Add all files again with v2 postfix to URL

v2/internal/index/codec_index.go

@@ -0,0 +1,134 @@
+package index
+
+import (
+	"encoding/binary"
+	"io"
+
+	"git.mills.io/prologic/bitcask/v2/internal"
+	"github.com/pkg/errors"
+	art "github.com/plar/go-adaptive-radix-tree"
+)
+
+var (
+	errTruncatedKeySize = errors.New("key size is truncated")
+	errTruncatedKeyData = errors.New("key data is truncated")
+	errTruncatedData    = errors.New("data is truncated")
+	errKeySizeTooLarge  = errors.New("key size too large")
+)
+
+const (
+	int32Size  = 4
+	int64Size  = 8
+	fileIDSize = int32Size
+	offsetSize = int64Size
+	sizeSize   = int64Size
+)
+
+func readKeyBytes(r io.Reader, maxKeySize uint32) ([]byte, error) {
+	s := make([]byte, int32Size)
+	_, err := io.ReadFull(r, s)
+	if err != nil {
+		if err == io.EOF {
+			return nil, err
+		}
+		return nil, errors.Wrap(errTruncatedKeySize, err.Error())
+	}
+	size := binary.BigEndian.Uint32(s)
+	if maxKeySize > 0 && size > uint32(maxKeySize) {
+		return nil, errKeySizeTooLarge
+	}
+
+	b := make([]byte, size)
+	_, err = io.ReadFull(r, b)
+	if err != nil {
+		return nil, errors.Wrap(errTruncatedKeyData, err.Error())
+	}
+	return b, nil
+}
+
+func writeBytes(b []byte, w io.Writer) error {
+	s := make([]byte, int32Size)
+	binary.BigEndian.PutUint32(s, uint32(len(b)))
+	_, err := w.Write(s)
+	if err != nil {
+		return err
+	}
+	_, err = w.Write(b)
+	if err != nil {
+		return err
+	}
+	return nil
+}
+
+func readItem(r io.Reader) (internal.Item, error) {
+	buf := make([]byte, (fileIDSize + offsetSize + sizeSize))
+	_, err := io.ReadFull(r, buf)
+	if err != nil {
+		return internal.Item{}, errors.Wrap(errTruncatedData, err.Error())
+	}
+
+	return internal.Item{
+		FileID: int(binary.BigEndian.Uint32(buf[:fileIDSize])),
+		Offset: int64(binary.BigEndian.Uint64(buf[fileIDSize:(fileIDSize + offsetSize)])),
+		Size:   int64(binary.BigEndian.Uint64(buf[(fileIDSize + offsetSize):])),
+	}, nil
+}
+
+func writeItem(item internal.Item, w io.Writer) error {
+	buf := make([]byte, (fileIDSize + offsetSize + sizeSize))
+	binary.BigEndian.PutUint32(buf[:fileIDSize], uint32(item.FileID))
+	binary.BigEndian.PutUint64(buf[fileIDSize:(fileIDSize+offsetSize)], uint64(item.Offset))
+	binary.BigEndian.PutUint64(buf[(fileIDSize+offsetSize):], uint64(item.Size))
+	_, err := w.Write(buf)
+	if err != nil {
+		return err
+	}
+	return nil
+}
+
+// ReadIndex reads a persisted from a io.Reader into a Tree
+func readIndex(r io.Reader, t art.Tree, maxKeySize uint32) error {
+	for {
+		key, err := readKeyBytes(r, maxKeySize)
+		if err != nil {
+			if err == io.EOF {
+				break
+			}
+			return err
+		}
+
+		item, err := readItem(r)
+		if err != nil {
+			return err
+		}
+
+		t.Insert(key, item)
+	}
+
+	return nil
+}
+
+func writeIndex(t art.Tree, w io.Writer) (err error) {
+	t.ForEach(func(node art.Node) bool {
+		err = writeBytes(node.Key(), w)
+		if err != nil {
+			return false
+		}
+
+		item := node.Value().(internal.Item)
+		err := writeItem(item, w)
+		return err == nil
+	})
+	return
+}
+
+// IsIndexCorruption returns a boolean indicating whether the error
+// is known to report a corruption data issue
+func IsIndexCorruption(err error) bool {
+	cause := errors.Cause(err)
+	switch cause {
+	case errKeySizeTooLarge, errTruncatedData, errTruncatedKeyData, errTruncatedKeySize:
+		return true
+	}
+	return false
+}

