Peernet-Archive/Abstraction
1
0
Fork 0
mirror of https://github.com/PeernetOfficial/Abstraction.git synced 2026-07-21 12:57:50 +01:00
Code Issues Packages Projects Releases Wiki Activity

added example package

Browse Source
This commit is contained in:
Akilan Selvacoumar
2022-12-11 06:18:00 +00:00
parent fc9ed17ecd
commit 73b8a7c2a3
811 changed files with 276135 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vendor/github.com/akrylysov/pogreb/.gitignore generated vendored Normal file
View File

@@ -0,0 +1,2 @@
/test.db
/fs/test

55
vendor/github.com/akrylysov/pogreb/CHANGELOG.md generated vendored Normal file
View File

@@ -0,0 +1,55 @@
# Changelog
## [0.10.1] - 2021-05-01
### Changed
- Improve error reporting.
### Fixed
- Fix compilation for 32-bit OS.
## [0.10.0] - 2021-02-09
### Added
- Memory-mapped file access can now be disabled by setting `Options.FileSystem` to `fs.OS`.
### Changed
- The default file system implementation is changed to `fs.OSMMap`.
## [0.9.2] - 2021-01-01
### Changed
- Write-ahead log doesn't rely on wall-clock time anymore. It prevents potential race conditions during compaction and recovery.
### Fixed
- Fix recovery writing extra delete records.
## [0.9.1] - 2020-04-03
### Changed
- Improve Go 1.14 compatibility (remove "unsafe" usage).
## [0.9.0] - 2020-03-08
### Changed
- Replace the unstructured data file for storing key-value pairs with a write-ahead log.
### Added
- In the event of a crash or a power loss the database is automatically recovered.
- Optional background compaction allows reclaiming disk space occupied by overwritten or deleted keys.
### Fixed
- Fix disk space overhead when storing small keys and values.
## [0.8.3] - 2019-11-03
### Fixed
- Fix slice bounds out of range error mapping files on Windows.
## [0.8.2] - 2019-09-04
### Fixed
- Race condition could lead to data corruption.
## [0.8.1] - 2019-06-30
### Fixed
- Fix panic when accessing closed database.
- Return error opening invalid database.
## [0.8] - 2019-03-30
### Changed
- ~2x write performance improvement on non-Windows.
## [0.7] - 2019-03-23
### Added
- Windows support (@mattn).
### Changed
- Improve freelist performance.

201
vendor/github.com/akrylysov/pogreb/LICENSE generated vendored Normal file
View File

@@ -0,0 +1,201 @@
Apache License
Version 2.0, January 2004
http://www.apache.org/licenses/
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
1. Definitions.
"License" shall mean the terms and conditions for use, reproduction,
and distribution as defined by Sections 1 through 9 of this document.
"Licensor" shall mean the copyright owner or entity authorized by
the copyright owner that is granting the License.
"Legal Entity" shall mean the union of the acting entity and all
other entities that control, are controlled by, or are under common
control with that entity. For the purposes of this definition,
"control" means (i) the power, direct or indirect, to cause the
direction or management of such entity, whether by contract or
otherwise, or (ii) ownership of fifty percent (50%) or more of the
outstanding shares, or (iii) beneficial ownership of such entity.
"You" (or "Your") shall mean an individual or Legal Entity
exercising permissions granted by this License.
"Source" form shall mean the preferred form for making modifications,
including but not limited to software source code, documentation
source, and configuration files.
"Object" form shall mean any form resulting from mechanical
transformation or translation of a Source form, including but
not limited to compiled object code, generated documentation,
and conversions to other media types.
"Work" shall mean the work of authorship, whether in Source or
Object form, made available under the License, as indicated by a
copyright notice that is included in or attached to the work
(an example is provided in the Appendix below).
"Derivative Works" shall mean any work, whether in Source or Object
form, that is based on (or derived from) the Work and for which the
editorial revisions, annotations, elaborations, or other modifications
represent, as a whole, an original work of authorship. For the purposes
of this License, Derivative Works shall not include works that remain
separable from, or merely link (or bind by name) to the interfaces of,
the Work and Derivative Works thereof.
"Contribution" shall mean any work of authorship, including
the original version of the Work and any modifications or additions
to that Work or Derivative Works thereof, that is intentionally
submitted to Licensor for inclusion in the Work by the copyright owner
or by an individual or Legal Entity authorized to submit on behalf of
the copyright owner. For the purposes of this definition, "submitted"
means any form of electronic, verbal, or written communication sent
to the Licensor or its representatives, including but not limited to
communication on electronic mailing lists, source code control systems,
and issue tracking systems that are managed by, or on behalf of, the
Licensor for the purpose of discussing and improving the Work, but
excluding communication that is conspicuously marked or otherwise
designated in writing by the copyright owner as "Not a Contribution."
"Contributor" shall mean Licensor and any individual or Legal Entity
on behalf of whom a Contribution has been received by Licensor and
subsequently incorporated within the Work.
2. Grant of Copyright License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
copyright license to reproduce, prepare Derivative Works of,
publicly display, publicly perform, sublicense, and distribute the
Work and such Derivative Works in Source or Object form.
3. Grant of Patent License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
(except as stated in this section) patent license to make, have made,
use, offer to sell, sell, import, and otherwise transfer the Work,
where such license applies only to those patent claims licensable
by such Contributor that are necessarily infringed by their
Contribution(s) alone or by combination of their Contribution(s)
with the Work to which such Contribution(s) was submitted. If You
institute patent litigation against any entity (including a
cross-claim or counterclaim in a lawsuit) alleging that the Work
or a Contribution incorporated within the Work constitutes direct
or contributory patent infringement, then any patent licenses
granted to You under this License for that Work shall terminate
as of the date such litigation is filed.
4. Redistribution. You may reproduce and distribute copies of the
Work or Derivative Works thereof in any medium, with or without
modifications, and in Source or Object form, provided that You
meet the following conditions:
(a) You must give any other recipients of the Work or
Derivative Works a copy of this License; and
(b) You must cause any modified files to carry prominent notices
stating that You changed the files; and
(c) You must retain, in the Source form of any Derivative Works
that You distribute, all copyright, patent, trademark, and
attribution notices from the Source form of the Work,
excluding those notices that do not pertain to any part of
the Derivative Works; and
(d) If the Work includes a "NOTICE" text file as part of its
distribution, then any Derivative Works that You distribute must
include a readable copy of the attribution notices contained
within such NOTICE file, excluding those notices that do not
pertain to any part of the Derivative Works, in at least one
of the following places: within a NOTICE text file distributed
as part of the Derivative Works; within the Source form or
documentation, if provided along with the Derivative Works; or,
within a display generated by the Derivative Works, if and
wherever such third-party notices normally appear. The contents
of the NOTICE file are for informational purposes only and
do not modify the License. You may add Your own attribution
notices within Derivative Works that You distribute, alongside
or as an addendum to the NOTICE text from the Work, provided
that such additional attribution notices cannot be construed
as modifying the License.
You may add Your own copyright statement to Your modifications and
may provide additional or different license terms and conditions
for use, reproduction, or distribution of Your modifications, or
for any such Derivative Works as a whole, provided Your use,
reproduction, and distribution of the Work otherwise complies with
the conditions stated in this License.
5. Submission of Contributions. Unless You explicitly state otherwise,
any Contribution intentionally submitted for inclusion in the Work
by You to the Licensor shall be under the terms and conditions of
this License, without any additional terms or conditions.
Notwithstanding the above, nothing herein shall supersede or modify
the terms of any separate license agreement you may have executed
with Licensor regarding such Contributions.
6. Trademarks. This License does not grant permission to use the trade
names, trademarks, service marks, or product names of the Licensor,
except as required for reasonable and customary use in describing the
origin of the Work and reproducing the content of the NOTICE file.
7. Disclaimer of Warranty. Unless required by applicable law or
agreed to in writing, Licensor provides the Work (and each
Contributor provides its Contributions) on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
implied, including, without limitation, any warranties or conditions
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
PARTICULAR PURPOSE. You are solely responsible for determining the
appropriateness of using or redistributing the Work and assume any
risks associated with Your exercise of permissions under this License.
8. Limitation of Liability. In no event and under no legal theory,
whether in tort (including negligence), contract, or otherwise,
unless required by applicable law (such as deliberate and grossly
negligent acts) or agreed to in writing, shall any Contributor be
liable to You for damages, including any direct, indirect, special,
incidental, or consequential damages of any character arising as a
result of this License or out of the use or inability to use the
Work (including but not limited to damages for loss of goodwill,
work stoppage, computer failure or malfunction, or any and all
other commercial damages or losses), even if such Contributor
has been advised of the possibility of such damages.
9. Accepting Warranty or Additional Liability. While redistributing
the Work or Derivative Works thereof, You may choose to offer,
and charge a fee for, acceptance of support, warranty, indemnity,
or other liability obligations and/or rights consistent with this
License. However, in accepting such obligations, You may act only
on Your own behalf and on Your sole responsibility, not on behalf
of any other Contributor, and only if You agree to indemnify,
defend, and hold each Contributor harmless for any liability
incurred by, or claims asserted against, such Contributor by reason
of your accepting any such warranty or additional liability.
END OF TERMS AND CONDITIONS
APPENDIX: How to apply the Apache License to your work.
To apply the Apache License to your work, attach the following
boilerplate notice, with the fields enclosed by brackets "[]"
replaced with your own identifying information. (Don't include
the brackets!) The text should be enclosed in the appropriate
comment syntax for the file format. We also recommend that a
file or class name and description of purpose be included on the
same "printed page" as the copyright notice for easier
identification within third-party archives.
Copyright [yyyy] [name of copyright owner]
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.

