object-log 0.4.0

A buffered, multiplexing append log over pluggable object storage, with a sequencing seam (opaque payloads, durable-on-return writes, group-commit).
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
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
//! Content-addressed immutable pages, sharing one store-level byte admission.
//! This bounds owned page/key sources; it is not a total HTTP/TLS/RSS bound.
use super::codec::{
    BRANCH_SLOTS, DESCRIPTOR_BYTES, Descriptor, HEADER, Kind, LEAF_SLOTS, LeafEntry, PAGE_BYTES,
    Page, Scope,
};
use crate::{
    AppendOwner, BlobStore, BoundedMetadataStore, MetadataLimits, MetadataOwners, MetadataSwap,
    ObjectLogError,
};
use bytes::Bytes;
mod batch;
mod immutable;
mod scoped_bytes;
mod transition_key;
pub(super) use batch::{Upload, UploadBatch};
pub(super) use scoped_bytes::{EncodedBytes, RetainedBytes};
use std::sync::{
    Arc,
    atomic::{AtomicU64, Ordering},
};

pub(super) trait NativeStore: BlobStore + BoundedMetadataStore {
    fn page_source_scratch(
        &self,
        key_len: usize,
        limits: MetadataLimits,
    ) -> Result<usize, ObjectLogError>;
    /// Observe absence of `key` without I/O, for a create-only conditional write.
    fn absent(&self, key: &str, retained: AppendOwner) -> crate::MetadataSnapshot;
}
impl NativeStore for crate::LocalBlobStore {
    fn page_source_scratch(
        &self,
        key_len: usize,
        limits: MetadataLimits,
    ) -> Result<usize, ObjectLogError> {
        limits.validate_local()?;
        self.native_page_source_scratch(key_len)
    }
    fn absent(&self, key: &str, retained: AppendOwner) -> crate::MetadataSnapshot {
        self.absent_metadata(key, retained)
    }
}

#[cfg(feature = "s3")]
impl NativeStore for crate::S3BlobStore {
    fn page_source_scratch(
        &self,
        key_len: usize,
        limits: MetadataLimits,
    ) -> Result<usize, ObjectLogError> {
        limits.validate()?;
        self.native_page_source_scratch(key_len, limits.version_bytes)
    }
    fn absent(&self, key: &str, retained: AppendOwner) -> crate::MetadataSnapshot {
        self.absent_metadata(key, retained)
    }
}

pub(super) struct ByteBudget {
    used: AtomicU64,
    capacity: u64,
}
impl ByteBudget {
    pub fn new(capacity: u64) -> Result<Arc<Self>, ObjectLogError> {
        if capacity == 0 || isize::try_from(capacity).is_err() {
            return Err(ObjectLogError::InvalidConfig("native byte capacity".into()));
        }
        Ok(Arc::new(Self {
            used: AtomicU64::new(0),
            capacity,
        }))
    }
    pub fn charge(self: &Arc<Self>, bytes: usize) -> Result<Arc<Charge>, ObjectLogError> {
        let count = u64::try_from(bytes)
            .map_err(|_| ObjectLogError::BudgetExceeded("native byte capacity".into()))?;
        if count == 0 || count > self.capacity {
            return Err(ObjectLogError::BudgetExceeded(
                "native byte capacity".into(),
            ));
        }
        // Install the local release owner before reservation; no fallible work
        // intervenes between the successful CAS and arming exact-once release.
        let mut charge = Charge {
            budget: self.clone(),
            count,
            reserved: false,
        };
        self.used
            .try_update(Ordering::AcqRel, Ordering::Acquire, |used| {
                used.checked_add(count)
                    .filter(|next| *next <= self.capacity)
            })
            .map_err(|_| ObjectLogError::BudgetExceeded("native shared byte admission".into()))?;
        charge.reserved = true;
        Ok(Arc::new(charge))
    }
    #[cfg(test)]
    pub fn available(&self) -> usize {
        usize::try_from(self.capacity - self.used.load(Ordering::Acquire))
            .expect("validated native byte capacity fits usize")
    }
}
pub(super) struct Charge {
    budget: Arc<ByteBudget>,
    count: u64,
    reserved: bool,
}
impl Drop for Charge {
    fn drop(&mut self) {
        if self.reserved {
            self.budget.used.fetch_sub(self.count, Ordering::AcqRel);
        }
    }
}

