Refactor and general tests for codec index (#83)
* codec_index: unexport const fields * codec_index: unexport internal functions and doc exported ones * codec_index: rename func & return errors for corruption * codec_index: new test for ReadIndex, WriteIndex, and read corruption * Update internal/codec_index.go Co-Authored-By: James Mills <1290234+prologic@users.noreply.github.com> * Update internal/codec_index.go Co-Authored-By: James Mills <1290234+prologic@users.noreply.github.com>
This commit is contained in:
parent
50d3971e86
commit
8041a4c1e7
@ -4,34 +4,44 @@ import (
|
||||
"encoding/binary"
|
||||
"io"
|
||||
|
||||
"github.com/pkg/errors"
|
||||
art "github.com/plar/go-adaptive-radix-tree"
|
||||
)
|
||||
|
||||
const (
|
||||
Int32Size = 4
|
||||
Int64Size = 8
|
||||
FileIDSize = Int32Size
|
||||
OffsetSize = Int64Size
|
||||
SizeSize = Int64Size
|
||||
var (
|
||||
errTruncatedKeySize = errors.New("key size is truncated")
|
||||
errTruncatedKeyData = errors.New("key data is truncated")
|
||||
errTruncatedData = errors.New("data is truncated")
|
||||
)
|
||||
|
||||
func ReadBytes(r io.Reader) ([]byte, error) {
|
||||
s := make([]byte, Int32Size)
|
||||
const (
|
||||
int32Size = 4
|
||||
int64Size = 8
|
||||
fileIDSize = int32Size
|
||||
offsetSize = int64Size
|
||||
sizeSize = int64Size
|
||||
)
|
||||
|
||||
func readKeyBytes(r io.Reader) ([]byte, error) {
|
||||
s := make([]byte, int32Size)
|
||||
_, err := io.ReadFull(r, s)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
if err == io.EOF {
|
||||
return nil, err
|
||||
}
|
||||
return nil, errors.Wrap(errTruncatedKeySize, err.Error())
|
||||
}
|
||||
size := binary.BigEndian.Uint32(s)
|
||||
b := make([]byte, size)
|
||||
_, err = io.ReadFull(r, b)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
return nil, errors.Wrap(errTruncatedKeyData, err.Error())
|
||||
}
|
||||
return b, nil
|
||||
}
|
||||
|
||||
func WriteBytes(b []byte, w io.Writer) (int, error) {
|
||||
s := make([]byte, Int32Size)
|
||||
func writeBytes(b []byte, w io.Writer) (int, error) {
|
||||
s := make([]byte, int32Size)
|
||||
binary.BigEndian.PutUint32(s, uint32(len(b)))
|
||||
n, err := w.Write(s)
|
||||
if err != nil {
|
||||
@ -44,25 +54,25 @@ func WriteBytes(b []byte, w io.Writer) (int, error) {
|
||||
return (n + m), nil
|
||||
}
|
||||
|
||||
func ReadItem(r io.Reader) (Item, error) {
|
||||
buf := make([]byte, (FileIDSize + OffsetSize + SizeSize))
|
||||
func readItem(r io.Reader) (Item, error) {
|
||||
buf := make([]byte, (fileIDSize + offsetSize + sizeSize))
|
||||
_, err := io.ReadFull(r, buf)
|
||||
if err != nil {
|
||||
return Item{}, err
|
||||
return Item{}, errors.Wrap(errTruncatedData, err.Error())
|
||||
}
|
||||
|
||||
return 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):])),
|
||||
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 Item, w io.Writer) (int, 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))
|
||||
func writeItem(item Item, w io.Writer) (int, 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))
|
||||
n, err := w.Write(buf)
|
||||
if err != nil {
|
||||
return 0, err
|
||||
@ -70,9 +80,10 @@ func WriteItem(item Item, w io.Writer) (int, error) {
|
||||
return n, nil
|
||||
}
|
||||
|
||||
// ReadIndex reads a persisted tree from a io.Reader into a Tree
|
||||
func ReadIndex(r io.Reader, t art.Tree) error {
|
||||
for {
|
||||
key, err := ReadBytes(r)
|
||||
key, err := readKeyBytes(r)
|
||||
if err != nil {
|
||||
if err == io.EOF {
|
||||
break
|
||||
@ -80,7 +91,7 @@ func ReadIndex(r io.Reader, t art.Tree) error {
|
||||
return err
|
||||
}
|
||||
|
||||
item, err := ReadItem(r)
|
||||
item, err := readItem(r)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
@ -91,15 +102,16 @@ func ReadIndex(r io.Reader, t art.Tree) error {
|
||||
return nil
|
||||
}
|
||||
|
||||
// WriteIndex persists a Tree into a io.Writer
|
||||
func WriteIndex(t art.Tree, w io.Writer) (err error) {
|
||||
t.ForEach(func(node art.Node) bool {
|
||||
_, err = WriteBytes(node.Key(), w)
|
||||
_, err = writeBytes(node.Key(), w)
|
||||
if err != nil {
|
||||
return false
|
||||
}
|
||||
|
||||
item := node.Value().(Item)
|
||||
_, err := WriteItem(item, w)
|
||||
_, err := writeItem(item, w)
|
||||
if err != nil {
|
||||
return false
|
||||
}
|
||||
|
91
internal/codec_index_test.go
Normal file
91
internal/codec_index_test.go
Normal file
@ -0,0 +1,91 @@
|
||||
package internal
|
||||
|
||||
import (
|
||||
"bytes"
|
||||
"encoding/base64"
|
||||
"testing"
|
||||
|
||||
"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)
|
||||
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)
|
||||
table := []struct {
|
||||
name string
|
||||
err error
|
||||
data []byte
|
||||
}{
|
||||
{name: "truncated-key-size-first-item", err: errTruncatedKeySize, data: sampleBytes[:2]},
|
||||
{name: "truncated-key-data-second-item", err: errTruncatedKeyData, data: sampleBytes[:6]},
|
||||
{name: "truncated-key-size-second-item", err: errTruncatedKeySize, data: sampleBytes[:(int32Size+4+fileIDSize+offsetSize+sizeSize)+2]},
|
||||
{name: "truncated-key-data-second-item", err: errTruncatedKeyData, data: sampleBytes[:(int32Size+4+fileIDSize+offsetSize+sizeSize)+6]},
|
||||
{name: "truncated-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()); 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], Item{FileID: i, Offset: int64(i), Size: int64(i)})
|
||||
expectedSerializedSize += int32Size + len(keys[i]) + fileIDSize + offsetSize + sizeSize
|
||||
}
|
||||
|
||||
return at, expectedSerializedSize
|
||||
}
|
Loading…
Reference in New Issue
Block a user