102
vendor/github.com/akrylysov/pogreb/README.md generated vendored Normal file
View File

@@ -0,0 +1,102 @@
<p align="center"><img src="https://akrylysov.github.io/pogreb/logo.svg" width="300"></p>
# Pogreb
[![Docs](https://godoc.org/github.com/akrylysov/pogreb?status.svg)](https://pkg.go.dev/github.com/akrylysov/pogreb)
[![Build Status](https://github.com/akrylysov/pogreb/actions/workflows/test.yaml/badge.svg?branch=master)](https://github.com/akrylysov/pogreb/actions)
[![Go Report Card](https://goreportcard.com/badge/github.com/akrylysov/pogreb)](https://goreportcard.com/report/github.com/akrylysov/pogreb)
[![Codecov](https://codecov.io/gh/akrylysov/pogreb/branch/master/graph/badge.svg)](https://codecov.io/gh/akrylysov/pogreb)
Pogreb is an embedded key-value store for read-heavy workloads written in Go.
## Key characteristics
- 100% Go.
- Optimized for fast random lookups and infrequent bulk inserts.
- Can store larger-than-memory data sets.
- Low memory usage.
- All DB methods are safe for concurrent use by multiple goroutines.
## Installation
```sh
$ go get -u github.com/akrylysov/pogreb
```
## Usage
### Opening a database
To open or create a new database, use the `pogreb.Open()` function:
```go
package main
import (
"log"
"github.com/akrylysov/pogreb"
)
func main() {
db, err := pogreb.Open("pogreb.test", nil)
if err != nil {
log.Fatal(err)
return
}
defer db.Close()
}
```
### Writing to a database
Use the `DB.Put()` function to insert a new key-value pair:
```go
err := db.Put([]byte("testKey"), []byte("testValue"))
if err != nil {
log.Fatal(err)
}
```
### Reading from a database
To retrieve the inserted value, use the `DB.Get()` function:
```go
val, err := db.Get([]byte("testKey"))
if err != nil {
log.Fatal(err)
}
log.Printf("%s", val)
```
### Iterating over items
To iterate over items, use `ItemIterator` returned by `DB.Items()`:
```go
it := db.Items()
for {
key, val, err := it.Next()
if err == pogreb.ErrIterationDone {
break
}
if err != nil {
log.Fatal(err)
}
log.Printf("%s %s", key, val)
}
```
## Performance
The benchmarking code can be found in the [pogreb-bench](https://github.com/akrylysov/pogreb-bench) repository.
Results of read performance benchmark of pogreb, goleveldb, bolt and badgerdb
on DigitalOcean 8 CPUs / 16 GB RAM / 160 GB SSD + Ubuntu 16.04.3 (higher is better):
<p align="center"><img src="https://akrylysov.github.io/pogreb/read-bench.png" width="609"></p>
## Internals
[Design document](/docs/design.md).

15
vendor/github.com/akrylysov/pogreb/appveyor.yml generated vendored Normal file
View File

@@ -0,0 +1,15 @@
version: "{build}"
clone_folder: c:\gopath\src\github.com\akrylysov/pogreb
environment:
GOPATH: c:\gopath
install:
- echo %PATH%
- echo %GOPATH%
- go version
- go env
build_script:
- go test -v ./...

126
vendor/github.com/akrylysov/pogreb/bucket.go generated vendored Normal file
View File

@@ -0,0 +1,126 @@
package pogreb
import (
"encoding/binary"
)
const (
bucketSize = 512
slotsPerBucket = 31 // Maximum number of slots possible to fit in a 512-byte bucket.
)
// slot corresponds to a single item in the hash table.
type slot struct {
hash uint32
segmentID uint16
keySize uint16
valueSize uint32
offset uint32 // Offset of the record in a segment.
}
func (sl slot) kvSize() uint32 {
return uint32(sl.keySize) + sl.valueSize
}
// bucket is an array of slots.
type bucket struct {
slots [slotsPerBucket]slot
next int64 // Offset of overflow bucket.
}
// bucketHandle is a bucket, plus its offset and the file it's written to.
type bucketHandle struct {
bucket
file *file
offset int64
}
func (b bucket) MarshalBinary() ([]byte, error) {
buf := make([]byte, bucketSize)
data := buf
for i := 0; i < slotsPerBucket; i++ {
sl := b.slots[i]
binary.LittleEndian.PutUint32(buf[:4], sl.hash)
binary.LittleEndian.PutUint16(buf[4:6], sl.segmentID)
binary.LittleEndian.PutUint16(buf[6:8], sl.keySize)
binary.LittleEndian.PutUint32(buf[8:12], sl.valueSize)
binary.LittleEndian.PutUint32(buf[12:16], sl.offset)
buf = buf[16:]
}
binary.LittleEndian.PutUint64(buf[:8], uint64(b.next))
return data, nil
}
func (b *bucket) UnmarshalBinary(data []byte) error {
for i := 0; i < slotsPerBucket; i++ {
_ = data[16] // bounds check hint to compiler; see golang.org/issue/14808
b.slots[i].hash = binary.LittleEndian.Uint32(data[:4])
b.slots[i].segmentID = binary.LittleEndian.Uint16(data[4:6])
b.slots[i].keySize = binary.LittleEndian.Uint16(data[6:8])
b.slots[i].valueSize = binary.LittleEndian.Uint32(data[8:12])
b.slots[i].offset = binary.LittleEndian.Uint32(data[12:16])
data = data[16:]
}
b.next = int64(binary.LittleEndian.Uint64(data[:8]))
return nil
}
func (b *bucket) del(slotIdx int) {
i := slotIdx
// Shift slots.
for ; i < slotsPerBucket-1; i++ {
b.slots[i] = b.slots[i+1]
}
b.slots[i] = slot{}
}
func (b *bucketHandle) read() error {
buf, err := b.file.Slice(b.offset, b.offset+int64(bucketSize))
if err != nil {
return err
}
return b.UnmarshalBinary(buf)
}
func (b *bucketHandle) write() error {
buf, err := b.MarshalBinary()
if err != nil {
return err
}
_, err = b.file.WriteAt(buf, b.offset)
return err
}
// slotWriter inserts and writes slots into a bucket.
type slotWriter struct {
bucket *bucketHandle
slotIdx int
prevBuckets []*bucketHandle
}
func (sw *slotWriter) insert(sl slot, idx *index) error {
if sw.slotIdx == slotsPerBucket {
// Bucket is full, create a new overflow bucket.
nextBucket, err := idx.createOverflowBucket()
if err != nil {
return err
}
sw.bucket.next = nextBucket.offset
sw.prevBuckets = append(sw.prevBuckets, sw.bucket)
sw.bucket = nextBucket
sw.slotIdx = 0
}
sw.bucket.slots[sw.slotIdx] = sl
sw.slotIdx++
return nil
}
func (sw *slotWriter) write() error {
// Write previous buckets first.
for i := len(sw.prevBuckets) - 1; i >= 0; i-- {
if err := sw.prevBuckets[i].write(); err != nil {
return err
}
}
return sw.bucket.write()
}

160
vendor/github.com/akrylysov/pogreb/compaction.go generated vendored Normal file
View File

@@ -0,0 +1,160 @@
package pogreb
import (
"sync/atomic"
"github.com/akrylysov/pogreb/internal/errors"
)
// promoteRecord writes the record to the current segment if the index still points to the record.
// Otherwise it discards the record.
func (db *DB) promoteRecord(rec record) (bool, error) {
hash := db.hash(rec.key)
it := db.index.newBucketIterator(db.index.bucketIndex(hash))
for {
b, err := it.next()
if err == ErrIterationDone {
// Exhausted all buckets and the slot wasn't found.
// The key was deleted or overwritten. The record is safe to discard.
return true, nil
}
if err != nil {
return false, err
}
for i := 0; i < slotsPerBucket; i++ {
sl := b.slots[i]
// No more slots in the bucket.
if sl.offset == 0 {
break
}
// Slot points to a different record.
if hash != sl.hash || rec.offset != sl.offset || rec.segmentID != sl.segmentID {
continue
}
// The record is in the index, write it to the current segment.
segmentID, offset, err := db.datalog.writeRecord(rec.data, rec.rtype) // TODO: batch writes
if err != nil {
return false, err
}
// Update index.
b.slots[i].segmentID = segmentID
b.slots[i].offset = offset
return false, b.write()
}
}
}
// CompactionResult holds the compaction result.
type CompactionResult struct {
CompactedSegments int
ReclaimedRecords int
ReclaimedBytes int
}
func (db *DB) compact(sourceSeg *segment) (CompactionResult, error) {
cr := CompactionResult{}
db.mu.Lock()
sourceSeg.meta.Full = true // Prevent writes to the compacted file.
db.mu.Unlock()
it, err := newSegmentIterator(sourceSeg)
if err != nil {
return cr, err
}
// Copy records from sourceSeg to the current segment.
for {
err := func() error {
db.mu.Lock()
defer db.mu.Unlock()
rec, err := it.next()
if err != nil {
return err
}
if rec.rtype == recordTypeDelete {
cr.ReclaimedRecords++
cr.ReclaimedBytes += len(rec.data)
return nil
}
reclaimed, err := db.promoteRecord(rec)
if reclaimed {
cr.ReclaimedRecords++
cr.ReclaimedBytes += len(rec.data)
}
return err
}()
if err == ErrIterationDone {
break
}
if err != nil {
return cr, err
}
}
db.mu.Lock()
defer db.mu.Unlock()
err = db.datalog.removeSegment(sourceSeg)
return cr, err
}
// pickForCompaction returns segments eligible for compaction.
func (db *DB) pickForCompaction() []*segment {
segments := db.datalog.segmentsBySequenceID()
var picked []*segment
for i := len(segments) - 1; i >= 0; i-- {
seg := segments[i]
if uint32(seg.size) < db.opts.compactionMinSegmentSize {
continue
}
fragmentation := float32(seg.meta.DeletedBytes) / float32(seg.size)
if fragmentation < db.opts.compactionMinFragmentation {
continue
}
if seg.meta.DeleteRecords > 0 {
// Delete records can be discarded only when older segments contain no put records
// for the corresponding keys.
// All segments older than the segment eligible for compaction have to be compacted.
return append(segments[:i+1], picked...)
}
picked = append([]*segment{seg}, picked...)
}
return picked
}
// Compact compacts the DB. Deleted and overwritten items are discarded.
// Returns an error if compaction is already in progress.
func (db *DB) Compact() (CompactionResult, error) {
cr := CompactionResult{}
// Run only a single compaction at a time.
if !atomic.CompareAndSwapInt32(&db.compactionRunning, 0, 1) {
return cr, errBusy
}
defer func() {
atomic.StoreInt32(&db.compactionRunning, 0)
}()
db.mu.RLock()
segments := db.pickForCompaction()
db.mu.RUnlock()
for _, seg := range segments {
segcr, err := db.compact(seg)
if err != nil {
return cr, errors.Wrapf(err, "compacting segment %s", seg.name)
}
cr.CompactedSegments++
cr.ReclaimedRecords += segcr.ReclaimedRecords
cr.ReclaimedBytes += segcr.ReclaimedBytes
}
return cr, nil
}

261
vendor/github.com/akrylysov/pogreb/datalog.go generated vendored Normal file
View File

