING_JOURNAL_REJECT_NEXT => Ok(writes
                .into_iter()
                .map(|_| AppendOutcome::Rejected {
                    reason: "injected journal append rejection".into(),
                })
                .collect()),
            _ => self.inner.append(writes).await,
        }
    }

    async fn highest_sequence_nr(
        &self,
        persistence_id: &PersistenceId,
        from_hint: SequenceNr,
    ) -> PersistenceResult<SequenceNr> {
        self.inner
            .highest_sequence_nr(persistence_id, from_hint)
            .await
    }

    async fn replay(&self, range: ReplayRange) -> PersistenceResult<EventStream> {
        self.inner.replay(range).await
    }

    async fn delete_to(
        &self,
        persistence_id: &PersistenceId,
        inclusive: SequenceNr,
    ) -> PersistenceResult<()> {
        self.inner.delete_to(persistence_id, inclusive).await
    }
}

/// Asserts the complete WP-PS2 journal storage contract.
pub async fn assert_journal_conformance(journal: &dyn Journal) -> PersistenceResult<()> {
    let namespace = TEST_NAMESPACE.fetch_add(1, Ordering::Relaxed);
    let unknown_id = test_id("journal-unknown", namespace);
    assert_eq!(
        journal
            .highest_sequence_nr(&unknown_id, SequenceNr::new(99))
            .await?,
        SequenceNr::new(0),
        "an unknown persistence id must have watermark zero regardless of the hint"
    );

    assert!(matches!(
        AtomicWrite::new(Vec::new()),
        Err(PersistenceError::InvalidAtomicWrite(_))
    ));
    let mismatched_id = test_id("journal-mismatched", namespace);
    assert!(matches!(
        AtomicWrite::new(vec![event(&unknown_id, 1, 1), event(&mismatched_id, 2, 2),]),
        Err(PersistenceError::InvalidAtomicWrite(_))
    ));
    assert!(matches!(
        AtomicWrite::new(vec![event(&unknown_id, 1, 1), event(&unknown_id, 3, 3),]),
        Err(PersistenceError::InvalidAtomicWrite(_))
    ));

    let ordered_id = test_id("journal-ordered", namespace);
    let outcomes = journal
        .append(vec![
            atomic(vec![
                event(&ordered_id, 1, 1),
                event(&ordered_id, 2, 2),
                event(&ordered_id, 3, 3),
            ]),
            atomic(vec![event(&ordered_id, 4, 4)]),
            atomic(vec![event(&ordered_id, 5, 5)]),
        ])
        .await?;
    assert_eq!(
        outcomes.len(),
        3,
        "append must return one outcome per input"
    );
    assert!(
        outcomes
            .iter()
            .all(|outcome| *outcome == AppendOutcome::Accepted),
        "valid sequential writes must be accepted"
    );

    let all = replay(journal, replay_range(&ordered_id, 1, 5, u64::MAX)).await?;
    assert_sequence_nrs(&all, &[1, 2, 3, 4, 5]);
    assert_eq!(
        all[1].metadata,
        Some(payload(102)),
        "journal replay must preserve optional metadata frames"
    );
    assert_eq!(all[1].tags, vec![Tag::new("even")]);

    let bounded = replay(journal, replay_range(&ordered_id, 2, 4, u64::MAX)).await?;
    assert_sequence_nrs(&bounded, &[2, 3, 4]);
    let limited = replay(journal, replay_range(&ordered_id, 2, 5, 2)).await?;
    assert_sequence_nrs(&limited, &[2, 3]);
    assert!(
        replay(journal, replay_range(&ordered_id, 5, 4, u64::MAX))
            .await?
            .is_empty(),
        "reversed replay bounds must be empty"
    );
    assert!(
        replay(journal, replay_range(&ordered_id, 1, 5, 0))
            .await?
            .is_empty(),
        "replay max zero must be empty"
    );

