Refactored configuration option handling. Fixes #3

:wq

@@ -0,0 +1,374 @@
+package bitcask
+
+import (
+	"fmt"
+	"hash/crc32"
+	"io"
+	"io/ioutil"
+	"os"
+	"path/filepath"
+	"strings"
+
+	"github.com/gofrs/flock"
+	"github.com/prologic/trie"
+)
+
+type Bitcask struct {
+	*flock.Flock
+
+	config    config
+	path      string
+	curr      *Datafile
+	keydir    *Keydir
+	datafiles []*Datafile
+	trie      *trie.Trie
+
+	maxDatafileSize int64
+}
+
+func (b *Bitcask) Close() error {
+	defer func() {
+		b.Flock.Unlock()
+		os.Remove(b.Flock.Path())
+	}()
+
+	for _, df := range b.datafiles {
+		df.Close()
+	}
+	return b.curr.Close()
+}
+
+func (b *Bitcask) Sync() error {
+	return b.curr.Sync()
+}
+
+func (b *Bitcask) Get(key string) ([]byte, error) {
+	var df *Datafile
+
+	item, ok := b.keydir.Get(key)
+	if !ok {
+		return nil, fmt.Errorf("error: key not found %s", key)
+	}
+
+	if item.FileID == b.curr.id {
+		df = b.curr
+	} else {
+		df = b.datafiles[item.FileID]
+	}
+
+	e, err := df.ReadAt(item.Offset)
+	if err != nil {
+		return nil, err
+	}
+
+	checksum := crc32.ChecksumIEEE(e.Value)
+	if checksum != e.Checksum {
+		return nil, fmt.Errorf("error: checksum falied %s %d != %d", key, e.Checksum, checksum)
+	}
+
+	return e.Value, nil
+}
+
+func (b *Bitcask) Put(key string, value []byte) error {
+	if len(key) > b.config.MaxKeySize {
+		return fmt.Errorf("error: key too large %d > %d", len(key), b.config.MaxKeySize)
+	}
+	if len(value) > b.config.MaxValueSize {
+		return fmt.Errorf("error: value too large %d > %d", len(value), b.config.MaxValueSize)
+	}
+
+	offset, err := b.put(key, value)
+	if err != nil {
+		return err
+	}
+
+	item := b.keydir.Add(key, b.curr.id, offset)
+	b.trie.Add(key, item)
+
+	return nil
+}
+
+func (b *Bitcask) Delete(key string) error {
+	_, err := b.put(key, []byte{})
+	if err != nil {
+		return err
+	}
+
+	b.keydir.Delete(key)
+	b.trie.Remove(key)
+
+	return nil
+}
+
+func (b *Bitcask) Scan(prefix string, f func(key string) error) error {
+	keys := b.trie.PrefixSearch(prefix)
+	for _, key := range keys {
+		if err := f(key); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func (b *Bitcask) Fold(f func(key string) error) error {
+	for key := range b.keydir.Keys() {
+		if err := f(key); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func (b *Bitcask) put(key string, value []byte) (int64, error) {
+	size, err := b.curr.Size()
+	if err != nil {
+		return -1, err
+	}
+
+	if size >= b.maxDatafileSize {
+		err := b.curr.Close()
+		if err != nil {
+			return -1, err
+		}
+
+		df, err := NewDatafile(b.path, b.curr.id, true)
+		if err != nil {
+			return -1, err
+		}
+
+		b.datafiles = append(b.datafiles, df)
+
+		id := b.curr.id + 1
+		curr, err := NewDatafile(b.path, id, false)
+		if err != nil {
+			return -1, err
+		}
+		b.curr = curr
+	}
+
+	e := NewEntry(key, value)
+	return b.curr.Write(e)
+}
+
+func (b *Bitcask) setMaxDatafileSize(size int64) error {
+	b.maxDatafileSize = size
+	return nil
+}
+
+func Merge(path string, force bool) error {
+	fns, err := getDatafiles(path)
+	if err != nil {
+		return err
+	}
+
+	ids, err := parseIds(fns)
+	if err != nil {
+		return err
+	}
+
+	// Do not merge if we only have 1 Datafile
+	if len(ids) <= 1 {
+		return nil
+	}
+
+	// Don't merge the Active Datafile (the last one)
+	fns = fns[:len(fns)-1]
+	ids = ids[:len(ids)-1]
+
+	temp, err := ioutil.TempDir("", "bitcask")
+	if err != nil {
+		return err
+	}
+
+	for i, fn := range fns {