@@ -0,0 +1,261 @@
package pogreb
import (
"fmt"
"math"
"os"
"path/filepath"
"sort"
"strconv"
"strings"
"github.com/akrylysov/pogreb/internal/errors"
)
const (
maxSegments = math.MaxInt16
)
// datalog is a write-ahead log.
type datalog struct {
opts *Options
curSeg *segment
segments [maxSegments]*segment
maxSequenceID uint64
}
func openDatalog(opts *Options) (*datalog, error) {
files, err := opts.FileSystem.ReadDir(".")
if err != nil {
return nil, err
}
dl := &datalog{
opts: opts,
}
// Open existing segments.
for _, file := range files {
name := file.Name()
ext := filepath.Ext(name)
if ext != segmentExt {
continue
}
id, seqID, err := parseSegmentName(name)
if err != nil {
return nil, err
}
seg, err := dl.openSegment(name, id, seqID)
if err != nil {
return nil, errors.Wrapf(err, "opening segment %s", name)
}
if seg.sequenceID > dl.maxSequenceID {
dl.maxSequenceID = seg.sequenceID
}
dl.segments[seg.id] = seg
}
if err := dl.swapSegment(); err != nil {
return nil, err
}
return dl, nil
}
func parseSegmentName(name string) (uint16, uint64, error) {
parts := strings.SplitN(strings.TrimSuffix(name, segmentExt), "-", 2)
id, err := strconv.ParseUint(parts[0], 10, 16)
if err != nil {
return 0, 0, err
}
var seqID uint64
if len(parts) == 2 {
seqID, err = strconv.ParseUint(parts[1], 10, 64)
if err != nil {
return 0, 0, err
}
}
return uint16(id), seqID, nil
}
func (dl *datalog) openSegment(name string, id uint16, seqID uint64) (*segment, error) {
f, err := openFile(dl.opts.FileSystem, name, false)
if err != nil {
return nil, err
}
meta := &segmentMeta{}
if !f.empty() {
metaName := name + metaExt
if err := readGobFile(dl.opts.FileSystem, metaName, &meta); err != nil {
logger.Printf("error reading segment meta %d: %v", id, err)
// TODO: rebuild meta?
}
}
seg := &segment{
file: f,
id: id,
sequenceID: seqID,
name: name,
meta: meta,
}
return seg, nil
}
func (dl *datalog) nextWritableSegmentID() (uint16, uint64, error) {
for id, seg := range dl.segments {
// Pick empty segment.
if seg == nil {
dl.maxSequenceID++
return uint16(id), dl.maxSequenceID, nil
}
}
return 0, 0, fmt.Errorf("number of segments exceeds %d", maxSegments)
}
func (dl *datalog) swapSegment() error {
// Pick unfilled segment.
for _, seg := range dl.segments {
if seg != nil && !seg.meta.Full {
dl.curSeg = seg
return nil
}
}
// Create new segment.
id, seqID, err := dl.nextWritableSegmentID()
if err != nil {
return err
}
name := segmentName(id, seqID)
seg, err := dl.openSegment(name, id, seqID)
if err != nil {
return err
}
dl.segments[id] = seg
dl.curSeg = seg
return nil
}
func (dl *datalog) removeSegment(seg *segment) error {
dl.segments[seg.id] = nil
if err := seg.Close(); err != nil {
return err
}
// Remove segment meta from FS.
metaName := seg.name + segmentExt
if err := dl.opts.FileSystem.Remove(metaName); err != nil && !os.IsNotExist(err) {
return err
}
// Remove segment from FS.
if err := dl.opts.FileSystem.Remove(seg.name); err != nil {
return err
}
return nil
}
func (dl *datalog) readKeyValue(sl slot) ([]byte, []byte, error) {
off := int64(sl.offset) + 6 // Skip key size and value size.
seg := dl.segments[sl.segmentID]
keyValue, err := seg.Slice(off, off+int64(sl.kvSize()))
if err != nil {
return nil, nil, err
}
return keyValue[:sl.keySize], keyValue[sl.keySize:], nil
}
func (dl *datalog) readKey(sl slot) ([]byte, error) {
off := int64(sl.offset) + 6
seg := dl.segments[sl.segmentID]
return seg.Slice(off, off+int64(sl.keySize))
}
// trackDel updates segment's metadata for deleted or overwritten items.
func (dl *datalog) trackDel(sl slot) {
meta := dl.segments[sl.segmentID].meta
meta.DeletedKeys++
meta.DeletedBytes += encodedRecordSize(sl.kvSize())
}
func (dl *datalog) del(key []byte) error {
rec := encodeDeleteRecord(key)
_, _, err := dl.writeRecord(rec, recordTypeDelete)
if err != nil {
return err
}
// Compaction removes delete records, increment DeletedBytes.
dl.curSeg.meta.DeletedBytes += uint32(len(rec))
return nil
}
func (dl *datalog) writeRecord(data []byte, rt recordType) (uint16, uint32, error) {
if dl.curSeg.meta.Full || dl.curSeg.size+int64(len(data)) > int64(dl.opts.maxSegmentSize) {
// Current segment is full, create a new one.
dl.curSeg.meta.Full = true
if err := dl.swapSegment(); err != nil {
return 0, 0, err
}
}
off, err := dl.curSeg.append(data)
if err != nil {
return 0, 0, err
}
switch rt {
case recordTypePut:
dl.curSeg.meta.PutRecords++
case recordTypeDelete:
dl.curSeg.meta.DeleteRecords++
}
return dl.curSeg.id, uint32(off), nil
}
func (dl *datalog) put(key []byte, value []byte) (uint16, uint32, error) {
return dl.writeRecord(encodePutRecord(key, value), recordTypePut)
}
func (dl *datalog) sync() error {
return dl.curSeg.Sync()
}
func (dl *datalog) close() error {
for _, seg := range dl.segments {
if seg == nil {
continue
}
if err := seg.Close(); err != nil {
return err
}
metaName := seg.name + metaExt
if err := writeGobFile(dl.opts.FileSystem, metaName, seg.meta); err != nil {
return err
}
}
return nil
}
// segmentsBySequenceID returns segments ordered from oldest to newest.
func (dl *datalog) segmentsBySequenceID() []*segment {
var segments []*segment
for _, seg := range dl.segments {
if seg == nil {
continue
}
segments = append(segments, seg)
}
sort.SliceStable(segments, func(i, j int) bool {
return segments[i].sequenceID < segments[j].sequenceID
})
return segments
}

403
vendor/github.com/akrylysov/pogreb/db.go generated vendored Normal file
View File