    journal.delete_to(&ordered_id, SequenceNr::new(3)).await?;
    assert_sequence_nrs(
        &replay(journal, replay_range(&ordered_id, 1, 5, u64::MAX)).await?,
        &[4, 5],
    );
    assert_eq!(
        journal
            .highest_sequence_nr(&ordered_id, SequenceNr::new(0))
            .await?,
        SequenceNr::new(5),
        "the high watermark must survive a prefix delete"
    );
    journal
        .delete_to(&ordered_id, SequenceNr::new(u64::MAX))
        .await?;
    assert!(
        replay(journal, replay_range(&ordered_id, 1, u64::MAX, u64::MAX))
            .await?
            .is_empty(),
        "delete-to-MAX must hide the entire stored range"
    );
    assert_eq!(
        journal
            .highest_sequence_nr(&ordered_id, SequenceNr::new(u64::MAX))
            .await?,
        SequenceNr::new(5),
        "the high watermark must survive delete-to-MAX"
    );
    assert_eq!(
        journal
            .append(vec![atomic(vec![event(&ordered_id, 6, 6)])])
            .await?,
        vec![AppendOutcome::Accepted],
        "a delete-to-MAX must not hide later events above the preserved watermark"
    );
    assert_sequence_nrs(
        &replay(journal, replay_range(&ordered_id, 1, u64::MAX, u64::MAX)).await?,
        &[6],
    );

    let rejected_id = test_id("journal-rejected", namespace);
    assert_eq!(
        journal
            .append(vec![atomic(vec![event(&rejected_id, 1, 10)])])
            .await?,
        vec![AppendOutcome::Accepted]
    );
    let sibling_id = test_id("journal-sibling", namespace);
    let outcomes = journal
        .append(vec![
            atomic(vec![event(&rejected_id, 3, 30), event(&rejected_id, 4, 40)]),
            atomic(vec![event(&sibling_id, 1, 50), event(&sibling_id, 2, 60)]),
        ])
        .await?;
    assert_eq!(
        outcomes.len(),
        2,
        "append must return one outcome per input"
    );
    assert!(matches!(
        &outcomes[0],
        AppendOutcome::Rejected { reason } if !reason.is_empty()
    ));
    assert_eq!(outcomes[1], AppendOutcome::Accepted);
    assert_sequence_nrs(
        &replay(journal, replay_range(&rejected_id, 1, u64::MAX, u64::MAX)).await?,
        &[1],
    );
    assert_sequence_nrs(
        &replay(journal, replay_range(&sibling_id, 1, u64::MAX, u64::MAX)).await?,
        &[1, 2],
    );

    Ok(())
}

/// Asserts the complete WP-PS2 snapshot-store contract.
pub async fn assert_snapshot_store_conformance(
    snapshots: &dyn SnapshotStore,
) -> PersistenceResult<()> {
    let namespace = TEST_NAMESPACE.fetch_add(1, Ordering::Relaxed);
    let persistence_id = test_id("snapshot", namespace);
    assert!(
        snapshots
            .load(&persistence_id, &SnapshotCriteria::latest())
            .await?
            .is_none(),
        "snapshot absence must be returned as None, not a load failure"
    );
    assert!(
        snapshots
            .load(&persistence_id, &SnapshotCriteria::none())
            .await?
            .is_none(),
        "SnapshotCriteria::none must never select a snapshot"
    );

    for (sequence_nr, timestamp_ms, marker) in [(1, 100, 1), (3, 300, 3), (2, 200, 2)] {
        snapshots
            .save(
                snapshot_metadata(&persistence_id, sequence_nr, timestamp_ms),
                format!("datum-tck-state-adapter-v{marker}"),
                payload(marker),
            )
            .await?;
    }
    let latest = require_snapshot(
        snapshots
            .load(&persistence_id, &SnapshotCriteria::latest())
            .await?,
    );
    assert_eq!(latest.metadata.sequence_nr, SequenceNr::new(3));
    assert_eq!(latest.adapter_manifest, "datum-tck-state-adapter-v3");
    assert_eq!(latest.state, payload(3));