/// Decoded capacity cannot escape its charge through the page-store API.
pub(super) struct OwnedPage {
    page: Page,
    store_identity: Arc<()>,
    _charge: Arc<Charge>,
}
pub(super) enum PageView<'a> {
    Leaf(&'a [LeafEntry]),
    Branch {
        height: u8,
        children: &'a [Descriptor],
    },
    Overflow {
        start: u64,
        bytes: &'a [u8],
        next: Option<Descriptor>,
    },
}
impl OwnedPage {
    pub fn view(&self) -> PageView<'_> {
        match &self.page {
            Page::Leaf(entries) => PageView::Leaf(entries),
            Page::Branch { height, children } => PageView::Branch {
                height: *height,
                children,
            },
            Page::Overflow { start, bytes, next } => PageView::Overflow {
                start: *start,
                bytes,
                next: *next,
            },
        }
    }
    pub fn push_leaf(&mut self, entry: LeafEntry) -> Result<(), ObjectLogError> {
        entry.end()?;
        match &mut self.page {
            Page::Leaf(entries)
                if entries.len() < LEAF_SLOTS && entries.last().unwrap().end()? == entry.base =>
            {
                entries.push(entry);
                Ok(())
            }
            _ => Err(ObjectLogError::InvalidBatch("bounded leaf append".into())),
        }
    }
    pub fn replace_last_child(&mut self, child: Descriptor) -> Result<(), ObjectLogError> {
        child.validate()?;
        match &mut self.page {
            Page::Branch { height, children }
                if child.kind != Kind::Overflow
                    && child.height + 1 == *height
                    && children.last().unwrap().start == child.start =>
            {
                *children.last_mut().unwrap() = child;
                Ok(())
            }
            _ => Err(ObjectLogError::InvalidBatch(
                "bounded branch replacement".into(),
            )),
        }
    }
    pub fn push_child(&mut self, child: Descriptor) -> Result<(), ObjectLogError> {
        child.validate()?;
        match &mut self.page {
            Page::Branch { height, children }
                if children.len() < BRANCH_SLOTS
                    && child.kind != Kind::Overflow
                    && child.height + 1 == *height
                    && children.last().unwrap().end == child.start =>
            {
                children.push(child);
                Ok(())
            }
            _ => Err(ObjectLogError::InvalidBatch("bounded branch append".into())),
        }
    }
}

/// An encoded immutable page awaiting its create-only upload.
pub(super) struct PendingPage {
    raw: Vec<u8>,
    hash: [u8; 32],
    // Released only after the upload completes.
    charge: Arc<Charge>,
}

struct UploadedPage {
    bytes: Vec<u8>,
    // Drop the complete source allocation before its reservation.
    _charge: Arc<Charge>,
}
impl AsRef<[u8]> for UploadedPage {
    fn as_ref(&self) -> &[u8] {
        &self.bytes
    }
}

pub(super) struct OwnedDataKey {
    key: String,
    _charge: Arc<Charge>,
}
impl OwnedDataKey {
    pub fn as_str(&self) -> &str {
        &self.key
    }
}