v2/internal/index/codec_index_test.go

@@ -0,0 +1,126 @@
+package index
+
+import (
+	"bytes"
+	"encoding/base64"
+	"encoding/binary"
+	"testing"
+
+	"git.mills.io/prologic/bitcask/v2/internal"
+	"github.com/pkg/errors"
+	art "github.com/plar/go-adaptive-radix-tree"
+)
+
+const (
+	base64SampleTree = "AAAABGFiY2QAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAARhYmNlAAAAAQAAAAAAAAABAAAAAAAAAAEAAAAEYWJjZgAAAAIAAAAAAAAAAgAAAAAAAAACAAAABGFiZ2QAAAADAAAAAAAAAAMAAAAAAAAAAw=="
+)
+
+func TestWriteIndex(t *testing.T) {
+	at, expectedSerializedSize := getSampleTree()
+
+	var b bytes.Buffer
+	err := writeIndex(at, &b)
+	if err != nil {
+		t.Fatalf("writing index failed: %v", err)
+	}
+	if b.Len() != expectedSerializedSize {
+		t.Fatalf("incorrect size of serialied index: expected %d, got: %d", expectedSerializedSize, b.Len())
+	}
+	sampleTreeBytes, _ := base64.StdEncoding.DecodeString(base64SampleTree)
+	if !bytes.Equal(b.Bytes(), sampleTreeBytes) {
+		t.Fatalf("unexpected serialization of the tree")
+	}
+}
+
+func TestReadIndex(t *testing.T) {
+	sampleTreeBytes, _ := base64.StdEncoding.DecodeString(base64SampleTree)
+	b := bytes.NewBuffer(sampleTreeBytes)
+
+	at := art.New()
+	err := readIndex(b, at, 1024)
+	if err != nil {
+		t.Fatalf("error while deserializing correct sample tree: %v", err)
+	}
+
+	atsample, _ := getSampleTree()
+	if atsample.Size() != at.Size() {
+		t.Fatalf("trees aren't the same size, expected %v, got %v", atsample.Size(), at.Size())
+	}
+	atsample.ForEach(func(node art.Node) bool {
+		_, found := at.Search(node.Key())
+		if !found {
+			t.Fatalf("expected node wasn't found: %s", node.Key())
+		}
+		return true
+	})
+}
+
+func TestReadCorruptedData(t *testing.T) {
+	sampleBytes, _ := base64.StdEncoding.DecodeString(base64SampleTree)
+
+	t.Run("truncated", func(t *testing.T) {
+		table := []struct {
+			name string
+			err  error
+			data []byte
+		}{
+			{name: "key-size-first-item", err: errTruncatedKeySize, data: sampleBytes[:2]},
+			{name: "key-data-second-item", err: errTruncatedKeyData, data: sampleBytes[:6]},
+			{name: "key-size-second-item", err: errTruncatedKeySize, data: sampleBytes[:(int32Size+4+fileIDSize+offsetSize+sizeSize)+2]},
+			{name: "key-data-second-item", err: errTruncatedKeyData, data: sampleBytes[:(int32Size+4+fileIDSize+offsetSize+sizeSize)+6]},
+			{name: "data", err: errTruncatedData, data: sampleBytes[:int32Size+4+(fileIDSize+offsetSize+sizeSize-3)]},
+		}
+
+		for i := range table {
+			t.Run(table[i].name, func(t *testing.T) {
+				bf := bytes.NewBuffer(table[i].data)
+
+				if err := readIndex(bf, art.New(), 1024); !IsIndexCorruption(err) || errors.Cause(err) != table[i].err {
+					t.Fatalf("expected %v, got %v", table[i].err, err)
+				}
+			})
+		}
+	})
+
+	t.Run("overflow", func(t *testing.T) {
+		overflowKeySize := make([]byte, len(sampleBytes))
+		copy(overflowKeySize, sampleBytes)
+		binary.BigEndian.PutUint32(overflowKeySize, 1025)
+
+		overflowDataSize := make([]byte, len(sampleBytes))
+		copy(overflowDataSize, sampleBytes)
+		binary.BigEndian.PutUint32(overflowDataSize[int32Size+4+fileIDSize+offsetSize:], 1025)
+
+		table := []struct {
+			name       string
+			err        error
+			maxKeySize uint32
+			data       []byte
+		}{
+			{name: "key-data-overflow", err: errKeySizeTooLarge, maxKeySize: 1024, data: overflowKeySize},
+		}
+
+		for i := range table {
+			t.Run(table[i].name, func(t *testing.T) {
+				bf := bytes.NewBuffer(table[i].data)
+
+				if err := readIndex(bf, art.New(), table[i].maxKeySize); !IsIndexCorruption(err) || errors.Cause(err) != table[i].err {
+					t.Fatalf("expected %v, got %v", table[i].err, err)
+				}
+			})
+		}
+	})
+
+}
+
+func getSampleTree() (art.Tree, int) {
+	at := art.New()
+	keys := [][]byte{[]byte("abcd"), []byte("abce"), []byte("abcf"), []byte("abgd")}
+	expectedSerializedSize := 0
+	for i := range keys {
+		at.Insert(keys[i], internal.Item{FileID: i, Offset: int64(i), Size: int64(i)})
+		expectedSerializedSize += int32Size + len(keys[i]) + fileIDSize + offsetSize + sizeSize
+	}
+
+	return at, expectedSerializedSize
+}