    let exact_two = SnapshotCriteria {
        min_sequence_nr: SequenceNr::new(2),
        max_sequence_nr: SequenceNr::new(2),
        min_timestamp_ms: 200,
        max_timestamp_ms: 200,
    };
    assert_eq!(
        require_snapshot(snapshots.load(&persistence_id, &exact_two).await?).state,
        payload(2),
        "all snapshot criteria bounds must be inclusive"
    );

    let replacement = snapshot_metadata(&persistence_id, 3, 350);
    snapshots
        .save(
            replacement.clone(),
            "datum-tck-state-adapter-v33".into(),
            payload(33),
        )
        .await?;
    let latest = require_snapshot(
        snapshots
            .load(&persistence_id, &SnapshotCriteria::latest())
            .await?,
    );
    assert_eq!(latest.metadata, replacement);
    assert_eq!(latest.adapter_manifest, "datum-tck-state-adapter-v33");
    assert_eq!(latest.state, payload(33));
    snapshots
        .delete(snapshot_metadata(&persistence_id, 3, 300))
        .await?;
    assert_eq!(
        require_snapshot(
            snapshots
                .load(&persistence_id, &SnapshotCriteria::latest())
                .await?,
        )
        .state,
        payload(33),
        "deleting replaced metadata must not remove the current same-sequence snapshot"
    );

    snapshots
        .delete(snapshot_metadata(&persistence_id, 2, 200))
        .await?;
    assert!(snapshots.load(&persistence_id, &exact_two).await?.is_none());

    for (sequence_nr, timestamp_ms) in [(4, 400), (5, 500), (6, 600)] {
        snapshots
            .save(
                snapshot_metadata(&persistence_id, sequence_nr, timestamp_ms),
                format!("datum-tck-state-adapter-v{sequence_nr}"),
                payload(sequence_nr),
            )
            .await?;
    }
    let delete_range = SnapshotCriteria {
        min_sequence_nr: SequenceNr::new(4),
        max_sequence_nr: SequenceNr::new(5),
        min_timestamp_ms: 400,
        max_timestamp_ms: 500,
    };
    snapshots
        .delete_matching(&persistence_id, &delete_range)
        .await?;
    assert!(
        snapshots
            .load(&persistence_id, &delete_range)
            .await?
            .is_none(),
        "delete_matching must remove every snapshot inside the inclusive criteria"
    );
    assert_eq!(
        require_snapshot(
            snapshots
                .load(&persistence_id, &SnapshotCriteria::latest())
                .await?,
        )
        .metadata
        .sequence_nr,
        SequenceNr::new(6),
        "delete_matching must preserve snapshots outside its criteria"
    );

    Ok(())
}

/// Asserts the complete WP-PS2 durable-state storage contract.
pub async fn assert_durable_state_store_conformance(
    durable_state: &dyn DurableStateStore,
) -> PersistenceResult<()> {
    let namespace = TEST_NAMESPACE.fetch_add(1, Ordering::Relaxed);
    let persistence_id = test_id("durable", namespace);
    let absent = durable_state.get(&persistence_id).await?;
    assert_eq!(absent.revision, Revision::new(0));
    assert!(
        absent.state.is_none(),
        "unknown durable state must be absent"
    );

    let invalid_first_revision = durable_state
        .upsert(&persistence_id, Revision::new(2), payload(2), None)
        .await
        .expect_err("the first durable-state write must use revision one");
    assert!(matches!(
        invalid_first_revision,
        PersistenceError::RevisionConflict {
            expected,
            actual,
            ..
        } if expected == Revision::new(1) && actual == Revision::new(2)
    ));
    assert_eq!(durable_state.get(&persistence_id).await?, absent);

    durable_state
        .upsert(
            &persistence_id,
            Revision::new(1),
            payload(1),
            Some(Tag::new("blue")),
        )
        .await?;
    let first = durable_state.get(&persistence_id).await?;
    assert_eq!(first.revision, Revision::new(1));
    assert_eq!(first.state, Some(payload(1)));
    assert_eq!(first.tag, Some(Tag::new("blue")));