#[derive(Clone)]
pub(super) struct PageBinding {
    identity: Arc<()>,
    store_identity: Arc<()>,
    prefix: Arc<str>,
    scope: Scope,
    allocation: usize,
    _binding_charge: Arc<Charge>,
}
impl PageBinding {
    pub(super) fn same(&self, other: &Self) -> bool {
        Arc::ptr_eq(&self.identity, &other.identity)
    }
}
pub(super) struct PageStore {
    store: Arc<dyn NativeStore>,
    identity: Arc<()>,
    store_identity: Arc<()>,
    prefix: Arc<str>,
    scope: Scope,
    budget: Arc<ByteBudget>,
    allocation: usize,
    limits: MetadataLimits,
    physical: Option<Arc<super::operation::Physical>>,
    _binding_charge: Option<Arc<Charge>>,
    _container_charge: Option<Arc<Charge>>,
}
impl PageStore {
    pub(super) fn planned_allocation(
        store: &dyn NativeStore,
        prefix: &str,
        limits: MetadataLimits,
    ) -> Result<usize, ObjectLogError> {
        Self::planned_allocation_with(prefix, limits, |key_len, limits| {
            store.page_source_scratch(key_len, limits)
        })
    }
    pub(super) fn planned_allocation_with(
        prefix: &str,
        limits: MetadataLimits,
        scratch: impl FnOnce(usize, MetadataLimits) -> Result<usize, ObjectLogError>,
    ) -> Result<usize, ObjectLogError> {
        Self::planned_allocation_for_len(prefix.len(), limits, scratch)
    }
    pub(super) fn planned_allocation_for_len(
        prefix_len: usize,
        limits: MetadataLimits,
        scratch: impl FnOnce(usize, MetadataLimits) -> Result<usize, ObjectLogError>,
    ) -> Result<usize, ObjectLogError> {
        // Config-owned prefix is shared. Reserve the complete generated key,
        // adapter key copy, encoded page, decoded capacity and source scratch.
        limits.validate()?;
        if limits.payload_bytes < PAGE_BYTES {
            return Err(ObjectLogError::InvalidConfig(
                "native metadata payload limit cannot hold a page".into(),
            ));
        }
        let key_len = prefix_len
            .checked_add(195)
            .ok_or_else(|| ObjectLogError::InvalidConfig("native key length overflow".into()))?;
        let allocation = scratch(key_len, limits)?
            .checked_add(PAGE_BYTES * 4)
            .ok_or_else(|| {
                ObjectLogError::InvalidConfig("native page allocation overflow".into())
            })?;
        Ok(allocation)
    }
    pub fn new(
        store: Arc<dyn NativeStore>,
        prefix: String,
        scope: Scope,
        budget: Arc<ByteBudget>,
        limits: MetadataLimits,
    ) -> Result<Arc<Self>, ObjectLogError> {
        if prefix.is_empty()
            || prefix.starts_with('/')
            || prefix.contains('\0')
            || prefix
                .split('/')
                .any(|part| part == ".." || part == "." || part.is_empty())
            || scope.namespace == [0; 32]
            || scope.incarnation == [0; 32]
        {
            return Err(ObjectLogError::InvalidConfig(
                "native page store binding".into(),
            ));
        }
        let allocation = Self::planned_allocation(store.as_ref(), &prefix, limits)?;
        let capacity = usize::try_from(budget.capacity)
            .map_err(|_| ObjectLogError::InvalidConfig("native byte capacity".into()))?;
        if allocation > capacity {
            return Err(ObjectLogError::InvalidConfig(
                "native budget cannot hold one page".into(),
            ));
        }
        let identity = Arc::new(());
        Ok(Arc::new(Self {
            store,
            store_identity: identity.clone(),
            identity,
            prefix: prefix.into(),
            scope,
            budget,
            allocation,
            limits,
            physical: None,
            _binding_charge: None,
            _container_charge: None,
        }))
    }
    /// Nested page operations reuse the admitted outer operation. Returned
    /// pages contain only byte guards and binding identity, never this session.
    pub fn in_operation(
        &self,
        physical: Arc<super::operation::Physical>,
    ) -> Result<Arc<Self>, ObjectLogError> {
        physical.validate_budget(&self.budget)?;
        Ok(Arc::new(Self {
            store: self.store.clone(),
            identity: self.identity.clone(),
            store_identity: self.store_identity.clone(),
            prefix: self.prefix.clone(),
            scope: self.scope,
            budget: self.budget.clone(),
            allocation: self.allocation,
            limits: self.limits,
            physical: Some(physical),
            _binding_charge: self._binding_charge.clone(),
            _container_charge: if self._binding_charge.is_some() {
                Some(self.charge_output(
                    std::mem::size_of::<Self>() + 2 * std::mem::size_of::<usize>(),
                )?)
            } else {
                None
            },
        }))
    }
    pub fn derive(&self, prefix: String, scope: Scope) -> Result<Arc<Self>, ObjectLogError> {
        self.require_operation()?;
        let bytes = prefix
            .len()
            .checked_mul(2)
            .and_then(|n| {
                n.checked_add(2 * std::mem::size_of::<Self>() + 4 * std::mem::size_of::<usize>())
            })
            .ok_or_else(|| {
                ObjectLogError::BudgetExceeded("derived binding allocation overflow".into())
            })?;
        let charge = self.charge_output(bytes)?;
        let mut pages = Self::new(
            self.store.clone(),
            prefix,
            scope,
            self.budget.clone(),
            self.limits,
        )?;
        let inner = Arc::get_mut(&mut pages).expect("new exclusive binding");
        inner.store_identity = self.store_identity.clone();
        inner._binding_charge = Some(charge);
        pages.in_operation(self.physical.as_ref().expect("checked operation").clone())
    }
    pub fn seal_binding(&self) -> Result<PageBinding, ObjectLogError> {
        let charge = self._binding_charge.clone().ok_or_else(|| {
            ObjectLogError::InvalidConfig("only admitted derived bindings can be sealed".into())
        })?;
        Ok(PageBinding {
            identity: self.identity.clone(),
            store_identity: self.store_identity.clone(),
            prefix: self.prefix.clone(),
            scope: self.scope,
            allocation: self.allocation,
            _binding_charge: charge,
        })
    }
    pub fn for_binding(&self, binding: &PageBinding) -> Result<Arc<Self>, ObjectLogError> {
        self.require_operation()?;
        if !Arc::ptr_eq(&self.store_identity, &binding.store_identity) {
            return Err(ObjectLogError::InvalidConfig(
                "selected binding belongs to another store".into(),
            ));
        }
        let charge =
            self.charge_output(std::mem::size_of::<Self>() + 2 * std::mem::size_of::<usize>())?;
        Ok(Arc::new(Self {
            store: self.store.clone(),
            identity: binding.identity.clone(),
            store_identity: self.store_identity.clone(),
            prefix: binding.prefix.clone(),
            scope: binding.scope,
            budget: self.budget.clone(),
            allocation: binding.allocation,
            limits: self.limits,
            physical: self.physical.clone(),
            _binding_charge: Some(binding._binding_charge.clone()),
            _container_charge: Some(charge),
        }))
    }
    fn owners(&self, retained: AppendOwner) -> MetadataOwners {
        MetadataOwners {
            physical: self
                .physical
                .clone()
                .map_or_else(|| retained.clone(), |physical| physical as AppendOwner),
            retained,
        }
    }
    pub fn require_operation(&self) -> Result<(), ObjectLogError> {
        if self.physical.is_none() {
            return Err(ObjectLogError::InvalidConfig(
                "native authority requires operation admission".into(),
            ));
        }
        Ok(())
    }
    pub fn scope(&self) -> Scope {
        self.scope
    }
    pub fn binding_identity(&self) -> Arc<()> {
        self.identity.clone()
    }
    pub fn validate_authority_key(&self, key: &str) -> Result<(), ObjectLogError> {
        if key == self.prefix.as_ref()
            || key
                .strip_prefix(self.prefix.as_ref())
                .is_some_and(|suffix| suffix.starts_with('/'))
            || key.is_empty()
            || key.starts_with('/')
            || key.contains('\0')
            || key
                .split('/')
                .any(|part| part.is_empty() || part == "." || part == "..")
        {
            return Err(ObjectLogError::InvalidConfig("native head key".into()));
        }
        Ok(())
    }
    pub async fn read_head(
        &self,
        key: &str,
        bytes: usize,
    ) -> Result<crate::MetadataSnapshot, ObjectLogError> {
        self.require_operation()?;
        self.validate_authority_key(key)?;
        let charge = self.budget.charge(
            self.store
                .page_source_scratch(key.len(), self.metadata_limits(bytes))?
                .checked_add(bytes)
                .ok_or_else(|| ObjectLogError::BudgetExceeded("head allocation overflow".into()))?,
        )?;
        self.store
            .read_metadata(
                key,
                self.metadata_limits(bytes),
                self.owners(charge as AppendOwner),
            )
            .await
    }
    pub fn data_key(&self, logical: &str) -> Result<OwnedDataKey, ObjectLogError> {
        let len = self
            .prefix
            .len()
            .checked_add(200)
            .ok_or_else(|| ObjectLogError::BudgetExceeded("data key overflow".into()))?;
        let charge = self.budget.charge(
            len.checked_add(std::mem::size_of::<String>())
                .ok_or_else(|| ObjectLogError::BudgetExceeded("data key overflow".into()))?,
        )?;
        use std::fmt::Write;
        let mut key = String::with_capacity(len);
        key.push_str(&self.prefix);
        for identity in [&self.scope.namespace, &self.scope.incarnation] {
            key.push('/');
            for byte in identity {
                write!(&mut key, "{byte:02x}").expect("String formatting");
            }
        }
        key.push_str("/data/");
        use sha2::{Digest, Sha256};
        for byte in Sha256::digest(logical.as_bytes()) {
            write!(&mut key, "{byte:02x}").expect("String formatting");
        }
        Ok(OwnedDataKey {