+		// Don't merge Datafiles whose .hint files we've already generated
+		// (they are already merged); unless we set the force flag to true
+		// (forcing a re-merge).
+		if filepath.Ext(fn) == ".hint" && !force {
+			// Already merged
+			continue
+		}
+
+		id := ids[i]
+
+		keydir := NewKeydir()
+
+		df, err := NewDatafile(path, id, true)
+		if err != nil {
+			return err
+		}
+		defer df.Close()
+
+		for {
+			e, err := df.Read()
+			if err != nil {
+				if err == io.EOF {
+					break
+				}
+				return err
+			}
+
+			// Tombstone value  (deleted key)
+			if len(e.Value) == 0 {
+				keydir.Delete(e.Key)
+				continue
+			}
+
+			keydir.Add(e.Key, ids[i], e.Offset)
+		}
+
+		tempdf, err := NewDatafile(temp, id, false)
+		if err != nil {
+			return err
+		}
+		defer tempdf.Close()
+
+		for key := range keydir.Keys() {
+			item, _ := keydir.Get(key)
+			e, err := df.ReadAt(item.Offset)
+			if err != nil {
+				return err
+			}
+
+			_, err = tempdf.Write(e)
+			if err != nil {
+				return err
+			}
+		}
+
+		err = tempdf.Close()
+		if err != nil {
+			return err
+		}
+
+		err = df.Close()
+		if err != nil {
+			return err
+		}
+
+		err = os.Rename(tempdf.Name(), df.Name())
+		if err != nil {
+			return err
+		}
+
+		hint := strings.TrimSuffix(df.Name(), ".data") + ".hint"
+		err = keydir.Save(hint)
+		if err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
+func Open(path string, options ...option) (*Bitcask, error) {
+	if err := os.MkdirAll(path, 0755); err != nil {
+		return nil, err
+	}
+
+	err := Merge(path, false)
+	if err != nil {
+		return nil, err
+	}
+
+	fns, err := getDatafiles(path)
+	if err != nil {
+		return nil, err
+	}
+
+	ids, err := parseIds(fns)
+	if err != nil {
+		return nil, err
+	}
+
+	var datafiles []*Datafile
+
+	keydir := NewKeydir()
+	trie := trie.New()
+
+	for i, fn := range fns {
+		df, err := NewDatafile(path, ids[i], true)
+		if err != nil {
+			return nil, err
+		}
+		datafiles = append(datafiles, df)
+
+		if filepath.Ext(fn) == ".hint" {
+			f, err := os.Open(filepath.Join(path, fn))
+			if err != nil {
+				return nil, err
+			}
+			defer f.Close()
+
+			hint, err := NewKeydirFromBytes(f)
+			if err != nil {
+				return nil, err
+			}
+
+			for key := range hint.Keys() {
+				item, _ := hint.Get(key)
+				_ = keydir.Add(key, item.FileID, item.Offset)
+				trie.Add(key, item)
+			}
+		} else {
+			for {
+				e, err := df.Read()
+				if err != nil {
+					if err == io.EOF {
+						break
+					}
+					return nil, err
+				}
+
+				// Tombstone value  (deleted key)
+				if len(e.Value) == 0 {
+					keydir.Delete(e.Key)
+					continue
+				}
+
+				item := keydir.Add(e.Key, ids[i], e.Offset)
+				trie.Add(e.Key, item)
+			}
+		}
+	}
+
+	var id int
+	if len(ids) > 0 {
+		id = ids[(len(ids) - 1)]
+	}
+
+	curr, err := NewDatafile(path, id, false)
+	if err != nil {
+		return nil, err
+	}
+
+	bitcask := &Bitcask{
+		Flock:     flock.New(filepath.Join(path, "lock")),
+		config:    NewDefaultConfig(),
+		path:      path,
+		curr:      curr,
+		keydir:    keydir,
+		datafiles: datafiles,
+		trie:      trie,
+
+		maxDatafileSize: DefaultMaxDatafileSize,
+	}
+
+	for _, opt := range options {
+		err = opt(bitcask.config)
+		if err != nil {
+			return nil, err
+		}
+	}
+
+	locked, err := bitcask.Flock.TryLock()
+	if err != nil {
+		return nil, err
+	}
+
+	if !locked {
+		return nil, fmt.Errorf("error: database locked %s", path)
+	}
+
+	return bitcask, nil
+}