    let conflict = durable_state
        .upsert(&persistence_id, Revision::new(3), payload(3), None)
        .await
        .expect_err("a skipped durable-state revision must conflict");
    assert!(matches!(
        conflict,
        PersistenceError::RevisionConflict {
            persistence_id: ref conflict_id,
            expected,
            actual,
        } if conflict_id == &persistence_id
            && expected == Revision::new(2)
            && actual == Revision::new(3)
    ));
    assert_eq!(
        durable_state.get(&persistence_id).await?,
        first,
        "a CAS conflict must not change stored durable state"
    );

    durable_state
        .upsert(&persistence_id, Revision::new(2), payload(2), None)
        .await?;
    assert_eq!(durable_state.get(&persistence_id).await?.tag, None);
    durable_state
        .delete(&persistence_id, Revision::new(3))
        .await?;
    let tombstone = durable_state.get(&persistence_id).await?;
    assert_eq!(tombstone.revision, Revision::new(3));
    assert!(
        tombstone.state.is_none(),
        "delete must leave an absent-state tombstone at the advanced revision"
    );
    assert_eq!(
        tombstone.tag, None,
        "delete must clear the durable-state tag"
    );

    let stale_delete = durable_state
        .delete(&persistence_id, Revision::new(3))
        .await
        .expect_err("reusing the tombstone revision must conflict");
    assert!(matches!(
        stale_delete,
        PersistenceError::RevisionConflict {
            expected,
            actual,
            ..
        } if expected == Revision::new(4) && actual == Revision::new(3)
    ));
    assert_eq!(durable_state.get(&persistence_id).await?, tombstone);

    Ok(())
}

/// Normative conformance contract for `events_by_persistence_id` backends.
///
/// Out-of-tree and future first-party query implementations must run this
/// suite against an isolated journal/query pair. It verifies inclusive bounds,
/// sequence ordering, the `Current` snapshot boundary, and later delivery plus
/// finite-range completion for `Live`.
pub async fn assert_events_by_persistence_id_query_conformance(
    journal: &dyn Journal,
    query: &dyn EventsByPersistenceIdQuery,
) -> PersistenceResult<()> {
    let namespace = TEST_NAMESPACE.fetch_add(1, Ordering::Relaxed);
    let persistence_id = test_id("query-pid-bounds", namespace);
    append_accepted(
        journal,
        vec![
            event(&persistence_id, 1, 1),
            event(&persistence_id, 2, 2),
            event(&persistence_id, 3, 3),
            event(&persistence_id, 4, 4),
            event(&persistence_id, 5, 5),
        ],
    )
    .await?;

    let bounded = collect_query(
        query
            .events_by_persistence_id(
                &persistence_id,
                SequenceNr::new(2),
                SequenceNr::new(4),
                QueryMode::Current,
            )
            .await?,
        "current per-id inclusive range",
    )
    .await?;
    assert_envelope_sequence_nrs(&bounded, &[2, 3, 4]);

    let current_at_boundary = query
        .events_by_persistence_id(
            &persistence_id,
            SequenceNr::new(1),
            SequenceNr::new(6),
            QueryMode::Current,
        )
        .await?;
    append_accepted(journal, vec![event(&persistence_id, 6, 6)]).await?;
    let boundary_events = collect_query(current_at_boundary, "current per-id boundary").await?;
    assert_envelope_sequence_nrs(&boundary_events, &[1, 2, 3, 4, 5]);

    let live_id = test_id("query-pid-live", namespace);
    append_accepted(journal, vec![event(&live_id, 1, 11)]).await?;
    let mut live = query
        .events_by_persistence_id(
            &live_id,
            SequenceNr::new(1),
            SequenceNr::new(3),
            QueryMode::Live,
        )
        .await?;
    let first = require_query_item(&mut live, "initial live per-id event").await?;
    assert_eq!(first.sequence_nr, SequenceNr::new(1));

    append_accepted(
        journal,
        vec![event(&live_id, 2, 12), event(&live_id, 3, 13)],
    )
    .await?;
    let second = require_query_item(&mut live, "second live per-id event").await?;
    let third = require_query_item(&mut live, "third live per-id event").await?;
    assert_eq!(second.sequence_nr, SequenceNr::new(2));
    assert_eq!(third.sequence_nr, SequenceNr::new(3));
    assert!(
        next_query_item(&mut live, "finite live per-id completion")
            .await?
            .is_none(),
        "a live per-id query must complete after its finite inclusive upper bound"
    );