            key,
            _charge: charge,
        })
    }
    pub fn validate_data_limit(&self, bytes: usize) -> Result<(), ObjectLogError> {
        if bytes > self.limits.payload_bytes {
            return Err(ObjectLogError::BudgetExceeded(
                "native data payload limit".into(),
            ));
        }
        Ok(())
    }
    pub fn preflight_data(&self, key: &str, bytes: usize) -> Result<(), ObjectLogError> {
        self.require_operation()?;
        self.validate_data_limit(bytes)?;
        let required =
            self.store
                .page_source_scratch(key.len(), self.metadata_limits(bytes.max(1)))?
                .checked_add(bytes.checked_mul(2).ok_or_else(|| {
                    ObjectLogError::BudgetExceeded("data allocation overflow".into())
                })?)
                .ok_or_else(|| ObjectLogError::BudgetExceeded("data allocation overflow".into()))?;
        // Capacity check before creating overflow/data objects. This probe is
        // released immediately; concurrent work may exhaust later admission.
        drop(self.budget.charge(required)?);
        Ok(())
    }
    pub async fn read_data(
        &self,
        key: &str,
        bytes: usize,
    ) -> Result<crate::MetadataSnapshot, ObjectLogError> {
        self.require_operation()?;
        if bytes == 0 || bytes > self.limits.payload_bytes {
            return Err(ObjectLogError::BudgetExceeded(
                "native data read limit".into(),
            ));
        }
        let charge = self.budget.charge(
            self.store
                .page_source_scratch(key.len(), self.metadata_limits(bytes))?
                .checked_add(bytes)
                .ok_or_else(|| {
                    ObjectLogError::BudgetExceeded("data read allocation overflow".into())
                })?,
        )?;
        self.store
            .read_metadata(
                key,
                self.metadata_limits(bytes),
                self.owners(charge as AppendOwner),
            )
            .await
    }
    pub async fn upload_data(&self, key: &str, raw: &[u8]) -> Result<(), ObjectLogError> {
        self.require_operation()?;
        super::data::DataFrame::decode(raw)?;
        if raw.len() > self.limits.payload_bytes {
            return Err(ObjectLogError::BudgetExceeded(
                "native data payload limit".into(),
            ));
        }
        let charge = self.budget.charge(
            self.store
                .page_source_scratch(key.len(), self.metadata_limits(raw.len().max(1)))?
                .checked_add(raw.len().checked_mul(2).ok_or_else(|| {
                    ObjectLogError::BudgetExceeded("data allocation overflow".into())
                })?)
                .ok_or_else(|| ObjectLogError::BudgetExceeded("data allocation overflow".into()))?,
        )?;
        let limits = self.metadata_limits(raw.len().max(1));
        let source = Bytes::from_owner(UploadedPage {
            bytes: raw.to_vec(),
            _charge: charge.clone(),
        });
        // Create first: an existing object is read only after the store rejects
        // the create, and must then match this frame exactly.
        let absent = self.store.absent(key, charge.clone() as AppendOwner);
        match self
            .store
            .swap_metadata(
                key,
                absent,
                source.clone(),
                limits,
                self.owners(charge as AppendOwner),
            )
            .await?
        {
            MetadataSwap::Stored(_) => Ok(()),
            MetadataSwap::Conflict(current) => {
                let existing = current.bytes().ok_or_else(|| {
                    ObjectLogError::Sequencer("immutable data publication conflicts".into())
                })?;
                // Framed data reuse requires exact literal identity and payload;
                // no physical-key digest can substitute for this comparison.
                let expected = super::data::DataFrame::decode(&source)?;
                let actual = super::data::DataFrame::decode(existing)?;
                if actual.identity() != expected.identity()
                    || actual.payload() != expected.payload()
                {
                    return Err(ObjectLogError::Sequencer(
                        "immutable data identity differs".into(),
                    ));
                }
                if existing == source.as_ref() {
                    return Ok(());
                }
                Err(ObjectLogError::Sequencer(
                    "immutable data location differs".into(),
                ))
            }
        }
    }
    pub async fn swap_head(
        &self,
        key: &str,
        observed: crate::MetadataSnapshot,
        raw: &[u8],
    ) -> Result<MetadataSwap, ObjectLogError> {
        self.require_operation()?;
        if observed.inner.key != key {
            return Err(ObjectLogError::InvalidConfig(
                "head literal key mismatch".into(),
            ));
        }
        let charge = self.budget.charge(
            self.store
                .page_source_scratch(key.len(), self.metadata_limits(raw.len()))?
                .checked_add(raw.len())
                .ok_or_else(|| ObjectLogError::BudgetExceeded("head allocation overflow".into()))?,
        )?;
        // Admission precedes the complete upload allocation. The adapter binds
        // the original observation to this exact provider and literal key.
        let source = Bytes::from_owner(UploadedPage {
            bytes: raw.to_vec(),
            _charge: charge.clone(),
        });
        self.store
            .swap_metadata(
                key,
                observed,
                source,
                self.metadata_limits(raw.len()),
                self.owners(charge as AppendOwner),
            )
            .await
    }
    fn metadata_limits(&self, payload_bytes: usize) -> MetadataLimits {
        MetadataLimits {
            payload_bytes,
            version_bytes: self.limits.version_bytes,
        }
    }
    #[cfg(test)]
    pub fn allocation(&self) -> usize {
        self.allocation
    }
    pub fn charge_output(&self, bytes: usize) -> Result<Arc<Charge>, ObjectLogError> {
        self.budget.charge(bytes)
    }
    fn owned(&self, page: Page, charge: Arc<Charge>) -> Result<OwnedPage, ObjectLogError> {
        // Field-ordered ownership exists before any fallible codec work.
        let owned = OwnedPage {
            page,
            store_identity: self.identity.clone(),
            _charge: charge,
        };
        owned.page.encode(self.scope)?;
        Ok(owned)
    }
    pub fn overflow(
        &self,
        start: u64,
        bytes: &[u8],
        next: Option<Descriptor>,
    ) -> Result<OwnedPage, ObjectLogError> {
        if bytes.is_empty() || bytes.len() > PAGE_BYTES - HEADER - DESCRIPTOR_BYTES {
            return Err(ObjectLogError::InvalidBatch(
                "bounded overflow length".into(),
            ));
        }
        let charge = self.budget.charge(self.allocation)?;
        self.owned(
            Page::Overflow {
                start,
                bytes: bytes.to_vec(),
                next,
            },
            charge,
        )
    }
    pub fn leaf(&self, entries: &[LeafEntry]) -> Result<OwnedPage, ObjectLogError> {
        if entries.is_empty() || entries.len() > LEAF_SLOTS {
            return Err(ObjectLogError::InvalidBatch("bounded leaf slots".into()));
        }
        let charge = self.budget.charge(self.allocation)?;
        let mut owned = Vec::with_capacity(LEAF_SLOTS);
        owned.extend_from_slice(entries);
        self.owned(Page::Leaf(owned), charge)
    }
    pub fn branch(&self, height: u8, children: &[Descriptor]) -> Result<OwnedPage, ObjectLogError> {
        if children.is_empty() || children.len() > BRANCH_SLOTS {
            return Err(ObjectLogError::InvalidBatch("bounded branch slots".into()));
        }
        let charge = self.budget.charge(self.allocation)?;
        let mut owned = Vec::with_capacity(BRANCH_SLOTS);
        owned.extend_from_slice(children);
        self.owned(
            Page::Branch {
                height,
                children: owned,
            },
            charge,
        )
    }
    pub async fn read(&self, descriptor: Descriptor) -> Result<OwnedPage, ObjectLogError> {
        descriptor.validate()?;
        let charge = self.budget.charge(self.allocation)?;
        let observed = self
            .read_immutable(&descriptor.hash, descriptor.encoded_len as usize, &charge)
            .await?;
        let raw = observed
            .bytes()
            .ok_or_else(|| ObjectLogError::MissingObject("native immutable page".into()))?;
        let mut page = Page::decode(raw, descriptor, self.scope)?;
        match &mut page {
            Page::Leaf(entries) => entries.reserve_exact(LEAF_SLOTS - entries.len()),
            Page::Branch { children, .. } => children.reserve_exact(BRANCH_SLOTS - children.len()),
            Page::Overflow { .. } => (),
        }
        Ok(OwnedPage {
            page,
            store_identity: self.identity.clone(),
            _charge: charge,
        })
    }
    /// Encode a page without I/O. The pending page keeps the page's reservation
    /// until it is uploaded, even after the page itself is dropped or mutated.
    pub fn seal(&self, page: &OwnedPage) -> Result<(Descriptor, PendingPage), ObjectLogError> {
        if !Arc::ptr_eq(&page.store_identity, &self.identity) {
            return Err(ObjectLogError::InvalidConfig(
                "page belongs to another store binding".into(),
            ));
        }
        // Encode under the existing page reservation, retaining both capacities.
        let (raw, descriptor) = page.page.encode(self.scope)?;
        Ok((
            descriptor,
            PendingPage {
                raw,
                hash: descriptor.hash,
                charge: page._charge.clone(),
            },
        ))
    }
    #[cfg(test)]
    pub async fn upload(&self, pending: PendingPage) -> Result<(), ObjectLogError> {
        self.write_immutable(pending.raw, &pending.hash, &pending.charge)
            .await
    }
    #[cfg(test)]
    pub async fn write(&self, page: OwnedPage) -> Result<Descriptor, ObjectLogError> {
        let (descriptor, pending) = self.seal(&page)?;
        self.upload(pending).await?;
        Ok(descriptor)
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::LocalBlobStore;