bitcask.go

@@ -2,12 +2,12 @@ package bitcask
 
 import (
 	"fmt"
+	"hash/crc32"
 	"io"
 	"io/ioutil"
 	"os"
 	"path/filepath"
 	"strings"
-	"time"
 
 	"github.com/gofrs/flock"
 	"github.com/prologic/trie"
@@ -16,7 +16,7 @@ import (
 type Bitcask struct {
 	*flock.Flock
 
-	opts      Options
+	config    *config
 	path      string
 	curr      *Datafile
 	keydir    *Keydir
@@ -56,28 +56,33 @@ func (b *Bitcask) Get(key string) ([]byte, error) {
 		df = b.datafiles[item.FileID]
 	}
 
-	e, err := df.ReadAt(item.Index)
+	e, err := df.ReadAt(item.Offset)
 	if err != nil {
 		return nil, err
 	}
 
+	checksum := crc32.ChecksumIEEE(e.Value)
+	if checksum != e.Checksum {
+		return nil, fmt.Errorf("error: checksum falied %s %d != %d", key, e.Checksum, checksum)
+	}
+
 	return e.Value, nil
 }
 
 func (b *Bitcask) Put(key string, value []byte) error {
-	if len(key) > b.opts.MaxKeySize {
-		return fmt.Errorf("error: key too large %d > %d", len(key), b.opts.MaxKeySize)
+	if len(key) > b.config.MaxKeySize {
+		return fmt.Errorf("error: key too large %d > %d", len(key), b.config.MaxKeySize)
 	}
-	if len(value) > b.opts.MaxValueSize {
-		return fmt.Errorf("error: value too large %d > %d", len(value), b.opts.MaxValueSize)
+	if len(value) > b.config.MaxValueSize {
+		return fmt.Errorf("error: value too large %d > %d", len(value), b.config.MaxValueSize)
 	}
 
-	index, err := b.put(key, value)
+	offset, err := b.put(key, value)
 	if err != nil {
 		return err
 	}
 
-	item := b.keydir.Add(key, b.curr.id, index, time.Now().Unix())
+	item := b.keydir.Add(key, b.curr.id, offset)
 	b.trie.Add(key, item)
 
 	return nil
@@ -127,6 +132,10 @@ func (b *Bitcask) put(key string, value []byte) (int64, error) {
 		}
 
 		df, err := NewDatafile(b.path, b.curr.id, true)
+		if err != nil {
+			return -1, err
+		}
+
 		b.datafiles = append(b.datafiles, df)
 
 		id := b.curr.id + 1
@@ -205,7 +214,7 @@ func Merge(path string, force bool) error {
 				continue
 			}
 
-			keydir.Add(e.Key, ids[i], e.Index, e.Timestamp)
+			keydir.Add(e.Key, ids[i], e.Offset)
 		}
 
 		tempdf, err := NewDatafile(temp, id, false)
@@ -216,7 +225,7 @@ func Merge(path string, force bool) error {
 
 		for key := range keydir.Keys() {
 			item, _ := keydir.Get(key)
-			e, err := df.ReadAt(item.Index)
+			e, err := df.ReadAt(item.Offset)
 			if err != nil {
 				return err
 			}
@@ -252,7 +261,7 @@ func Merge(path string, force bool) error {
 	return nil
 }
 
-func Open(path string, options ...func(*Bitcask) error) (*Bitcask, error) {
+func Open(path string, options ...option) (*Bitcask, error) {
 	if err := os.MkdirAll(path, 0755); err != nil {
 		return nil, err
 	}
@@ -298,7 +307,7 @@ func Open(path string, options ...func(*Bitcask) error) (*Bitcask, error) {
 
 			for key := range hint.Keys() {
 				item, _ := hint.Get(key)
-				_ = keydir.Add(key, item.FileID, item.Index, item.Timestamp)
+				_ = keydir.Add(key, item.FileID, item.Offset)
 				trie.Add(key, item)
 			}
 		} else {
@@ -317,7 +326,7 @@ func Open(path string, options ...func(*Bitcask) error) (*Bitcask, error) {
 					continue
 				}
 
