Auto recovery (#153)

* implement autorepair

Signed-off-by: Ignacio Hagopian <jsign.uy@gmail.com>

* fix misspell

Signed-off-by: Ignacio Hagopian <jsign.uy@gmail.com>

* Update internal/data/recover.go

Co-authored-by: James Mills <prologic@shortcircuit.net.au>

* Update internal/utils.go

Co-authored-by: James Mills <prologic@shortcircuit.net.au>

* Update internal/data/recover.go

Co-authored-by: James Mills <prologic@shortcircuit.net.au>

* skip failing test on windows

Signed-off-by: Ignacio Hagopian <jsign.uy@gmail.com>

Co-authored-by: James Mills <prologic@shortcircuit.net.au>

internal/data/recover.go

@@ -0,0 +1,89 @@
+package data
+
+import (
+	"fmt"
+	"io"
+	"os"
+	"path/filepath"
+
+	"github.com/prologic/bitcask/internal"
+	"github.com/prologic/bitcask/internal/config"
+	"github.com/prologic/bitcask/internal/data/codec"
+)
+
+// CheckAndRecover checks and recovers the last datafile.
+// If the datafile isn't corrupted, this is a noop. If it is,
+// the longest non-corrupted prefix will be kept and the rest
+// will be *deleted*. Also, the index file is also *deleted* which
+// will be automatically recreated on next startup.
+func CheckAndRecover(path string, cfg *config.Config) error {
+	dfs, err := internal.GetDatafiles(path)
+	if err != nil {
+		return fmt.Errorf("scanning datafiles: %s", err)
+	}
+	if len(dfs) == 0 {
+		return nil
+	}
+	f := dfs[len(dfs)-1]
+	recovered, err := recoverDatafile(f, cfg)
+	if err != nil {
+		return fmt.Errorf("recovering data file")
+	}
+	if recovered {
+		if err := os.Remove(filepath.Join(path, "index")); err != nil {
+			return fmt.Errorf("error deleting the index on recovery: %s", err)
+		}
+	}
+	return nil
+}
+
+func recoverDatafile(path string, cfg *config.Config) (recovered bool, err error) {
+	f, err := os.Open(path)
+	if err != nil {
+		return false, fmt.Errorf("opening the datafile: %s", err)
+	}
+	defer func() {
+		err = f.Close()
+	}()
+	_, file := filepath.Split(path)
+	rPath := fmt.Sprintf("%s.recovered", file)
+	fr, err := os.OpenFile(rPath, os.O_CREATE|os.O_WRONLY, os.ModePerm)
+	if err != nil {
+		return false, fmt.Errorf("creating the recovered datafile: %w", err)
+	}
+	defer func() {
+		err = fr.Close()
+	}()
+
+	dec := codec.NewDecoder(f, cfg.MaxKeySize, cfg.MaxValueSize)
+	enc := codec.NewEncoder(fr)
+	e := internal.Entry{}
+
+	corrupted := false
+	for !corrupted {
+		_, err = dec.Decode(&e)
+		if err == io.EOF {
+			break
+		}
+		if codec.IsCorruptedData(err) {
+			corrupted = true
+			continue
+		}
+		if err != nil {
+			return false, fmt.Errorf("unexpected error while reading datafile: %w", err)
+		}
+		if _, err := enc.Encode(e); err != nil {
+			return false, fmt.Errorf("writing to recovered datafile: %w", err)
+		}
+	}
+	if !corrupted {
+		if err := os.Remove(fr.Name()); err != nil {
+			return false, fmt.Errorf("can't remove temporal recovered datafile: %w", err)
+		}
+		return false, nil
+	}
+	if err := os.Rename(rPath, path); err != nil {
+		return false, fmt.Errorf("removing corrupted file: %s", err)
+	}
+	return true, nil
+}