    fn scope() -> Scope {
        Scope {
            namespace: [1; 32],
            incarnation: [2; 32],
        }
    }

    #[tokio::test]
    async fn real_local_pages_keep_old_roots_and_fail_missing_or_corrupt() {
        let root = tempfile::tempdir().unwrap();
        let local = Arc::new(LocalBlobStore::new(root.path()));
        let budget = ByteBudget::new(1 << 20).unwrap();
        let store = PageStore::new(
            local.clone(),
            "pages".into(),
            scope(),
            budget.clone(),
            crate::MetadataLimits {
                payload_bytes: PAGE_BYTES,
                version_bytes: 56,
            },
        )
        .unwrap();
        let first = store
            .write(store.overflow(0, b"old/key", None).unwrap())
            .await
            .unwrap();
        let second = store
            .write(store.overflow(0, b"new/key", None).unwrap())
            .await
            .unwrap();
        assert_ne!(first.hash, second.hash);
        let old = store.read(first).await.unwrap();
        assert!(matches!(
            old.view(),
            PageView::Overflow {
                bytes: b"old/key",
                ..
            }
        ));
        assert!(budget.available() < 1 << 20);
        let mut new = store.read(second).await.unwrap();
        // Mutating decoded memory cannot alter a committed immutable page.
        if let Page::Overflow { bytes, .. } = &mut new.page {
            bytes[0] ^= 1;
        }
        drop(new);
        assert!(matches!(
            store.read(second).await.unwrap().view(),
            PageView::Overflow {
                bytes: b"new/key",
                ..
            }
        ));
        drop(old);
        assert_eq!(budget.available(), 1 << 20);
        let key = store.key(first);
        local
            .put(&key, Bytes::from_static(b"corrupt"))
            .await
            .unwrap();
        assert!(store.read(first).await.is_err());
        local.delete(&key).await.unwrap();
        assert!(matches!(
            store.read(first).await,
            Err(ObjectLogError::MissingObject(_))
        ));
        assert_eq!(budget.available(), 1 << 20);
    }