-				item := keydir.Add(e.Key, ids[i], e.Index, e.Timestamp)
+				item := keydir.Add(e.Key, ids[i], e.Offset)
 				trie.Add(e.Key, item)
 			}
 		}
@@ -335,7 +344,7 @@ func Open(path string, options ...func(*Bitcask) error) (*Bitcask, error) {
 
 	bitcask := &Bitcask{
 		Flock:     flock.New(filepath.Join(path, "lock")),
-		opts:      NewDefaultOptions(),
+		config:    NewDefaultConfig(),
 		path:      path,
 		curr:      curr,
 		keydir:    keydir,
@@ -345,8 +354,8 @@ func Open(path string, options ...func(*Bitcask) error) (*Bitcask, error) {
 		maxDatafileSize: DefaultMaxDatafileSize,
 	}
 
-	for _, option := range options {
-		err = option(bitcask)
+	for _, opt := range options {
+		err = opt(bitcask.config)
 		if err != nil {
 			return nil, err
 		}

bitcask_test.go

@@ -287,10 +287,9 @@ func TestConcurrent(t *testing.T) {
 			wg := &sync.WaitGroup{}
 
 			go f(wg, 2)
-			wg.Add(1)
-
 			go f(wg, 3)
-			wg.Add(1)
+			go f(wg, 5)
+			wg.Add(3)
 
 			wg.Wait()
 		})
@@ -310,10 +309,9 @@ func TestConcurrent(t *testing.T) {
 			wg := &sync.WaitGroup{}
 
 			go f(wg, 100)
-			wg.Add(1)
-
 			go f(wg, 100)
-			wg.Add(1)
+			go f(wg, 100)
+			wg.Add(3)
 
 			wg.Wait()
 		})

datafile.go

@@ -5,7 +5,6 @@ import (
 	"os"
 	"path/filepath"
 	"sync"
-	"time"
 
 	"github.com/pkg/errors"
 
@@ -128,8 +127,7 @@ func (df *Datafile) Write(e pb.Entry) (int64, error) {
 	df.Lock()
 	defer df.Unlock()
 
-	e.Index = df.offset
-	e.Timestamp = time.Now().Unix()
+	e.Offset = df.offset
 
 	n, err := df.enc.Encode(&e)
 	if err != nil {
@@ -137,5 +135,5 @@ func (df *Datafile) Write(e pb.Entry) (int64, error) {
 	}
 	df.offset += n
 
-	return e.Index, nil
+	return e.Offset, nil
 }

entry.go

@@ -7,11 +7,11 @@ import (
 )
 
 func NewEntry(key string, value []byte) pb.Entry {
-	crc := crc32.ChecksumIEEE(value)
+	checksum := crc32.ChecksumIEEE(value)
 
 	return pb.Entry{
-		CRC:   crc,
-		Key:   key,
-		Value: value,
+		Checksum: checksum,
+		Key:      key,
+		Value:    value,
 	}
 }

keydir.go

@@ -9,9 +9,8 @@ import (
 )
 
 type Item struct {
-	FileID    int
-	Index     int64
-	Timestamp int64
+	FileID int
+	Offset int64
 }
 
 type Keydir struct {
@@ -25,11 +24,10 @@ func NewKeydir() *Keydir {
 	}
 }
 
-func (k *Keydir) Add(key string, fileid int, index, timestamp int64) Item {
+func (k *Keydir) Add(key string, fileid int, offset int64) Item {
 	item := Item{
-		FileID:    fileid,
-		Index:     index,
-		Timestamp: timestamp,
+		FileID: fileid,
+		Offset: offset,
 	}
 
 	k.Lock()

options.go

@@ -6,37 +6,42 @@ const (
 	DefaultMaxValueSize    = 1 << 16 // 65KB
 )
 
-type Options struct {
+// Option ...
+type Option option
+
+type option func(*config) error
+
+type config struct {
 	MaxDatafileSize int
 	MaxKeySize      int
 	MaxValueSize    int
 }
 