@@ -0,0 +1,403 @@
package pogreb
import (
"bytes"
"context"
"math"
"os"
"sync"
"time"
"github.com/akrylysov/pogreb/fs"
"github.com/akrylysov/pogreb/internal/errors"
"github.com/akrylysov/pogreb/internal/hash"
)
const (
// MaxKeyLength is the maximum size of a key in bytes.
MaxKeyLength = math.MaxUint16
// MaxValueLength is the maximum size of a value in bytes.
MaxValueLength = 512 << 20 // 512 MiB
// MaxKeys is the maximum numbers of keys in the DB.
MaxKeys = math.MaxUint32
metaExt = ".pmt"
dbMetaName = "db" + metaExt
)
// DB represents the key-value storage.
// All DB methods are safe for concurrent use by multiple goroutines.
type DB struct {
mu sync.RWMutex // Allows multiple database readers or a single writer.
opts *Options
index *index
datalog *datalog
lock fs.LockFile // Prevents opening multiple instances of the same database.
hashSeed uint32
metrics *Metrics
syncWrites bool
cancelBgWorker context.CancelFunc
closeWg sync.WaitGroup
compactionRunning int32 // Prevents running compactions concurrently.
}
type dbMeta struct {
HashSeed uint32
}
// Open opens or creates a new DB.
// The DB must be closed after use, by calling Close method.
func Open(path string, opts *Options) (*DB, error) {
opts = opts.copyWithDefaults(path)
if err := os.MkdirAll(path, 0755); err != nil {
return nil, err
}
// Try to acquire a file lock.
lock, acquiredExistingLock, err := createLockFile(opts)
if err != nil {
if err == os.ErrExist {
err = errLocked
}
return nil, errors.Wrap(err, "creating lock file")
}
clean := lock.Unlock
defer func() {
if clean != nil {
_ = clean()
}
}()
if acquiredExistingLock {
// Lock file already existed, but the process managed to acquire it.
// It means the database wasn't closed properly.
// Start recovery process.
if err := backupNonsegmentFiles(opts.FileSystem); err != nil {
return nil, err
}
}
index, err := openIndex(opts)
if err != nil {
return nil, errors.Wrap(err, "opening index")
}
datalog, err := openDatalog(opts)
if err != nil {
return nil, errors.Wrap(err, "opening datalog")
}
db := &DB{
opts: opts,
index: index,
datalog: datalog,
lock: lock,
metrics: &Metrics{},
syncWrites: opts.BackgroundSyncInterval == -1,
}
if index.count() == 0 {
// The index is empty, make a new hash seed.
seed, err := hash.RandSeed()
if err != nil {
return nil, err
}
db.hashSeed = seed
} else {
if err := db.readMeta(); err != nil {
return nil, errors.Wrap(err, "reading db meta")
}
}
if acquiredExistingLock {
if err := db.recover(); err != nil {
return nil, errors.Wrap(err, "recovering")
}
}
if db.opts.BackgroundSyncInterval > 0 || db.opts.BackgroundCompactionInterval > 0 {
db.startBackgroundWorker()
}
clean = nil
return db, nil
}
func cloneBytes(src []byte) []byte {
dst := make([]byte, len(src))
copy(dst, src)
return dst
}
func (db *DB) writeMeta() error {
m := dbMeta{
HashSeed: db.hashSeed,
}
return writeGobFile(db.opts.FileSystem, dbMetaName, m)
}
func (db *DB) readMeta() error {
m := dbMeta{}
if err := readGobFile(db.opts.FileSystem, dbMetaName, &m); err != nil {
return err
}
db.hashSeed = m.HashSeed
return nil
}
func (db *DB) hash(data []byte) uint32 {
return hash.Sum32WithSeed(data, db.hashSeed)
}
// newNullableTicker is a wrapper around time.NewTicker that allows creating a nil ticker.
// A nil ticker never ticks.
func newNullableTicker(d time.Duration) (<-chan time.Time, func()) {
if d > 0 {
t := time.NewTicker(d)
return t.C, t.Stop
}
return nil, func() {}
}
func (db *DB) startBackgroundWorker() {
ctx, cancel := context.WithCancel(context.Background())
db.cancelBgWorker = cancel
db.closeWg.Add(1)
go func() {
defer db.closeWg.Done()
syncC, syncStop := newNullableTicker(db.opts.BackgroundSyncInterval)
defer syncStop()
compactC, compactStop := newNullableTicker(db.opts.BackgroundCompactionInterval)
defer compactStop()
for {
select {
case <-ctx.Done():
return
case <-syncC:
if err := db.Sync(); err != nil {
logger.Printf("error synchronizing database: %v", err)
}
case <-compactC:
if cr, err := db.Compact(); err != nil {
logger.Printf("error compacting database: %v", err)
} else if cr.CompactedSegments > 0 {
logger.Printf("compacted database: %+v", cr)
}
}
}
}()
}
// Get returns the value for the given key stored in the DB or nil if the key doesn't exist.
func (db *DB) Get(key []byte) ([]byte, error) {
h := db.hash(key)
db.metrics.Gets.Add(1)
db.mu.RLock()
defer db.mu.RUnlock()
var retValue []byte
err := db.index.get(h, func(sl slot) (bool, error) {
if uint16(len(key)) != sl.keySize {
return false, nil
}
slKey, value, err := db.datalog.readKeyValue(sl)
if err != nil {
return true, err
}
if bytes.Equal(key, slKey) {
retValue = cloneBytes(value)
return true, nil
}
db.metrics.HashCollisions.Add(1)
return false, nil
})
if err != nil {
return nil, err
}
return retValue, nil
}
// Has returns true if the DB contains the given key.
func (db *DB) Has(key []byte) (bool, error) {
h := db.hash(key)
found := false
db.mu.RLock()
defer db.mu.RUnlock()
err := db.index.get(h, func(sl slot) (bool, error) {
if uint16(len(key)) != sl.keySize {
return false, nil
}
slKey, err := db.datalog.readKey(sl)
if err != nil {
return true, err
}
if bytes.Equal(key, slKey) {
found = true
return true, nil
}
return false, nil
})
if err != nil {
return false, err
}
return found, nil
}
func (db *DB) put(sl slot, key []byte) error {
return db.index.put(sl, func(cursl slot) (bool, error) {
if uint16(len(key)) != cursl.keySize {
return false, nil
}
slKey, err := db.datalog.readKey(cursl)
if err != nil {
return true, err
}
if bytes.Equal(key, slKey) {
db.datalog.trackDel(cursl) // Overwriting existing key.
return true, nil
}
return false, nil
})
}
// Put sets the value for the given key. It updates the value for the existing key.
func (db *DB) Put(key []byte, value []byte) error {
if len(key) > MaxKeyLength {
return errKeyTooLarge
}
if len(value) > MaxValueLength {
return errValueTooLarge
}
h := db.hash(key)
db.metrics.Puts.Add(1)
db.mu.Lock()
defer db.mu.Unlock()
segID, offset, err := db.datalog.put(key, value)
if err != nil {
return err
}
sl := slot{
hash: h,
segmentID: segID,
keySize: uint16(len(key)),
valueSize: uint32(len(value)),
offset: offset,
}
if err := db.put(sl, key); err != nil {
return err
}
if db.syncWrites {
return db.sync()
}
return nil
}
func (db *DB) del(h uint32, key []byte, writeWAL bool) error {
err := db.index.delete(h, func(sl slot) (b bool, e error) {
if uint16(len(key)) != sl.keySize {
return false, nil
}
slKey, err := db.datalog.readKey(sl)
if err != nil {
return true, err
}
if bytes.Equal(key, slKey) {
db.datalog.trackDel(sl)
var err error
if writeWAL {
err = db.datalog.del(key)
}
return true, err
}
return false, nil
})
return err
}
// Delete deletes the given key from the DB.
func (db *DB) Delete(key []byte) error {
h := db.hash(key)
db.metrics.Dels.Add(1)
db.mu.Lock()
defer db.mu.Unlock()
if err := db.del(h, key, true); err != nil {
return err
}
if db.syncWrites {
return db.sync()
}
return nil
}
// Close closes the DB.
func (db *DB) Close() error {
if db.cancelBgWorker != nil {
db.cancelBgWorker()
}
db.closeWg.Wait()
db.mu.Lock()
defer db.mu.Unlock()
if err := db.writeMeta(); err != nil {
return err
}
if err := db.datalog.close(); err != nil {
return err
}
if err := db.index.close(); err != nil {
return err
}
if err := db.lock.Unlock(); err != nil {
return err
}
return nil
}
func (db *DB) sync() error {
return db.datalog.sync()
}
// Items returns a new ItemIterator.
func (db *DB) Items() *ItemIterator {
return &ItemIterator{db: db}
}
// Sync commits the contents of the database to the backing FileSystem.
func (db *DB) Sync() error {
db.mu.Lock()
defer db.mu.Unlock()
return db.sync()
}
// Count returns the number of keys in the DB.
func (db *DB) Count() uint32 {
db.mu.RLock()
defer db.mu.RUnlock()
return db.index.count()
}
// Metrics returns the DB metrics.
func (db *DB) Metrics() *Metrics {
return db.metrics
}
// FileSize returns the total size of the disk storage used by the DB.
func (db *DB) FileSize() (int64, error) {
var size int64
files, err := db.opts.FileSystem.ReadDir(".")
if err != nil {
return 0, err
}
for _, file := range files {
size += file.Size()
}
return size, nil
}

4
vendor/github.com/akrylysov/pogreb/doc.go generated vendored Normal file
View File

@@ -0,0 +1,4 @@
/*
Package pogreb implements an embedded key-value store for read-heavy workloads.
*/
package pogreb

14
vendor/github.com/akrylysov/pogreb/errors.go generated vendored Normal file
View File

@@ -0,0 +1,14 @@
package pogreb
import (
"github.com/akrylysov/pogreb/internal/errors"
)
var (
errKeyTooLarge = errors.New("key is too large")
errValueTooLarge = errors.New("value is too large")
errFull = errors.New("database is full")
errCorrupted = errors.New("database is corrupted")
errLocked = errors.New("database is locked")
errBusy = errors.New("database is busy")
)

97
vendor/github.com/akrylysov/pogreb/file.go generated vendored Normal file
View File

@@ -0,0 +1,97 @@
package pogreb
import (
"io"
"os"
"github.com/akrylysov/pogreb/fs"
)
// file is a database file.
// When stored in a file system, the file starts with a header.
type file struct {
fs.File
size int64
}
func openFile(fsyst fs.FileSystem, name string, truncate bool) (*file, error) {
flag := os.O_CREATE | os.O_RDWR
if truncate {
flag |= os.O_TRUNC
}
fi, err := fsyst.OpenFile(name, flag, os.FileMode(0640))
f := &file{}
if err != nil {
return f, err
}
clean := fi.Close
defer func() {
if clean != nil {
_ = clean()
}
}()
f.File = fi
stat, err := fi.Stat()
if err != nil {
return f, err
}
f.size = stat.Size()
if f.size == 0 {
// It's a new file - write header.
if err := f.writeHeader(); err != nil {
return nil, err
}
} else {
if err := f.readHeader(); err != nil {
return nil, err
}
}
if _, err := f.Seek(int64(headerSize), io.SeekStart); err != nil {
return nil, err
}
clean = nil
return f, nil
}
func (f *file) writeHeader() error {
h := newHeader()
data, err := h.MarshalBinary()
if err != nil {
return err
}
if _, err = f.append(data); err != nil {
return err
}
return nil
}
func (f *file) readHeader() error {
h := &header{}
buf := make([]byte, headerSize)
if _, err := io.ReadFull(f, buf); err != nil {
return err
}
return h.UnmarshalBinary(buf)
}
func (f *file) empty() bool {
return f.size == int64(headerSize)
}
func (f *file) extend(size uint32) (int64, error) {
off := f.size
if err := f.Truncate(off + int64(size)); err != nil {
return 0, err
}
f.size += int64(size)
return off, nil
}
func (f *file) append(data []byte) (int64, error) {
off := f.size
if _, err := f.WriteAt(data, off); err != nil {
return 0, err
}
f.size += int64(len(data))
return off, nil
}

63
vendor/github.com/akrylysov/pogreb/fs/fs.go generated vendored Normal file
View File

@@ -0,0 +1,63 @@
/*
Package fs provides a file system interface.
*/
package fs
import (
"errors"
"io"
"os"
)
var (
errAppendModeNotSupported = errors.New("append mode is not supported")
)
// File is the interface compatible with os.File.
type File interface {
io.Closer
io.Reader
io.ReaderAt
io.Seeker
io.Writer
io.WriterAt
// Stat returns os.FileInfo describing the file.
Stat() (os.FileInfo, error)
// Sync commits the current contents of the file.
Sync() error
// Truncate changes the size of the file.
Truncate(size int64) error
// Slice reads and returns the contents of file from offset start to offset end.
Slice(start int64, end int64) ([]byte, error)
}
// LockFile represents a lock file.
type LockFile interface {
// Unlock and removes the lock file.
Unlock() error
}
// FileSystem represents a file system.
type FileSystem interface {
// OpenFile opens the file with specified flag.
OpenFile(name string, flag int, perm os.FileMode) (File, error)
// Stat returns os.FileInfo describing the file.
Stat(name string) (os.FileInfo, error)
// Remove removes the file.
Remove(name string) error
// Rename renames oldpath to newpath.
Rename(oldpath, newpath string) error
// ReadDir reads the directory and returns a list of directory entries.
ReadDir(name string) ([]os.FileInfo, error)
// CreateLockFile creates a lock file.
CreateLockFile(name string, perm os.FileMode) (LockFile, bool, error)
}

234
vendor/github.com/akrylysov/pogreb/fs/mem.go generated vendored Normal file
View File