    #[tokio::test]
    async fn exact_charge_precedes_owned_allocation_and_io_and_is_store_bound() {
        let root = tempfile::tempdir().unwrap();
        let local = Arc::new(LocalBlobStore::new(root.path()));
        let sizing = PageStore::new(
            local.clone(),
            "pages".into(),
            scope(),
            ByteBudget::new(1 << 20).unwrap(),
            crate::MetadataLimits {
                payload_bytes: PAGE_BYTES,
                version_bytes: 56,
            },
        )
        .unwrap();
        let bytes = sizing.allocation();
        assert!(
            PageStore::new(
                local.clone(),
                "pages".into(),
                scope(),
                ByteBudget::new(u64::try_from(bytes).unwrap() - 1).unwrap(),
                crate::MetadataLimits {
                    payload_bytes: PAGE_BYTES,
                    version_bytes: 56
                }
            )
            .is_err()
        );
        let budget = ByteBudget::new(u64::try_from(bytes).unwrap()).unwrap();
        let store = PageStore::new(
            local.clone(),
            "pages".into(),
            scope(),
            budget.clone(),
            crate::MetadataLimits {
                payload_bytes: PAGE_BYTES,
                version_bytes: 56,
            },
        )
        .unwrap();
        let held = store.overflow(0, b"key", None).unwrap();
        assert_eq!(budget.available(), 0);
        assert!(store.overflow(0, b"key", None).is_err());
        assert!(local.list("pages").await.unwrap().is_empty());
        let foreign = PageStore::new(
            local.clone(),
            "foreign".into(),
            scope(),
            ByteBudget::new(1 << 20).unwrap(),
            crate::MetadataLimits {
                payload_bytes: PAGE_BYTES,
                version_bytes: 56,
            },
        )
        .unwrap();
        assert!(foreign.write(held).await.is_err());
        assert_eq!(budget.available(), bytes);
        assert!(local.list("pages").await.unwrap().is_empty());
        let descriptor = store
            .write(store.overflow(0, b"key", None).unwrap())
            .await
            .unwrap();
        assert_eq!(budget.available(), bytes);
        let read = store.read(descriptor).await.unwrap();
        assert_eq!(budget.available(), 0);
        assert!(store.read(descriptor).await.is_err());
        drop(read);
        assert_eq!(budget.available(), bytes);
    }