    Ok(())
}

/// Normative conformance contract for `events_by_tag` backends.
///
/// The suite verifies backend-global tag ordering and gapless offset resume
/// after a simulated disconnect. Duplicate re-emission at resume is accepted;
/// omission of any not-yet-observed event is not.
pub async fn assert_events_by_tag_query_conformance(
    journal: &dyn Journal,
    query: &dyn EventsByTagQuery,
) -> PersistenceResult<()> {
    let namespace = TEST_NAMESPACE.fetch_add(1, Ordering::Relaxed);
    let tag = Tag::new(format!("query-order-{namespace}"));
    let first_id = test_id("query-tag-first", namespace);
    let second_id = test_id("query-tag-second", namespace);
    append_accepted(journal, vec![tagged_event(&first_id, 1, 101, &tag)]).await?;
    append_accepted(journal, vec![tagged_event(&second_id, 1, 102, &tag)]).await?;
    append_accepted(journal, vec![tagged_event(&first_id, 2, 103, &tag)]).await?;

    let ordered = collect_query(
        query
            .events_by_tag(&tag, Offset::NoOffset, QueryMode::Current)
            .await?,
        "current tag ordering",
    )
    .await?;
    assert_eq!(
        ordered
            .iter()
            .map(|envelope| (&envelope.persistence_id, envelope.sequence_nr.as_u64()))
            .collect::<Vec<_>>(),
        vec![(&first_id, 1), (&second_id, 1), (&first_id, 2)],
        "tag queries must preserve the backend's global append order"
    );
    assert_strict_sequence_offsets(&ordered);

    let resume_tag = Tag::new(format!("query-resume-{namespace}"));
    let resume_id = test_id("query-tag-resume", namespace);
    append_accepted(
        journal,
        vec![
            tagged_event(&resume_id, 1, 201, &resume_tag),
            tagged_event(&resume_id, 2, 202, &resume_tag),
            tagged_event(&resume_id, 3, 203, &resume_tag),
        ],
    )
    .await?;

    let mut disconnected = query
        .events_by_tag(&resume_tag, Offset::NoOffset, QueryMode::Live)
        .await?;
    let first = require_query_item(&mut disconnected, "first pre-disconnect tag event").await?;
    let second = require_query_item(&mut disconnected, "second pre-disconnect tag event").await?;
    let resume_offset = second.offset;
    drop(disconnected);

    append_accepted(journal, vec![tagged_event(&resume_id, 4, 204, &resume_tag)]).await?;
    let resumed = collect_query(
        query
            .events_by_tag(&resume_tag, resume_offset, QueryMode::Current)
            .await?,
        "resumed current tag query",
    )
    .await?;

    let observed = std::iter::once(first)
        .chain(std::iter::once(second))
        .chain(resumed)
        .filter(|envelope| envelope.persistence_id == resume_id)
        .map(|envelope| envelope.sequence_nr.as_u64())
        .collect::<HashSet<_>>();
    assert_eq!(
        observed,
        HashSet::from([1, 2, 3, 4]),
        "resuming from the last observed offset must not lose unseen tag events"
    );