v2/internal/index/index.go

@@ -0,0 +1,59 @@
+package index
+
+import (
+	"os"
+
+	"git.mills.io/prologic/bitcask/v2/internal"
+	art "github.com/plar/go-adaptive-radix-tree"
+)
+
+// Indexer is an interface for loading and saving the index (an Adaptive Radix Tree)
+type Indexer interface {
+	Load(path string, maxkeySize uint32) (art.Tree, bool, error)
+	Save(t art.Tree, path string) error
+}
+
+// NewIndexer returns an instance of the default `Indexer` implemtnation
+// which perists the index (an Adaptive Radix Tree) as a binary blob on file
+func NewIndexer() Indexer {
+	return &indexer{}
+}
+
+type indexer struct{}
+
+func (i *indexer) Load(path string, maxKeySize uint32) (art.Tree, bool, error) {
+	t := art.New()
+
+	if !internal.Exists(path) {
+		return t, false, nil
+	}
+
+	f, err := os.Open(path)
+	if err != nil {
+		return t, true, err
+	}
+	defer f.Close()
+
+	if err := readIndex(f, t, maxKeySize); err != nil {
+		return t, true, err
+	}
+	return t, true, nil
+}
+
+func (i *indexer) Save(t art.Tree, path string) error {
+	f, err := os.OpenFile(path, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, 0600)
+	if err != nil {
+		return err
+	}
+	defer f.Close()
+
+	if err := writeIndex(t, f); err != nil {
+		return err
+	}
+
+	if err := f.Sync(); err != nil {
+		return err
+	}
+
+	return f.Close()
+}