    #[tokio::test(flavor = "multi_thread", worker_threads = 2)]
    async fn independent_same_page_writers_create_once_and_refuse_existing_corruption() {
        let root = tempfile::tempdir().unwrap();
        let first_local = Arc::new(LocalBlobStore::new(root.path()));
        let second_local = Arc::new(LocalBlobStore::new(root.path()));
        let budget = ByteBudget::new(1 << 20).unwrap();
        let first = PageStore::new(
            first_local.clone(),
            "pages".into(),
            scope(),
            budget.clone(),
            crate::MetadataLimits {
                payload_bytes: PAGE_BYTES,
                version_bytes: 56,
            },
        )
        .unwrap();
        let second = PageStore::new(
            second_local.clone(),
            "pages".into(),
            scope(),
            budget.clone(),
            crate::MetadataLimits {
                payload_bytes: PAGE_BYTES,
                version_bytes: 56,
            },
        )
        .unwrap();
        let (entered, release) = first_local.pause_metadata_swap(false);
        let writer = first.clone();
        let task = tokio::spawn(async move {
            writer
                .write(writer.overflow(0, b"same/key", None).unwrap())
                .await
        });
        tokio::task::spawn_blocking(move || {
            entered
                .recv_timeout(std::time::Duration::from_secs(10))
                .unwrap()
        })
        .await
        .unwrap();
        let writer = second.clone();
        let other = tokio::spawn(async move {
            writer
                .write(writer.overflow(0, b"same/key", None).unwrap())
                .await
        });
        // Closing this channel on an assertion failure also releases the first CAS.
        release.send(()).unwrap();
        let first_descriptor = task.await.unwrap().unwrap();
        assert_eq!(other.await.unwrap().unwrap(), first_descriptor);
        assert!(matches!(
            second.read(first_descriptor).await.unwrap().view(),
            PageView::Overflow {
                bytes: b"same/key",
                ..
            }
        ));
        assert_eq!(budget.available(), 1 << 20);
        let key = first.key(first_descriptor);
        let original = first_local.get(&key).await.unwrap().unwrap();
        let mut corrupt = original.to_vec();
        *corrupt.last_mut().unwrap() ^= 1;
        first_local
            .put(&key, Bytes::from(corrupt.clone()))
            .await
            .unwrap();
        assert!(
            second
                .write(second.overflow(0, b"same/key", None).unwrap())
                .await
                .is_err()
        );
        assert_eq!(
            first_local.get(&key).await.unwrap().unwrap().as_ref(),
            corrupt
        );
        assert_eq!(budget.available(), 1 << 20);
        assert!(
            PageStore::new(
                first_local,
                "/absolute".into(),
                scope(),
                budget,
                crate::MetadataLimits {
                    payload_bytes: PAGE_BYTES,
                    version_bytes: 56
                }
            )
            .is_err()
        );
    }
}

#[cfg(test)]
mod density_tests {
    use super::*;
    use crate::native_v2::operation::{ExecutionConfig, Resources};

    #[tokio::test]
    async fn thousand_known_bindings_share_executor_and_byte_admission() {
        let directory = tempfile::tempdir().unwrap();
        let local = Arc::new(crate::LocalBlobStore::new(directory.path()));
        let resources = Resources::new(ExecutionConfig {
            workers: 2,
            blocking_workers: 2,
            operations: 1,
            bytes: 1 << 20,
        })
        .unwrap();
        let mut bindings = Vec::with_capacity(1000);
        for namespace in 1u64..=1000 {
            let mut identity = [0; 32];
            identity[..8].copy_from_slice(&namespace.to_be_bytes());
            bindings.push(
                PageStore::new(
                    local.clone(),
                    "pages".into(),
                    Scope {
                        namespace: identity,
                        incarnation: [2; 32],
                    },
                    resources.bytes.clone(),
                    MetadataLimits {
                        payload_bytes: PAGE_BYTES,
                        version_bytes: 56,
                    },
                )
                .unwrap(),
            );
        }
        assert!(local.list("pages").await.unwrap().is_empty());
        let budget = resources.bytes.clone();
        let (held, identity) = resources
            .run(move |physical| async move {
                let mut held = Vec::new();
                let mut rejected = false;
                for binding in &bindings {
                    assert!(Arc::ptr_eq(&binding.budget, &budget));
                    let scoped = binding.in_operation(physical.clone())?;
                    match scoped.overflow(0, b"physical/key", None) {
                        Ok(page) => held.push(page),
                        Err(ObjectLogError::BudgetExceeded(_)) => {
                            rejected = true;
                            break;
                        }
                        Err(error) => return Err(error),
                    }
                }
                assert!(
                    rejected,
                    "all namespaces must share a single finite byte pool"
                );
                assert!(!held.is_empty());
                let weak = Arc::downgrade(&physical);
                Ok((held, weak))
            })
            .await
            .unwrap();
        resources.close_and_drain().await.unwrap();
        assert!(identity.upgrade().is_none());
        assert!(resources.bytes.available() < 1 << 20);
        drop(held);
        assert_eq!(resources.bytes.available(), 1 << 20);
    }
}

#[cfg(test)]
mod data_namespace_tests {
    use super::*;
    use crate::native_v2::operation::{ExecutionConfig, Resources};