-func NewDefaultOptions() Options {
-	return Options{
+func NewDefaultConfig() *config {
+	return &config{
 		MaxDatafileSize: DefaultMaxDatafileSize,
 		MaxKeySize:      DefaultMaxKeySize,
 		MaxValueSize:    DefaultMaxValueSize,
 	}
 }
 
-func WithMaxDatafileSize(size int) func(*Bitcask) error {
-	return func(b *Bitcask) error {
-		b.opts.MaxDatafileSize = size
+func WithMaxDatafileSize(size int) option {
+	return func(cfg *config) error {
+		cfg.MaxDatafileSize = size
 		return nil
 	}
 }
 
-func WithMaxKeySize(size int) func(*Bitcask) error {
-	return func(b *Bitcask) error {
-		b.opts.MaxKeySize = size
+func WithMaxKeySize(size int) option {
+	return func(cfg *config) error {
+		cfg.MaxKeySize = size
 		return nil
 	}
 }
 
-func WithMaxValueSize(size int) func(*Bitcask) error {
-	return func(b *Bitcask) error {
-		b.opts.MaxValueSize = size
+func WithMaxValueSize(size int) option {
+	return func(cfg *config) error {
+		cfg.MaxValueSize = size
 		return nil
 	}
 }

proto/entry.pb.go

@@ -19,11 +19,10 @@ var _ = math.Inf
 const _ = proto.ProtoPackageIsVersion2 // please upgrade the proto package
 
 type Entry struct {
-	CRC                  uint32   `protobuf:"varint,1,opt,name=CRC,proto3" json:"CRC,omitempty"`
+	Checksum             uint32   `protobuf:"varint,1,opt,name=Checksum,proto3" json:"Checksum,omitempty"`
 	Key                  string   `protobuf:"bytes,2,opt,name=Key,proto3" json:"Key,omitempty"`
-	Index                int64    `protobuf:"varint,3,opt,name=Index,proto3" json:"Index,omitempty"`
+	Offset               int64    `protobuf:"varint,3,opt,name=Offset,proto3" json:"Offset,omitempty"`
 	Value                []byte   `protobuf:"bytes,4,opt,name=Value,proto3" json:"Value,omitempty"`
-	Timestamp            int64    `protobuf:"varint,5,opt,name=Timestamp,proto3" json:"Timestamp,omitempty"`
 	XXX_NoUnkeyedLiteral struct{} `json:"-"`
 	XXX_unrecognized     []byte   `json:"-"`
 	XXX_sizecache        int32    `json:"-"`
@@ -33,7 +32,7 @@ func (m *Entry) Reset()         { *m = Entry{} }
 func (m *Entry) String() string { return proto.CompactTextString(m) }
 func (*Entry) ProtoMessage()    {}
 func (*Entry) Descriptor() ([]byte, []int) {
-	return fileDescriptor_entry_4f5906245d08394f, []int{0}
+	return fileDescriptor_entry_3e91842c99935ae2, []int{0}
 }
 func (m *Entry) XXX_Unmarshal(b []byte) error {
 	return xxx_messageInfo_Entry.Unmarshal(m, b)
@@ -53,9 +52,9 @@ func (m *Entry) XXX_DiscardUnknown() {
 
 var xxx_messageInfo_Entry proto.InternalMessageInfo
 
-func (m *Entry) GetCRC() uint32 {
+func (m *Entry) GetChecksum() uint32 {
 	if m != nil {
-		return m.CRC
+		return m.Checksum
 	}
 	return 0
 }
@@ -67,9 +66,9 @@ func (m *Entry) GetKey() string {
 	return ""
 }
 
-func (m *Entry) GetIndex() int64 {
+func (m *Entry) GetOffset() int64 {
 	if m != nil {
-		return m.Index
+		return m.Offset
 	}
 	return 0
 }
@@ -81,28 +80,20 @@ func (m *Entry) GetValue() []byte {
 	return nil
 }
 
-func (m *Entry) GetTimestamp() int64 {
-	if m != nil {
-		return m.Timestamp
-	}
-	return 0
-}
-
 func init() {
 	proto.RegisterType((*Entry)(nil), "proto.Entry")
 }
 