@@ -0,0 +1,234 @@
package fs
import (
"io"
"os"
"path/filepath"
"time"
)
type memFS struct {
files map[string]*memFile
}
// Mem is a file system backed by memory.
var Mem FileSystem = &memFS{files: map[string]*memFile{}}
func (fs *memFS) OpenFile(name string, flag int, perm os.FileMode) (File, error) {
if flag&os.O_APPEND != 0 {
// memFS doesn't support opening files in append-only mode.
// The database doesn't currently use O_APPEND.
return nil, errAppendModeNotSupported
}
f := fs.files[name]
if f == nil || (flag&os.O_TRUNC) != 0 {
f = &memFile{
name: name,
perm: perm, // Perm is saved to return it in Mode, but don't do anything else with it yet.
}
fs.files[name] = f
} else if !f.closed {
return nil, os.ErrExist
} else {
f.offset = 0
f.closed = false
}
return f, nil
}
func (fs *memFS) CreateLockFile(name string, perm os.FileMode) (LockFile, bool, error) {
_, exists := fs.files[name]
_, err := fs.OpenFile(name, 0, perm)
if err != nil {
return nil, false, err
}
return fs.files[name], exists, nil
}
func (fs *memFS) Stat(name string) (os.FileInfo, error) {
if f, ok := fs.files[name]; ok {
return f, nil
}
return nil, os.ErrNotExist
}
func (fs *memFS) Remove(name string) error {
if _, ok := fs.files[name]; ok {
delete(fs.files, name)
return nil
}
return os.ErrNotExist
}
func (fs *memFS) Rename(oldpath, newpath string) error {
if f, ok := fs.files[oldpath]; ok {
delete(fs.files, oldpath)
fs.files[newpath] = f
f.name = newpath
return nil
}
return os.ErrNotExist
}
func (fs *memFS) ReadDir(dir string) ([]os.FileInfo, error) {
dir = filepath.Clean(dir)
var fis []os.FileInfo
for name, f := range fs.files {
if filepath.Dir(name) == dir {
fis = append(fis, f)
}
}
return fis, nil
}
type memFile struct {
name string
perm os.FileMode
buf []byte
size int64
offset int64
closed bool
}
func (f *memFile) Close() error {
if f.closed {
return os.ErrClosed
}
f.closed = true
return nil
}
func (f *memFile) Unlock() error {
if err := f.Close(); err != nil {
return err
}
return Mem.Remove(f.name)
}
func (f *memFile) ReadAt(p []byte, off int64) (int, error) {
if f.closed {
return 0, os.ErrClosed
}
if off >= f.size {
return 0, io.EOF
}
n := int64(len(p))
if n > f.size-off {
copy(p, f.buf[off:])
return int(f.size - off), nil
}
copy(p, f.buf[off:off+n])
return int(n), nil
}
func (f *memFile) Read(p []byte) (int, error) {
n, err := f.ReadAt(p, f.offset)
if err != nil {
return n, err
}
f.offset += int64(n)
return n, err
}
func (f *memFile) WriteAt(p []byte, off int64) (int, error) {
if f.closed {
return 0, os.ErrClosed
}
n := int64(len(p))
if off+n > f.size {
f.truncate(off + n)
}
copy(f.buf[off:off+n], p)
return int(n), nil
}
func (f *memFile) Write(p []byte) (int, error) {
n, err := f.WriteAt(p, f.offset)
if err != nil {
return n, err
}
f.offset += int64(n)
return n, err
}
func (f *memFile) Seek(offset int64, whence int) (int64, error) {
if f.closed {
return 0, os.ErrClosed
}
switch whence {
case io.SeekEnd:
f.offset = f.size + offset
case io.SeekStart:
f.offset = offset
case io.SeekCurrent:
f.offset += offset
}
return f.offset, nil
}
func (f *memFile) Stat() (os.FileInfo, error) {
if f.closed {
return f, os.ErrClosed
}
return f, nil
}
func (f *memFile) Sync() error {
if f.closed {
return os.ErrClosed
}
return nil
}
func (f *memFile) truncate(size int64) {
if size > f.size {
diff := int(size - f.size)
f.buf = append(f.buf, make([]byte, diff)...)
} else {
f.buf = f.buf[:size]
}
f.size = size
}
func (f *memFile) Truncate(size int64) error {
if f.closed {
return os.ErrClosed
}
f.truncate(size)
return nil
}
func (f *memFile) Name() string {
_, name := filepath.Split(f.name)
return name
}
func (f *memFile) Size() int64 {
return f.size
}
func (f *memFile) Mode() os.FileMode {
return f.perm
}
func (f *memFile) ModTime() time.Time {
return time.Now()
}
func (f *memFile) IsDir() bool {
return false
}
func (f *memFile) Sys() interface{} {
return nil
}
func (f *memFile) Slice(start int64, end int64) ([]byte, error) {
if f.closed {
return nil, os.ErrClosed
}
if end > f.size {
return nil, io.EOF
}
return f.buf[start:end], nil
}

64
vendor/github.com/akrylysov/pogreb/fs/os.go generated vendored Normal file
View File

@@ -0,0 +1,64 @@
package fs
import (
"io/ioutil"
"os"
)
type osFS struct{}
// OS is a file system backed by the os package.
var OS FileSystem = &osFS{}
func (fs *osFS) OpenFile(name string, flag int, perm os.FileMode) (File, error) {
f, err := os.OpenFile(name, flag, perm)
if err != nil {
return nil, err
}
return &osFile{File: f}, nil
}
func (fs *osFS) CreateLockFile(name string, perm os.FileMode) (LockFile, bool, error) {
return createLockFile(name, perm)
}
func (fs *osFS) Stat(name string) (os.FileInfo, error) {
return os.Stat(name)
}
func (fs *osFS) Remove(name string) error {
return os.Remove(name)
}
func (fs *osFS) Rename(oldpath, newpath string) error {
return os.Rename(oldpath, newpath)
}
func (fs *osFS) ReadDir(name string) ([]os.FileInfo, error) {
return ioutil.ReadDir(name)
}
type osFile struct {
*os.File
}
func (f *osFile) Slice(start int64, end int64) ([]byte, error) {
buf := make([]byte, end-start)
_, err := f.ReadAt(buf, start)
if err != nil {
return nil, err
}
return buf, nil
}
type osLockFile struct {
*os.File
path string
}
func (f *osLockFile) Unlock() error {
if err := os.Remove(f.path); err != nil {
return err
}
return f.Close()
}

163
vendor/github.com/akrylysov/pogreb/fs/os_mmap.go generated vendored Normal file
View File

@@ -0,0 +1,163 @@
package fs
import (
"io"
"os"
)
const (
initialMmapSize = 1024 << 20 // 1 GiB
)
type osMMapFS struct {
osFS
}
// OSMMap is a file system backed by the os package and memory-mapped files.
var OSMMap FileSystem = &osMMapFS{}
func (fs *osMMapFS) OpenFile(name string, flag int, perm os.FileMode) (File, error) {
if flag&os.O_APPEND != 0 {
// osMMapFS doesn't support opening files in append-only mode.
// The database doesn't currently use O_APPEND.
return nil, errAppendModeNotSupported
}
f, err := os.OpenFile(name, flag, perm)
if err != nil {
return nil, err
}
stat, err := f.Stat()
if err != nil {
return nil, err
}
mf := &osMMapFile{
File: f,
size: stat.Size(),
}
if err := mf.mremap(); err != nil {
return nil, err
}
return mf, nil
}
type osMMapFile struct {
*os.File
data []byte
offset int64
size int64
mmapSize int64
}
func (f *osMMapFile) WriteAt(p []byte, off int64) (int, error) {
n, err := f.File.WriteAt(p, off)
if err != nil {
return 0, err
}
writeOff := off + int64(n)
if writeOff > f.size {
f.size = writeOff
}
return n, f.mremap()
}
func (f *osMMapFile) Write(p []byte) (int, error) {
n, err := f.File.Write(p)
if err != nil {
return 0, err
}
f.offset += int64(n)
if f.offset > f.size {
f.size = f.offset
}
return n, f.mremap()
}
func (f *osMMapFile) Seek(offset int64, whence int) (int64, error) {
off, err := f.File.Seek(offset, whence)
f.offset = off
return off, err
}
func (f *osMMapFile) Read(p []byte) (int, error) {
n, err := f.File.Read(p)
f.offset += int64(n)
return n, err
}
func (f *osMMapFile) Slice(start int64, end int64) ([]byte, error) {
if end > f.size {
return nil, io.EOF
}
if f.data == nil {
return nil, os.ErrClosed
}
return f.data[start:end], nil
}
func (f *osMMapFile) munmap() error {
if f.data == nil {
return nil
}
if err := munmap(f.data); err != nil {
return err
}
f.data = nil
f.mmapSize = 0
return nil
}
func (f *osMMapFile) mmap(fileSize int64, mappingSize int64) error {
if f.data != nil {
if err := munmap(f.data); err != nil {
return err
}
}
data, err := mmap(f.File, fileSize, mappingSize)
if err != nil {
return err
}
_ = madviceRandom(data)
f.data = data
return nil
}
func (f *osMMapFile) mremap() error {
mmapSize := f.mmapSize
if mmapSize >= f.size {
return nil
}
if mmapSize == 0 {
mmapSize = initialMmapSize
if mmapSize < f.size {
mmapSize = f.size
}
} else {
if err := f.munmap(); err != nil {
return err
}
mmapSize *= 2
}
if err := f.mmap(f.size, mmapSize); err != nil {
return err
}
// On Windows mmap may memory-map less than the requested size.
f.mmapSize = int64(len(f.data))
return nil
}
func (f *osMMapFile) Close() error {
if err := f.munmap(); err != nil {
return err
}
return f.File.Close()
}

34
vendor/github.com/akrylysov/pogreb/fs/os_mmap_unix.go generated vendored Normal file
View File

@@ -0,0 +1,34 @@
// +build !windows
package fs
import (
"os"
"syscall"
"unsafe"
)
func mmap(f *os.File, fileSize int64, mappingSize int64) ([]byte, error) {
p, err := syscall.Mmap(int(f.Fd()), 0, int(mappingSize), syscall.PROT_READ, syscall.MAP_SHARED)
return p, err
}
func munmap(data []byte) error {
return syscall.Munmap(data)
}
func madviceRandom(data []byte) error {
_, _, errno := syscall.Syscall(syscall.SYS_MADVISE, uintptr(unsafe.Pointer(&data[0])), uintptr(len(data)), uintptr(syscall.MADV_RANDOM))
if errno != 0 {
return errno
}
return nil
}
func (f *osMMapFile) Truncate(size int64) error {
if err := f.File.Truncate(size); err != nil {
return err
}
f.size = size
return f.mremap()
}

45
vendor/github.com/akrylysov/pogreb/fs/os_mmap_windows.go generated vendored Normal file
View File