    Ok(())
}

/// Normative conformance contract for persistence-id discovery backends.
///
/// The suite verifies that `Current` completes at its captured boundary and
/// that `Live` discovers an id first written after the query begins.
pub async fn assert_persistence_ids_query_conformance(
    journal: &dyn Journal,
    query: &dyn PersistenceIdsQuery,
) -> PersistenceResult<()> {
    let namespace = TEST_NAMESPACE.fetch_add(1, Ordering::Relaxed);
    let first_id = test_id("query-ids-first", namespace);
    let second_id = test_id("query-ids-second", namespace);
    let third_id = test_id("query-ids-third", namespace);
    append_accepted(journal, vec![event(&first_id, 1, 301)]).await?;

    let current_at_boundary = query.persistence_ids(QueryMode::Current).await?;
    append_accepted(journal, vec![event(&second_id, 1, 302)]).await?;
    let current = collect_query(current_at_boundary, "current persistence ids boundary").await?;
    assert!(current.contains(&first_id));
    assert!(
        !current.contains(&second_id),
        "Current persistence_ids must exclude ids first written after its boundary"
    );

    let mut live = query.persistence_ids(QueryMode::Live).await?;
    let initial_count = 2;
    let mut initial = Vec::with_capacity(initial_count);
    for _ in 0..initial_count {
        initial.push(require_query_item(&mut live, "initial live persistence id").await?);
    }
    assert!(initial.contains(&first_id));
    assert!(initial.contains(&second_id));

    append_accepted(journal, vec![event(&third_id, 1, 303)]).await?;
    assert_eq!(
        require_query_item(&mut live, "later live persistence id").await?,
        third_id
    );

    Ok(())
}

/// Normative conformance contract for durable-state change query backends.
///
/// The suite verifies latest-per-id coalescing, eventual observation of the
/// newest rapid revision, deletion tombstones, and gapless resume. Intermediate
/// revisions and duplicate resume emissions are both permitted.
pub async fn assert_durable_state_changes_query_conformance(
    durable_state: &dyn DurableStateStore,
    query: &dyn DurableStateChangesQuery,
) -> PersistenceResult<()> {
    let namespace = TEST_NAMESPACE.fetch_add(1, Ordering::Relaxed);
    let tag = Tag::new(format!("query-durable-{namespace}"));
    let persistence_id = test_id("query-durable", namespace);
    for revision in 1..=3 {
        durable_state
            .upsert(
                &persistence_id,
                Revision::new(revision),
                payload(revision),
                Some(tag.clone()),
            )
            .await?;
    }

    let current = collect_query(
        query
            .changes(&tag, Offset::NoOffset, QueryMode::Current)
            .await?,
        "current durable-state changes",
    )
    .await?;
    let matching = current
        .iter()
        .filter(|change| change.persistence_id() == &persistence_id)
        .collect::<Vec<_>>();
    assert_eq!(
        matching.len(),
        1,
        "Current durable changes must coalesce to one latest value per id"
    );
    assert_eq!(matching[0].revision(), Revision::new(3));
    assert!(matches!(matching[0], DurableStateChange::Updated(_)));

    let mut live = query
        .changes(&tag, Offset::NoOffset, QueryMode::Live)
        .await?;
    let initial = require_query_item(&mut live, "initial live durable change").await?;
    assert_eq!(initial.revision(), Revision::new(3));

    for revision in 4..=6 {
        durable_state
            .upsert(
                &persistence_id,
                Revision::new(revision),
                payload(revision),
                Some(tag.clone()),
            )
            .await?;
    }
    let latest = timeout(QUERY_WAIT, async {
        loop {
            match live.next().await {
                Some(Ok(change)) if change.persistence_id() == &persistence_id => {
                    if change.revision() == Revision::new(6) {
                        break Ok(change);
                    }
                }
                Some(Ok(_)) => {}
                Some(Err(error)) => break Err(error),
                None => {
                    break Err(PersistenceError::Failure(
                        "live durable query completed before the latest revision".into(),
                    ));
                }
            }
        }
    })
    .await
    .map_err(|_| PersistenceError::Failure("latest durable revision timed out".into()))??;
    let resume_offset = latest.offset();
    drop(live);

    durable_state
        .delete(&persistence_id, Revision::new(7))
        .await?;
    let resumed = collect_query(
        query
            .changes(&tag, resume_offset, QueryMode::Current)
            .await?,
        "resumed durable-state changes",
    )
    .await?;
    assert!(
        resumed.iter().any(|change| {
            matches!(
                change,
                DurableStateChange::Deleted(deleted)
                    if deleted.persistence_id == persistence_id
                        && deleted.revision == Revision::new(7)
            )
        }),
        "resuming after the latest observed durable offset must expose the later tombstone"
    );