    #[tokio::test]
    async fn logical_aliases_and_long_ids_have_disjoint_canonical_data_objects() {
        let directory = tempfile::tempdir().unwrap();
        let local = Arc::new(crate::LocalBlobStore::new(directory.path()));
        let resources = Resources::new(ExecutionConfig {
            workers: 2,
            blocking_workers: 2,
            operations: 1,
            bytes: 1 << 20,
        })
        .unwrap();
        let pages = PageStore::new(
            local.clone(),
            "pages".into(),
            Scope {
                namespace: [1; 32],
                incarnation: [2; 32],
            },
            resources.bytes.clone(),
            MetadataLimits {
                payload_bytes: PAGE_BYTES,
                version_bytes: 56,
            },
        )
        .unwrap();
        resources
            .run(move |physical| async move {
                let pages = pages.in_operation(physical)?;
                let before = pages.budget.available();
                let owned = pages.data_key("reservation-control")?;
                let required = before - pages.budget.available();
                assert!(required >= owned.as_str().len());
                drop(owned);
                assert_eq!(pages.budget.available(), before);
                let pressure = pages.budget.charge(before - required + 1)?;
                assert!(pages.data_key("reservation-control").is_err());
                drop(pressure);
                let pressure = pages.budget.charge(before - required)?;
                let owned = pages.data_key("reservation-control")?;
                assert_eq!(pages.budget.available(), 0);
                drop(owned);
                assert_eq!(pages.budget.available(), required);
                drop(pressure);
                let identities = [
                    "head".to_owned(),
                    "head.olog-tmp".to_owned(),
                    "head.caslock".to_owned(),
                    "/absolute".to_owned(),
                    "../head".to_owned(),
                    "a//./b".to_owned(),
                    "x".repeat(64),
                    "x".repeat(300),
                ];
                let mut keys = Vec::new();
                for (index, logical) in identities.iter().enumerate() {
                    let key = pages.data_key(logical)?;
                    assert!(!key.as_str().starts_with('/'));
                    assert!(key.as_str().split('/').all(|part| !part.is_empty()
                        && part != "."
                        && part != ".."
                        && part.len() <= 128));
                    assert!(!key.as_str().contains('.'));
                    assert!(
                        !keys
                            .iter()
                            .any(|existing: &OwnedDataKey| existing.as_str() == key.as_str())
                    );
                    let frame =
                        super::super::data::DataFrame::encode(&pages, logical, &[index as u8])?;
                    pages.upload_data(key.as_str(), frame.bytes()).await?;
                    keys.push(key);
                }
                for (index, key) in keys.iter().enumerate() {
                    assert_eq!(
                        super::super::data::DataFrame::decode(
                            &local.get(key.as_str()).await?.unwrap()
                        )?
                        .payload(),
                        &[index as u8]
                    );
                    assert!(pages.read_head(key.as_str(), 240).await.is_err());
                }
                assert!(local.get("head").await?.is_none());
                assert!(local.get("head.olog-tmp").await?.is_none());
                Ok(())
            })
            .await
            .unwrap();
        resources.close_and_drain().await.unwrap();
    }
}

#[cfg(test)]
mod byte_ledger_tests {
    use super::*;
    #[test]
    fn exact_admission_aliases_and_unwind_release_once() {
        assert!(ByteBudget::new(0).is_err());
        assert!(ByteBudget::new(u64::MAX).is_err());
        let budget = ByteBudget::new(17).unwrap();
        assert!(budget.charge(0).is_err());
        assert!(budget.charge(18).is_err());
        assert!(budget.charge(usize::MAX).is_err());
        assert_eq!(budget.available(), 17);
        let first = budget.charge(16).unwrap();
        assert!(budget.charge(2).is_err());
        let last = budget.charge(1).unwrap();
        assert_eq!(budget.available(), 0);
        let alias = first.clone();
        drop(first);
        assert_eq!(budget.available(), 0);
        drop(last);
        assert_eq!(budget.available(), 1);
        drop(alias);
        assert_eq!(budget.available(), 17);
        let result = std::panic::catch_unwind(|| {
            let charge = budget.charge(17).unwrap();
            let _alias = charge.clone();
            panic!("unwind exact byte ownership");
        });
        assert!(result.is_err());
        assert_eq!(budget.available(), 17);
    }
    #[cfg(target_pointer_width = "64")]
    #[test]
    fn accounting_above_u32_requires_no_payload_allocation() {
        let count = u64::from(u32::MAX) + 128;
        let budget = ByteBudget::new(count).unwrap();
        let retained = budget.charge(usize::try_from(count).unwrap()).unwrap();
        assert_eq!(budget.available(), 0);
        assert!(budget.charge(1).is_err());
        let alias = retained.clone();
        drop(retained);
        assert_eq!(budget.available(), 0);
        drop(alias);
        assert_eq!(budget.available(), usize::try_from(count).unwrap());
    }
    #[test]
    fn simultaneous_contenders_cannot_oversubscribe() {
        use std::sync::{Barrier, atomic::AtomicUsize};
        let budget = ByteBudget::new(100).unwrap();
        let start = Arc::new(Barrier::new(33));
        let attempted = Arc::new(Barrier::new(33));
        let release = Arc::new(Barrier::new(33));
        let successes = Arc::new(AtomicUsize::new(0));
        let mut threads = Vec::new();
        for _ in 0..32 {
            let (budget, start, attempted, release, successes) = (
                budget.clone(),
                start.clone(),
                attempted.clone(),
                release.clone(),
                successes.clone(),
            );
            threads.push(std::thread::spawn(move || {
                start.wait();
                let charge = budget.charge(7).ok();
                if charge.is_some() {
                    successes.fetch_add(1, Ordering::AcqRel);
                }
                attempted.wait();
                release.wait();
                drop(charge);
            }));
        }
        start.wait();
        attempted.wait();
        assert_eq!(successes.load(Ordering::Acquire), 14);
        assert_eq!(budget.available(), 2);
        release.wait();
        for thread in threads {
            thread.join().unwrap();
        }
        assert_eq!(budget.available(), 100);
    }
}