@@ -0,0 +1,45 @@
// +build windows
package fs
import (
"os"
"syscall"
"unsafe"
)
func mmap(f *os.File, fileSize int64, mappingSize int64) ([]byte, error) {
size := fileSize
low, high := uint32(size), uint32(size>>32)
fmap, err := syscall.CreateFileMapping(syscall.Handle(f.Fd()), nil, syscall.PAGE_READONLY, high, low, nil)
if err != nil {
return nil, err
}
defer syscall.CloseHandle(fmap)
ptr, err := syscall.MapViewOfFile(fmap, syscall.FILE_MAP_READ, 0, 0, uintptr(size))
if err != nil {
return nil, err
}
data := (*[maxMmapSize]byte)(unsafe.Pointer(ptr))[:size]
return data, nil
}
func munmap(data []byte) error {
return syscall.UnmapViewOfFile(uintptr(unsafe.Pointer(&data[0])))
}
func madviceRandom(data []byte) error {
return nil
}
func (f *osMMapFile) Truncate(size int64) error {
// Truncating a memory-mapped file fails on Windows. Unmap it first.
if err := f.munmap(); err != nil {
return err
}
if err := f.File.Truncate(size); err != nil {
return err
}
f.size = size
return f.mremap()
}

7
vendor/github.com/akrylysov/pogreb/fs/os_mmap_windows_386.go generated vendored Normal file
View File

@@ -0,0 +1,7 @@
package fs
import (
"math"
)
const maxMmapSize = math.MaxInt32

3
vendor/github.com/akrylysov/pogreb/fs/os_mmap_windows_amd64.go generated vendored Normal file
View File

@@ -0,0 +1,3 @@
package fs
const maxMmapSize = 1 << 48

26
vendor/github.com/akrylysov/pogreb/fs/os_unix.go generated vendored Normal file
View File

@@ -0,0 +1,26 @@
// +build !windows
package fs
import (
"os"
"syscall"
)
func createLockFile(name string, perm os.FileMode) (LockFile, bool, error) {
acquiredExisting := false
if _, err := os.Stat(name); err == nil {
acquiredExisting = true
}
f, err := os.OpenFile(name, os.O_RDWR|os.O_CREATE, perm)
if err != nil {
return nil, false, err
}
if err := syscall.Flock(int(f.Fd()), syscall.LOCK_EX|syscall.LOCK_NB); err != nil {
if err == syscall.EWOULDBLOCK {
err = os.ErrExist
}
return nil, false, err
}
return &osLockFile{f, name}, acquiredExisting, nil
}

60
vendor/github.com/akrylysov/pogreb/fs/os_windows.go generated vendored Normal file
View File

@@ -0,0 +1,60 @@
// +build windows
package fs
import (
"os"
"syscall"
"unsafe"
)
var (
modkernel32 = syscall.NewLazyDLL("kernel32.dll")
procLockFileEx = modkernel32.NewProc("LockFileEx")
)
const (
errorLockViolation = 0x21
)
func lockfile(f *os.File) error {
var ol syscall.Overlapped
r1, _, err := syscall.Syscall6(
procLockFileEx.Addr(),
6,
uintptr(f.Fd()), // handle
uintptr(0x0003),
uintptr(0), // reserved
uintptr(1), // locklow
uintptr(0), // lockhigh
uintptr(unsafe.Pointer(&ol)),
)
if r1 == 0 && (err == syscall.ERROR_FILE_EXISTS || err == errorLockViolation) {
return os.ErrExist
}
return nil
}
func createLockFile(name string, perm os.FileMode) (LockFile, bool, error) {
acquiredExisting := false
if _, err := os.Stat(name); err == nil {
acquiredExisting = true
}
fd, err := syscall.CreateFile(&(syscall.StringToUTF16(name)[0]),
syscall.GENERIC_READ|syscall.GENERIC_WRITE,
syscall.FILE_SHARE_READ|syscall.FILE_SHARE_WRITE|syscall.FILE_SHARE_DELETE,
nil,
syscall.CREATE_ALWAYS,
syscall.FILE_ATTRIBUTE_NORMAL,
0)
if err != nil {
return nil, false, os.ErrExist
}
f := os.NewFile(uintptr(fd), name)
if err := lockfile(f); err != nil {
f.Close()
return nil, false, err
}
return &osLockFile{f, name}, acquiredExisting, nil
}

52
vendor/github.com/akrylysov/pogreb/fs/sub.go generated vendored Normal file
View File

@@ -0,0 +1,52 @@
package fs
import (
"os"
"path/filepath"
)
// Sub returns a new file system rooted at dir.
func Sub(fsys FileSystem, dir string) FileSystem {
return &subFS{
fsys: fsys,
root: dir,
}
}
type subFS struct {
fsys FileSystem
root string
}
func (fs *subFS) OpenFile(name string, flag int, perm os.FileMode) (File, error) {
subName := filepath.Join(fs.root, name)
return fs.fsys.OpenFile(subName, flag, perm)
}
func (fs *subFS) Stat(name string) (os.FileInfo, error) {
subName := filepath.Join(fs.root, name)
return fs.fsys.Stat(subName)
}
func (fs *subFS) Remove(name string) error {
subName := filepath.Join(fs.root, name)
return fs.fsys.Remove(subName)
}
func (fs *subFS) Rename(oldpath, newpath string) error {
subOldpath := filepath.Join(fs.root, oldpath)
subNewpath := filepath.Join(fs.root, newpath)
return fs.fsys.Rename(subOldpath, subNewpath)
}
func (fs *subFS) ReadDir(name string) ([]os.FileInfo, error) {
subName := filepath.Join(fs.root, name)
return fs.fsys.ReadDir(subName)
}
func (fs *subFS) CreateLockFile(name string, perm os.FileMode) (LockFile, bool, error) {
subName := filepath.Join(fs.root, name)
return fs.fsys.CreateLockFile(subName, perm)
}
var _ FileSystem = &subFS{}

27
vendor/github.com/akrylysov/pogreb/gobfile.go generated vendored Normal file
View File

@@ -0,0 +1,27 @@
package pogreb
import (
"encoding/gob"
"github.com/akrylysov/pogreb/fs"
)
func readGobFile(fsys fs.FileSystem, name string, v interface{}) error {
f, err := openFile(fsys, name, false)
if err != nil {
return err
}
defer f.Close()
dec := gob.NewDecoder(f)
return dec.Decode(v)
}
func writeGobFile(fsys fs.FileSystem, name string, v interface{}) error {
f, err := openFile(fsys, name, true)
if err != nil {
return err
}
defer f.Close()
enc := gob.NewEncoder(f)
return enc.Encode(v)
}

43
vendor/github.com/akrylysov/pogreb/header.go generated vendored Normal file
View File

@@ -0,0 +1,43 @@
package pogreb
import (
"bytes"
"encoding/binary"
)
const (
formatVersion = 2 // File format version.
headerSize = 512
)
var (
signature = [8]byte{'p', 'o', 'g', 'r', 'e', 'b', '\x0e', '\xfd'}
)
type header struct {
signature [8]byte
formatVersion uint32
}
func newHeader() *header {
return &header{
signature: signature,
formatVersion: formatVersion,
}
}
func (h header) MarshalBinary() ([]byte, error) {
buf := make([]byte, headerSize)
copy(buf[:8], h.signature[:])
binary.LittleEndian.PutUint32(buf[8:12], h.formatVersion)
return buf, nil
}
func (h *header) UnmarshalBinary(data []byte) error {
if !bytes.Equal(data[:8], signature[:]) {
return errCorrupted
}
copy(h.signature[:], data[:8])
h.formatVersion = binary.LittleEndian.Uint32(data[8:12])
return nil
}

363
vendor/github.com/akrylysov/pogreb/index.go generated vendored Normal file
View File