    Ok(())
}

fn test_id(kind: &str, namespace: u64) -> PersistenceId {
    PersistenceId::new("datum-tck", format!("{kind}-{namespace}"))
        .expect("conformance-suite ids are valid")
}

fn payload(marker: u64) -> SerializedPayload {
    SerializedPayload::new(
        SerializerId::new(7),
        "datum-tck-v1",
        marker.to_be_bytes().to_vec(),
    )
}

fn event(persistence_id: &PersistenceId, sequence_nr: u64, marker: u64) -> StoredEvent {
    let is_even = sequence_nr & 1 == 0;
    StoredEvent {
        persistence_id: persistence_id.clone(),
        sequence_nr: SequenceNr::new(sequence_nr),
        adapter_manifest: "datum-tck-adapter-v1".to_owned(),
        writer_uuid: "datum-tck-writer".to_owned(),
        timestamp_ms: i64::try_from(marker).expect("test marker fits i64"),
        metadata: is_even.then(|| payload(marker + 100)),
        payload: payload(marker),
        tags: is_even.then(|| Tag::new("even")).into_iter().collect(),
    }
}

fn tagged_event(
    persistence_id: &PersistenceId,
    sequence_nr: u64,
    marker: u64,
    tag: &Tag,
) -> StoredEvent {
    let mut event = event(persistence_id, sequence_nr, marker);
    event.tags = vec![tag.clone()];
    event
}

fn atomic(events: Vec<StoredEvent>) -> AtomicWrite {
    AtomicWrite::new(events).expect("conformance-suite writes are internally valid")
}

fn replay_range(persistence_id: &PersistenceId, from: u64, to: u64, max: u64) -> ReplayRange {
    ReplayRange {
        persistence_id: persistence_id.clone(),
        from: SequenceNr::new(from),
        to: SequenceNr::new(to),
        max,
    }
}

async fn replay(journal: &dyn Journal, range: ReplayRange) -> PersistenceResult<Vec<StoredEvent>> {
    let stream = journal.replay(range).await?;
    collect_events(stream).await
}

async fn collect_events(mut stream: EventStream) -> PersistenceResult<Vec<StoredEvent>> {
    let mut events = Vec::new();
    while let Some(event) = stream.next().await {
        events.push(event?);
    }
    Ok(events)
}

async fn append_accepted(journal: &dyn Journal, events: Vec<StoredEvent>) -> PersistenceResult<()> {
    let outcome = journal.append_one(atomic(events)).await?;
    assert_eq!(
        outcome,
        AppendOutcome::Accepted,
        "conformance-suite setup writes must be accepted"
    );
    Ok(())
}

async fn collect_query<T>(stream: QueryStream<T>, context: &str) -> PersistenceResult<Vec<T>> {
    timeout(QUERY_WAIT, async move {
        let mut stream = stream;
        let mut items = Vec::new();
        while let Some(item) = stream.next().await {
            items.push(item?);
        }
        Ok(items)
    })
    .await
    .map_err(|_| PersistenceError::Failure(format!("{context} timed out")))?
}

async fn next_query_item<T>(
    stream: &mut QueryStream<T>,
    context: &str,
) -> PersistenceResult<Option<T>> {
    timeout(QUERY_WAIT, stream.next())
        .await
        .map_err(|_| PersistenceError::Failure(format!("{context} timed out")))?
        .transpose()
}

async fn require_query_item<T>(stream: &mut QueryStream<T>, context: &str) -> PersistenceResult<T> {
    next_query_item(stream, context).await?.ok_or_else(|| {
        PersistenceError::Failure(format!("{context} completed without an expected item"))
    })
}