v2/internal/index/ttl_index.go

@@ -0,0 +1,71 @@
+package index
+
+import (
+	"encoding/binary"
+	"io"
+	"os"
+	"time"
+
+	"git.mills.io/prologic/bitcask/v2/internal"
+	art "github.com/plar/go-adaptive-radix-tree"
+)
+
+type ttlIndexer struct{}
+
+func NewTTLIndexer() Indexer {
+	return ttlIndexer{}
+}
+
+func (i ttlIndexer) Save(t art.Tree, path string) error {
+	f, err := os.OpenFile(path, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, 0600)
+	if err != nil {
+		return err
+	}
+	buf := make([]byte, int64Size)
+	for it := t.Iterator(); it.HasNext(); {
+		node, err := it.Next()
+		if err != nil {
+			return err
+		}
+		// save key
+		err = writeBytes(node.Key(), f)
+		if err != nil {
+			return err
+		}
+		// save key ttl
+		binary.BigEndian.PutUint64(buf, uint64(node.Value().(time.Time).Unix()))
+		_, err = f.Write(buf)
+		if err != nil {
+			return err
+		}
+	}
+	return f.Sync()
+}
+
+func (i ttlIndexer) Load(path string, maxKeySize uint32) (art.Tree, bool, error) {
+	t := art.New()
+	if !internal.Exists(path) {
+		return t, false, nil
+	}
+	f, err := os.Open(path)
+	if err != nil {
+		return t, true, err
+	}
+	buf := make([]byte, int64Size)
+	for {
+		key, err := readKeyBytes(f, maxKeySize)
+		if err != nil {
+			if err == io.EOF {
+				break
+			}
+			return t, true, err
+		}
+		_, err = io.ReadFull(f, buf)
+		if err != nil {
+			return t, true, err
+		}
+		expiry := time.Unix(int64(binary.BigEndian.Uint64(buf)), 0).UTC()
+		t.Insert(key, expiry)
+	}
+	return t, true, nil
+}

v2/internal/index/ttl_index_test.go

@@ -0,0 +1,54 @@
+package index
+
+import (
+	"io/ioutil"
+	"os"
+	"path/filepath"
+	"testing"
+	"time"
+
+	art "github.com/plar/go-adaptive-radix-tree"
+	assert2 "github.com/stretchr/testify/assert"
+)
+
+func Test_TTLIndexer(t *testing.T) {
+	assert := assert2.New(t)
+	tempDir, err := ioutil.TempDir("", "bitcask")
+	assert.NoError(err)
+	defer os.RemoveAll(tempDir)
+
+	currTime := time.Date(2020, 12, 27, 0, 0, 0, 0, time.UTC)
+	trie := art.New()
+
+	t.Run("LoadEmpty", func(t *testing.T) {
+		newTrie, found, err := NewTTLIndexer().Load(filepath.Join(tempDir, "ttl_index"), 4)
+		assert.NoError(err)
+		assert.False(found)
+		assert.Equal(trie, newTrie)
+	})
+
+	t.Run("Save", func(t *testing.T) {
+		trie.Insert([]byte("key"), currTime)
+		err := NewTTLIndexer().Save(trie, filepath.Join(tempDir, "ttl_index"))
+		assert.NoError(err)
+		trie.Insert([]byte("foo"), currTime.Add(24*time.Hour))
+		err = NewTTLIndexer().Save(trie, filepath.Join(tempDir, "ttl_index"))
+		assert.NoError(err)
+		trie.Insert([]byte("key"), currTime.Add(-24*time.Hour))
+		err = NewTTLIndexer().Save(trie, filepath.Join(tempDir, "ttl_index"))
+		assert.NoError(err)
+	})
+
+	t.Run("Load", func(t *testing.T) {
+		newTrie, found, err := NewTTLIndexer().Load(filepath.Join(tempDir, "ttl_index"), 4)
+		assert.NoError(err)
+		assert.True(found)
+		assert.Equal(2, newTrie.Size())
+		value, found := newTrie.Search([]byte("key"))
+		assert.True(found)
+		assert.Equal(currTime.Add(-24*time.Hour), value)
+		value, found = newTrie.Search([]byte("foo"))
+		assert.True(found)
+		assert.Equal(currTime.Add(24*time.Hour), value)
+	})
+}