@@ -0,0 +1,363 @@
package pogreb
import (
"github.com/akrylysov/pogreb/internal/errors"
)
const (
indexExt = ".pix"
indexMainName = "main" + indexExt
indexOverflowName = "overflow" + indexExt
indexMetaName = "index" + metaExt
loadFactor = 0.7
)
// index is an on-disk linear hashing hash table.
// It uses two files to store the hash table on disk - "main" and "overflow" index files.
// Each index file holds an array of buckets.
type index struct {
opts *Options
main *file // Main index file.
overflow *file // Overflow index file.
freeBucketOffs []int64 // Offsets of freed buckets.
level uint8 // Maximum number of buckets on a logarithmic scale.
numKeys uint32 // Number of keys.
numBuckets uint32 // Number of buckets.
splitBucketIdx uint32 // Index of the bucket to split on next split.
}
type indexMeta struct {
Level uint8
NumKeys uint32
NumBuckets uint32
SplitBucketIndex uint32
FreeOverflowBuckets []int64
}
// matchKeyFunc returns whether the slot matches the key sought.
type matchKeyFunc func(slot) (bool, error)
func openIndex(opts *Options) (*index, error) {
main, err := openFile(opts.FileSystem, indexMainName, false)
if err != nil {
return nil, errors.Wrap(err, "opening main index")
}
overflow, err := openFile(opts.FileSystem, indexOverflowName, false)
if err != nil {
_ = main.Close()
return nil, errors.Wrap(err, "opening overflow index")
}
idx := &index{
opts: opts,
main: main,
overflow: overflow,
numBuckets: 1,
}
if main.empty() {
// Add an empty bucket.
if _, err = idx.main.extend(bucketSize); err != nil {
_ = main.Close()
_ = overflow.Close()
return nil, err
}
} else if err := idx.readMeta(); err != nil {
_ = main.Close()
_ = overflow.Close()
return nil, errors.Wrap(err, "opening index meta")
}
return idx, nil
}
func (idx *index) writeMeta() error {
m := indexMeta{
Level: idx.level,
NumKeys: idx.numKeys,
NumBuckets: idx.numBuckets,
SplitBucketIndex: idx.splitBucketIdx,
FreeOverflowBuckets: idx.freeBucketOffs,
}
return writeGobFile(idx.opts.FileSystem, indexMetaName, m)
}
func (idx *index) readMeta() error {
m := indexMeta{}
if err := readGobFile(idx.opts.FileSystem, indexMetaName, &m); err != nil {
return err
}
idx.level = m.Level
idx.numKeys = m.NumKeys
idx.numBuckets = m.NumBuckets
idx.splitBucketIdx = m.SplitBucketIndex
idx.freeBucketOffs = m.FreeOverflowBuckets
return nil
}
func (idx *index) bucketIndex(hash uint32) uint32 {
bidx := hash & ((1 << idx.level) - 1)
if bidx < idx.splitBucketIdx {
return hash & ((1 << (idx.level + 1)) - 1)
}
return bidx
}
type bucketIterator struct {
off int64 // Offset of the next bucket.
f *file // Current index file.
overflow *file // Overflow index file.
}
// bucketOffset returns on-disk bucket offset by the bucket index.
func bucketOffset(idx uint32) int64 {
return int64(headerSize) + (int64(bucketSize) * int64(idx))
}
func (idx *index) newBucketIterator(startBucketIdx uint32) *bucketIterator {
return &bucketIterator{
off: bucketOffset(startBucketIdx),
f: idx.main,
overflow: idx.overflow,
}
}
func (it *bucketIterator) next() (bucketHandle, error) {
if it.off == 0 {
return bucketHandle{}, ErrIterationDone
}
b := bucketHandle{file: it.f, offset: it.off}
if err := b.read(); err != nil {
return bucketHandle{}, err
}
it.f = it.overflow
it.off = b.next
return b, nil
}
func (idx *index) get(hash uint32, matchKey matchKeyFunc) error {
it := idx.newBucketIterator(idx.bucketIndex(hash))
for {
b, err := it.next()
if err == ErrIterationDone {
return nil
}
if err != nil {
return err
}
for i := 0; i < slotsPerBucket; i++ {
sl := b.slots[i]
// No more slots in the bucket.
if sl.offset == 0 {
break
}
if hash != sl.hash {
continue
}
if match, err := matchKey(sl); match || err != nil {
return err
}
}
}
}
func (idx *index) findInsertionBucket(newSlot slot, matchKey matchKeyFunc) (*slotWriter, bool, error) {
sw := &slotWriter{}
it := idx.newBucketIterator(idx.bucketIndex(newSlot.hash))
for {
b, err := it.next()
if err == ErrIterationDone {
return nil, false, errors.New("failed to insert a new slot")
}
if err != nil {
return nil, false, err
}
sw.bucket = &b
var i int
for i = 0; i < slotsPerBucket; i++ {
sl := b.slots[i]
if sl.offset == 0 {
// Found an empty slot.
sw.slotIdx = i
return sw, false, nil
}
if newSlot.hash != sl.hash {
continue
}
match, err := matchKey(sl)
if err != nil {
return nil, false, err
}
if match {
// Key already in the index.
// The slot writer will overwrite the existing slot.
sw.slotIdx = i
return sw, true, nil
}
}
if b.next == 0 {
// No more buckets in the chain.
sw.slotIdx = i
return sw, false, nil
}
}
}
func (idx *index) put(newSlot slot, matchKey matchKeyFunc) error {
if idx.numKeys == MaxKeys {
return errFull
}
sw, overwritingExisting, err := idx.findInsertionBucket(newSlot, matchKey)
if err != nil {
return err
}
if err := sw.insert(newSlot, idx); err != nil {
return err
}
if err := sw.write(); err != nil {
return err
}
if overwritingExisting {
return nil
}
idx.numKeys++
if float64(idx.numKeys)/float64(idx.numBuckets*slotsPerBucket) > loadFactor {
if err := idx.split(); err != nil {
return err
}
}
return nil
}
func (idx *index) delete(hash uint32, matchKey matchKeyFunc) error {
it := idx.newBucketIterator(idx.bucketIndex(hash))
for {
b, err := it.next()
if err == ErrIterationDone {
return nil
}
if err != nil {
return err
}
for i := 0; i < slotsPerBucket; i++ {
sl := b.slots[i]
if sl.offset == 0 {
break
}
if hash != sl.hash {
continue
}
match, err := matchKey(sl)
if err != nil {
return err
}
if !match {
continue
}
b.del(i)
if err := b.write(); err != nil {
return err
}
idx.numKeys--
return nil
}
}
}
func (idx *index) createOverflowBucket() (*bucketHandle, error) {
var off int64
if len(idx.freeBucketOffs) > 0 {
off = idx.freeBucketOffs[0]
idx.freeBucketOffs = idx.freeBucketOffs[1:]
} else {
var err error
off, err = idx.overflow.extend(bucketSize)
if err != nil {
return nil, err
}
}
return &bucketHandle{file: idx.overflow, offset: off}, nil
}
func (idx *index) freeOverflowBucket(offsets ...int64) {
idx.freeBucketOffs = append(idx.freeBucketOffs, offsets...)
}
func (idx *index) split() error {
updatedBucketIdx := idx.splitBucketIdx
updatedBucketOff := bucketOffset(updatedBucketIdx)
updatedBucket := slotWriter{
bucket: &bucketHandle{file: idx.main, offset: updatedBucketOff},
}
newBucketOff, err := idx.main.extend(bucketSize)
if err != nil {
return err
}
sw := slotWriter{
bucket: &bucketHandle{file: idx.main, offset: newBucketOff},
}
idx.splitBucketIdx++
if idx.splitBucketIdx == 1<<idx.level {
idx.level++
idx.splitBucketIdx = 0
}
var overflowBuckets []int64
it := idx.newBucketIterator(updatedBucketIdx)
for {
b, err := it.next()
if err == ErrIterationDone {
break
}
if err != nil {
return err
}
for j := 0; j < slotsPerBucket; j++ {
sl := b.slots[j]
if sl.offset == 0 {
break
}
if idx.bucketIndex(sl.hash) == updatedBucketIdx {
if err := updatedBucket.insert(sl, idx); err != nil {
return err
}
} else {
if err := sw.insert(sl, idx); err != nil {
return err
}
}
}
if b.next != 0 {
overflowBuckets = append(overflowBuckets, b.next)
}
}
idx.freeOverflowBucket(overflowBuckets...)
if err := sw.write(); err != nil {
return err
}
if err := updatedBucket.write(); err != nil {
return err
}
idx.numBuckets++
return nil
}
func (idx *index) close() error {
if err := idx.writeMeta(); err != nil {
return err
}
if err := idx.main.Close(); err != nil {
return err
}
if err := idx.overflow.Close(); err != nil {
return err
}
return nil
}
func (idx *index) count() uint32 {
return idx.numKeys
}

42
vendor/github.com/akrylysov/pogreb/internal/errors/errors.go generated vendored Normal file
View File

@@ -0,0 +1,42 @@
package errors
import (
"errors"
"fmt"
)
type wrappedError struct {
cause error
msg string
}
func (we wrappedError) Error() string {
return we.msg + ": " + we.cause.Error()
}
func (we wrappedError) Unwrap() error {
return we.cause
}
// New returns an error that formats as the given text.
func New(text string) error {
return errors.New(text)
}
// Wrap returns an error annotating err with an additional message.
// Compatible with Go 1.13 error chains.
func Wrap(cause error, message string) error {
return wrappedError{
cause: cause,
msg: message,
}
}
// Wrapf returns an error annotating err with an additional formatted message.
// Compatible with Go 1.13 error chains.
func Wrapf(cause error, format string, a ...interface{}) error {
return wrappedError{
cause: cause,
msg: fmt.Sprintf(format, a...),
}
}

55
vendor/github.com/akrylysov/pogreb/internal/hash/murmurhash32.go generated vendored Normal file
View File

@@ -0,0 +1,55 @@
package hash
import (
"math/bits"
)
const (
c1 uint32 = 0xcc9e2d51
c2 uint32 = 0x1b873593
)
// Sum32WithSeed is a port of MurmurHash3_x86_32 function.
func Sum32WithSeed(data []byte, seed uint32) uint32 {
h1 := seed
dlen := len(data)
for len(data) >= 4 {
k1 := uint32(data[0]) | uint32(data[1])<<8 | uint32(data[2])<<16 | uint32(data[3])<<24
data = data[4:]
k1 *= c1
k1 = bits.RotateLeft32(k1, 15)
k1 *= c2
h1 ^= k1
h1 = bits.RotateLeft32(h1, 13)
h1 = h1*5 + 0xe6546b64
}
var k1 uint32
switch len(data) {
case 3:
k1 ^= uint32(data[2]) << 16
fallthrough
case 2:
k1 ^= uint32(data[1]) << 8
fallthrough
case 1:
k1 ^= uint32(data[0])
k1 *= c1
k1 = bits.RotateLeft32(k1, 15)
k1 *= c2
h1 ^= k1
}
h1 ^= uint32(dlen)
h1 ^= h1 >> 16
h1 *= 0x85ebca6b
h1 ^= h1 >> 13
h1 *= 0xc2b2ae35
h1 ^= h1 >> 16
return h1
}

15
vendor/github.com/akrylysov/pogreb/internal/hash/seed.go generated vendored Normal file
View File

@@ -0,0 +1,15 @@
package hash
import (
"crypto/rand"
"encoding/binary"
)
// RandSeed generates a random hash seed.
func RandSeed() (uint32, error) {
b := make([]byte, 4)
if _, err := rand.Read(b); err != nil {
return 0, err
}
return binary.LittleEndian.Uint32(b), nil
}

75
vendor/github.com/akrylysov/pogreb/iterator.go generated vendored Normal file
View File

@@ -0,0 +1,75 @@
package pogreb
import (
"errors"
"sync"
)
// ErrIterationDone is returned by ItemIterator.Next calls when there are no more items to return.
var ErrIterationDone = errors.New("no more items in iterator")
type item struct {
key []byte
value []byte
}
// ItemIterator is an iterator over DB key-value pairs. It iterates the items in an unspecified order.
type ItemIterator struct {
db *DB
nextBucketIdx uint32
queue []item
mu sync.Mutex
}
// fetchItems adds items to the iterator queue from a bucket located at nextBucketIdx.
func (it *ItemIterator) fetchItems(nextBucketIdx uint32) error {
bit := it.db.index.newBucketIterator(nextBucketIdx)
for {
b, err := bit.next()
if err == ErrIterationDone {
return nil
}
if err != nil {
return err
}
for i := 0; i < slotsPerBucket; i++ {
sl := b.slots[i]
if sl.offset == 0 {
// No more items in the bucket.
break
}
key, value, err := it.db.datalog.readKeyValue(sl)
if err != nil {
return err
}
key = cloneBytes(key)
value = cloneBytes(value)
it.queue = append(it.queue, item{key: key, value: value})
}
}
}
// Next returns the next key-value pair if available, otherwise it returns ErrIterationDone error.
func (it *ItemIterator) Next() ([]byte, []byte, error) {
it.mu.Lock()
defer it.mu.Unlock()
it.db.mu.RLock()
defer it.db.mu.RUnlock()
// The iterator queue is empty and we have more buckets to check.
for len(it.queue) == 0 && it.nextBucketIdx < it.db.index.numBuckets {
if err := it.fetchItems(it.nextBucketIdx); err != nil {
return nil, nil, err
}
it.nextBucketIdx++
}
if len(it.queue) > 0 {
item := it.queue[0]
it.queue = it.queue[1:]
return item.key, item.value, nil
}
return nil, nil, ErrIterationDone
}