-func init() { proto.RegisterFile("entry.proto", fileDescriptor_entry_4f5906245d08394f) }
+func init() { proto.RegisterFile("entry.proto", fileDescriptor_entry_3e91842c99935ae2) }
 
-var fileDescriptor_entry_4f5906245d08394f = []byte{
-	// 134 bytes of a gzipped FileDescriptorProto
+var fileDescriptor_entry_3e91842c99935ae2 = []byte{
+	// 126 bytes of a gzipped FileDescriptorProto
 	0x1f, 0x8b, 0x08, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0xff, 0xe2, 0xe2, 0x4e, 0xcd, 0x2b, 0x29,
-	0xaa, 0xd4, 0x2b, 0x28, 0xca, 0x2f, 0xc9, 0x17, 0x62, 0x05, 0x53, 0x4a, 0xa5, 0x5c, 0xac, 0xae,
-	0x20, 0x51, 0x21, 0x01, 0x2e, 0x66, 0xe7, 0x20, 0x67, 0x09, 0x46, 0x05, 0x46, 0x0d, 0xde, 0x20,
-	0x10, 0x13, 0x24, 0xe2, 0x9d, 0x5a, 0x29, 0xc1, 0xa4, 0xc0, 0xa8, 0xc1, 0x19, 0x04, 0x62, 0x0a,
-	0x89, 0x70, 0xb1, 0x7a, 0xe6, 0xa5, 0xa4, 0x56, 0x48, 0x30, 0x2b, 0x30, 0x6a, 0x30, 0x07, 0x41,
-	0x38, 0x20, 0xd1, 0xb0, 0xc4, 0x9c, 0xd2, 0x54, 0x09, 0x16, 0x05, 0x46, 0x0d, 0x9e, 0x20, 0x08,
-	0x47, 0x48, 0x86, 0x8b, 0x33, 0x24, 0x33, 0x37, 0xb5, 0xb8, 0x24, 0x31, 0xb7, 0x40, 0x82, 0x15,
-	0xac, 0x1e, 0x21, 0x90, 0xc4, 0x06, 0xb6, 0xdd, 0x18, 0x10, 0x00, 0x00, 0xff, 0xff, 0x07, 0x99,
-	0x47, 0xb9, 0x93, 0x00, 0x00, 0x00,
+	0xaa, 0xd4, 0x2b, 0x28, 0xca, 0x2f, 0xc9, 0x17, 0x62, 0x05, 0x53, 0x4a, 0xc9, 0x5c, 0xac, 0xae,
+	0x20, 0x51, 0x21, 0x29, 0x2e, 0x0e, 0xe7, 0x8c, 0xd4, 0xe4, 0xec, 0xe2, 0xd2, 0x5c, 0x09, 0x46,
+	0x05, 0x46, 0x0d, 0xde, 0x20, 0x38, 0x5f, 0x48, 0x80, 0x8b, 0xd9, 0x3b, 0xb5, 0x52, 0x82, 0x49,
+	0x81, 0x51, 0x83, 0x33, 0x08, 0xc4, 0x14, 0x12, 0xe3, 0x62, 0xf3, 0x4f, 0x4b, 0x2b, 0x4e, 0x2d,
+	0x91, 0x60, 0x56, 0x60, 0xd4, 0x60, 0x0e, 0x82, 0xf2, 0x84, 0x44, 0xb8, 0x58, 0xc3, 0x12, 0x73,
+	0x4a, 0x53, 0x25, 0x58, 0x14, 0x18, 0x35, 0x78, 0x82, 0x20, 0x9c, 0x24, 0x36, 0xb0, 0x5d, 0xc6,
+	0x80, 0x00, 0x00, 0x00, 0xff, 0xff, 0x76, 0xd2, 0x3e, 0x83, 0x81, 0x00, 0x00, 0x00,
 }

proto/entry.proto

@@ -3,9 +3,8 @@ syntax = "proto3";
 package proto;
 
 message Entry {
-	uint32 CRC = 1;
+	uint32 Checksum = 1;
 	string Key = 2;
-	int64 Index = 3;
+	int64 Offset = 3;
 	bytes Value = 4;
-	int64 Timestamp = 5;
 }