fn assert_sequence_nrs(events: &[StoredEvent], expected: &[u64]) {
    let actual = events
        .iter()
        .map(|event| event.sequence_nr.as_u64())
        .collect::<Vec<_>>();
    assert_eq!(actual, expected, "replay sequence numbers differ");
}

fn assert_envelope_sequence_nrs(events: &[EventEnvelope], expected: &[u64]) {
    let actual = events
        .iter()
        .map(|event| event.sequence_nr.as_u64())
        .collect::<Vec<_>>();
    assert_eq!(actual, expected, "query sequence numbers differ");
}

fn assert_strict_sequence_offsets(events: &[EventEnvelope]) {
    // Sequence offsets must strictly increase. Timestamp offsets order
    // lexicographically on (epoch_micros, seq_nr) and may repeat an equal pair
    // (duplicate re-emission is allowed; gaps are not). Mixed offset kinds or
    // NoOffset in emitted events are conformance failures.
    let mut sequences = Vec::new();
    let mut timestamps = Vec::new();
    for event in events {
        match event.offset {
            Offset::Sequence(offset) => sequences.push(offset),
            Offset::Timestamp {
                epoch_micros,
                seq_nr,
            } => timestamps.push((epoch_micros, seq_nr)),
            other => panic!("an emitted event must carry a concrete offset, got {other:?}"),
        }
    }
    assert!(
        sequences.is_empty() || timestamps.is_empty(),
        "a backend must not mix offset kinds in one stream"
    );
    assert!(
        sequences.windows(2).all(|pair| pair[0] < pair[1]),
        "tag-query sequence offsets must follow strict global append order: {sequences:?}"
    );
    assert!(
        timestamps.windows(2).all(|pair| pair[0] <= pair[1]),
        "tag-query timestamp offsets must be non-decreasing: {timestamps:?}"
    );
}

fn snapshot_metadata(
    persistence_id: &PersistenceId,
    sequence_nr: u64,
    timestamp_ms: i64,
) -> SnapshotMetadata {
    SnapshotMetadata {
        persistence_id: persistence_id.clone(),
        sequence_nr: SequenceNr::new(sequence_nr),
        timestamp_ms,
    }
}

fn require_snapshot(snapshot: Option<SelectedSnapshot>) -> SelectedSnapshot {
    snapshot.expect("a matching saved snapshot must be returned")
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::{MemoryDurableStateStore, MemoryJournal, MemorySnapshotStore};

    #[tokio::test]
    async fn memory_backends_pass_the_public_conformance_suite() -> PersistenceResult<()> {
        assert_journal_conformance(&MemoryJournal::new()).await?;
        assert_snapshot_store_conformance(&MemorySnapshotStore::new()).await?;
        assert_durable_state_store_conformance(&MemoryDurableStateStore::new()).await?;
        Ok(())
    }

    struct FailingSnapshotStore;

    #[async_trait::async_trait]
    impl SnapshotStore for FailingSnapshotStore {
        async fn load(
            &self,
            _persistence_id: &PersistenceId,
            _criteria: &SnapshotCriteria,
        ) -> PersistenceResult<Option<SelectedSnapshot>> {
            Err(PersistenceError::Failure(
                "injected snapshot load failure".to_owned(),
            ))
        }

        async fn save(
            &self,
            _metadata: SnapshotMetadata,
            _adapter_manifest: String,
            _state: SerializedPayload,
        ) -> PersistenceResult<()> {
            Ok(())
        }

        async fn delete(&self, _metadata: SnapshotMetadata) -> PersistenceResult<()> {
            Ok(())
        }

        async fn delete_matching(
            &self,
            _persistence_id: &PersistenceId,
            _criteria: &SnapshotCriteria,
        ) -> PersistenceResult<()> {
            Ok(())
        }
    }

    #[tokio::test]
    async fn snapshot_conformance_does_not_treat_load_failure_as_absence() {
        let error = assert_snapshot_store_conformance(&FailingSnapshotStore)
            .await
            .expect_err("the injected load failure must propagate");
        assert!(
            matches!(error, PersistenceError::Failure(message) if message.contains("injected"))
        );
    }
}