15
vendor/github.com/akrylysov/pogreb/lock.go generated vendored Normal file
View File

@@ -0,0 +1,15 @@
package pogreb
import (
"os"
"github.com/akrylysov/pogreb/fs"
)
const (
lockName = "lock"
)
func createLockFile(opts *Options) (fs.LockFile, bool, error) {
return opts.FileSystem.CreateLockFile(lockName, os.FileMode(0644))
}

15
vendor/github.com/akrylysov/pogreb/logger.go generated vendored Normal file
View File

@@ -0,0 +1,15 @@
package pogreb
import (
"log"
"os"
)
var logger = log.New(os.Stderr, "pogreb: ", 0)
// SetLogger sets the global logger.
func SetLogger(l *log.Logger) {
if l != nil {
logger = l
}
}

11
vendor/github.com/akrylysov/pogreb/metrics.go generated vendored Normal file
View File

@@ -0,0 +1,11 @@
package pogreb
import "expvar"
// Metrics holds the DB metrics.
type Metrics struct {
Puts expvar.Int
Dels expvar.Int
Gets expvar.Int
HashCollisions expvar.Int
}

52
vendor/github.com/akrylysov/pogreb/options.go generated vendored Normal file
View File

@@ -0,0 +1,52 @@
package pogreb
import (
"math"
"time"
"github.com/akrylysov/pogreb/fs"
)
// Options holds the optional DB parameters.
type Options struct {
// BackgroundSyncInterval sets the amount of time between background Sync() calls.
//
// Setting the value to 0 disables the automatic background synchronization.
// Setting the value to -1 makes the DB call Sync() after every write operation.
BackgroundSyncInterval time.Duration
// BackgroundCompactionInterval sets the amount of time between background Compact() calls.
//
// Setting the value to 0 disables the automatic background compaction.
BackgroundCompactionInterval time.Duration
// FileSystem sets the file system implementation.
//
// Default: fs.OSMMap.
FileSystem fs.FileSystem
maxSegmentSize uint32
compactionMinSegmentSize uint32
compactionMinFragmentation float32
}
func (src *Options) copyWithDefaults(path string) *Options {
opts := Options{}
if src != nil {
opts = *src
}
if opts.FileSystem == nil {
opts.FileSystem = fs.OSMMap
}
opts.FileSystem = fs.Sub(opts.FileSystem, path)
if opts.maxSegmentSize == 0 {
opts.maxSegmentSize = math.MaxUint32
}
if opts.compactionMinSegmentSize == 0 {
opts.compactionMinSegmentSize = 32 << 20
}
if opts.compactionMinFragmentation == 0 {
opts.compactionMinFragmentation = 0.5
}
return &opts
}

161
vendor/github.com/akrylysov/pogreb/recovery.go generated vendored Normal file
View File

@@ -0,0 +1,161 @@
package pogreb
import (
"io"
"path/filepath"
"github.com/akrylysov/pogreb/fs"
)
const (
recoveryBackupExt = ".bac"
)
func backupNonsegmentFiles(fsys fs.FileSystem) error {
logger.Println("moving non-segment files...")
files, err := fsys.ReadDir(".")
if err != nil {
return err
}
for _, file := range files {
name := file.Name()
ext := filepath.Ext(name)
if ext == segmentExt || name == lockName {
continue
}
dst := name + recoveryBackupExt
if err := fsys.Rename(name, dst); err != nil {
return err
}
logger.Printf("moved %s to %s", name, dst)
}
return nil
}
func removeRecoveryBackupFiles(fsys fs.FileSystem) error {
logger.Println("removing recovery backup files...")
files, err := fsys.ReadDir(".")
if err != nil {
return err
}
for _, file := range files {
name := file.Name()
ext := filepath.Ext(name)
if ext != recoveryBackupExt {
continue
}
if err := fsys.Remove(name); err != nil {
return err
}
logger.Printf("removed %s", name)
}
return nil
}
// recoveryIterator iterates over records of all datalog segments in insertion order.
// Corrupted segments are truncated to the last valid record.
type recoveryIterator struct {
segments []*segment
segit *segmentIterator
}
func newRecoveryIterator(segments []*segment) *recoveryIterator {
return &recoveryIterator{
segments: segments,
}
}
func (it *recoveryIterator) next() (record, error) {
for {
if it.segit == nil {
if len(it.segments) == 0 {
return record{}, ErrIterationDone
}
var err error
it.segit, err = newSegmentIterator(it.segments[0])
if err != nil {
return record{}, err
}
it.segments = it.segments[1:]
}
rec, err := it.segit.next()
if err == io.EOF || err == io.ErrUnexpectedEOF || err == errCorrupted {
// Truncate file to the last valid offset.
if err := it.segit.f.Truncate(int64(it.segit.offset)); err != nil {
return record{}, err
}
fi, fierr := it.segit.f.Stat()
if fierr != nil {
return record{}, fierr
}
logger.Printf("truncated segment %s to offset %d", fi.Name(), it.segit.offset)
err = ErrIterationDone
}
if err == ErrIterationDone {
it.segit = nil
continue
}
if err != nil {
return record{}, err
}
return rec, nil
}
}
func (db *DB) recover() error {
logger.Println("started recovery")
logger.Println("rebuilding index...")
segments := db.datalog.segmentsBySequenceID()
it := newRecoveryIterator(segments)
for {
rec, err := it.next()
if err == ErrIterationDone {
break
}
if err != nil {
return err
}
h := db.hash(rec.key)
meta := db.datalog.segments[rec.segmentID].meta
if rec.rtype == recordTypePut {
sl := slot{
hash: h,
segmentID: rec.segmentID,
keySize: uint16(len(rec.key)),
valueSize: uint32(len(rec.value)),
offset: rec.offset,
}
if err := db.put(sl, rec.key); err != nil {
return err
}
meta.PutRecords++
} else {
if err := db.del(h, rec.key, false); err != nil {
return err
}
meta.DeleteRecords++
meta.DeletedBytes += uint32(len(rec.data))
}
}
// Mark all segments except the newest as full.
for i := 0; i < len(segments)-1; i++ {
segments[i].meta.Full = true
}
if err := removeRecoveryBackupFiles(db.opts.FileSystem); err != nil {
logger.Printf("error removing recovery backups files: %v", err)
}
logger.Println("successfully recovered database")
return nil
}

156
vendor/github.com/akrylysov/pogreb/segment.go generated vendored Normal file
View File

@@ -0,0 +1,156 @@
package pogreb
import (
"bufio"
"encoding/binary"
"fmt"
"hash/crc32"
"io"
)
type recordType int
const (
recordTypePut recordType = iota
recordTypeDelete
segmentExt = ".psg"
)
// segment is a write-ahead log segment.
// It consists of a sequence of binary-encoded variable length records.
type segment struct {
*file
id uint16 // Physical segment identifier.
sequenceID uint64 // Logical monotonically increasing segment identifier.
name string
meta *segmentMeta
}
func segmentName(id uint16, sequenceID uint64) string {
return fmt.Sprintf("%05d-%d%s", id, sequenceID, segmentExt)
}
type segmentMeta struct {
Full bool
PutRecords uint32
DeleteRecords uint32
DeletedKeys uint32
DeletedBytes uint32
}
func segmentMetaName(id uint16, sequenceID uint64) string {
return segmentName(id, sequenceID) + metaExt
}
// Binary representation of a segment record:
// +---------------+------------------+------------------+-...-+--...--+----------+
// | Key Size (2B) | Record Type (1b) | Value Size (31b) | Key | Value | CRC (4B) |
// +---------------+------------------+------------------+-...-+--...--+----------+
type record struct {
rtype recordType
segmentID uint16
offset uint32
data []byte
key []byte
value []byte
}
func encodedRecordSize(kvSize uint32) uint32 {
// key size, value size, key, value, crc32
return 2 + 4 + kvSize + 4
}
func encodeRecord(key []byte, value []byte, rt recordType) []byte {
size := encodedRecordSize(uint32(len(key) + len(value)))
data := make([]byte, size)
binary.LittleEndian.PutUint16(data[:2], uint16(len(key)))
valLen := uint32(len(value))
if rt == recordTypeDelete { // Set delete bit.
valLen |= 1 << 31
}
binary.LittleEndian.PutUint32(data[2:], valLen)
copy(data[6:], key)
copy(data[6+len(key):], value)
checksum := crc32.ChecksumIEEE(data[:6+len(key)+len(value)])
binary.LittleEndian.PutUint32(data[size-4:size], checksum)
return data
}
func encodePutRecord(key []byte, value []byte) []byte {
return encodeRecord(key, value, recordTypePut)
}
func encodeDeleteRecord(key []byte) []byte {
return encodeRecord(key, nil, recordTypeDelete)
}
// segmentIterator iterates over segment records.
type segmentIterator struct {
f *segment
offset uint32
r *bufio.Reader
buf []byte // kv size and crc32 reusable buffer.
}
func newSegmentIterator(f *segment) (*segmentIterator, error) {
if _, err := f.Seek(int64(headerSize), io.SeekStart); err != nil {
return nil, err
}
return &segmentIterator{
f: f,
offset: headerSize,
r: bufio.NewReader(f),
buf: make([]byte, 6),
}, nil
}
func (it *segmentIterator) next() (record, error) {
// Read key and value size.
kvSizeBuf := it.buf
if _, err := io.ReadFull(it.r, kvSizeBuf); err != nil {
if err == io.EOF {
return record{}, ErrIterationDone
}
return record{}, err
}
// Decode key size.
keySize := uint32(binary.LittleEndian.Uint16(kvSizeBuf[:2]))
// Decode value size and record type.
rt := recordTypePut
valueSize := binary.LittleEndian.Uint32(kvSizeBuf[2:])
if valueSize&(1<<31) != 0 {
rt = recordTypeDelete
valueSize &^= 1 << 31
}
// Read key, value and checksum.
recordSize := encodedRecordSize(keySize + valueSize)
data := make([]byte, recordSize)
copy(data, kvSizeBuf)
if _, err := io.ReadFull(it.r, data[6:]); err != nil {
return record{}, err
}
// Verify checksum.
checksum := binary.LittleEndian.Uint32(data[len(data)-4:])
if checksum != crc32.ChecksumIEEE(data[:len(data)-4]) {
return record{}, errCorrupted
}
offset := it.offset
it.offset += recordSize
rec := record{
rtype: rt,
segmentID: it.f.id,
offset: offset,
data: data,
key: data[6 : 6+keySize],
value: data[6+keySize : 6+keySize+valueSize],
}
return rec, nil
}