From 981c981fa8862c34f0a56ec2823f66a029c17826 Mon Sep 17 00:00:00 2001
From: chrislusf <chris.lu@gmail.com>
Date: Tue, 17 Mar 2015 11:15:37 -0700
Subject: [PATCH] one-off fix for issues/93

https://github.com/chrislusf/weed-fs/issues/93
---
 unmaintained/README.txt         |   9 +++
 unmaintained/fix_dat/fix_dat.go | 130 ++++++++++++++++++++++++++++++++
 2 files changed, 139 insertions(+)
 create mode 100644 unmaintained/README.txt
 create mode 100644 unmaintained/fix_dat/fix_dat.go

diff --git a/unmaintained/README.txt b/unmaintained/README.txt
new file mode 100644
index 000000000..f34136959
--- /dev/null
+++ b/unmaintained/README.txt
@@ -0,0 +1,9 @@
+Here is the folder of unmaintained go code.
+
+Although its content is useful sometimes, 
+Its content would not be released.
+And the code can be ugly, e.g., not checking exceptions.
+
+If any code is used often, it will be productionized.
+So it is helpful if anyone wants to cleanup a bit.
+
diff --git a/unmaintained/fix_dat/fix_dat.go b/unmaintained/fix_dat/fix_dat.go
new file mode 100644
index 000000000..486da4a78
--- /dev/null
+++ b/unmaintained/fix_dat/fix_dat.go
@@ -0,0 +1,130 @@
+package main
+
+import (
+	"flag"
+	"fmt"
+	"io"
+	"os"
+	"path"
+	"strconv"
+
+	"github.com/chrislusf/weed-fs/go/glog"
+	"github.com/chrislusf/weed-fs/go/storage"
+	"github.com/chrislusf/weed-fs/go/util"
+)
+
+var (
+	fixVolumePath       = flag.String("dir", "/tmp", "data directory to store files")
+	fixVolumeCollection = flag.String("collection", "", "the volume collection name")
+	fixVolumeId         = flag.Int("volumeId", -1, "a volume id. The volume should already exist in the dir. The volume index file should not exist.")
+)
+
+/*
+This is to resolve an one-time issue that caused inconsistency with .dat and .idx files.
+1. fix the .dat file, a new .dat_fixed file will be generated.
+	go run fix_dat.go -volumeId=9 -dir=/Users/chrislu/Downloads
+2. move the original .dat and .idx files to some backup folder, and rename .dat_fixed to .dat file
+    mv 9.dat_fixed 9.dat
+3. fix the .idx file with the "weed fix"
+	weed fix -volumeId=9 -dir=/Users/chrislu/Downloads
+*/
+func main() {
+	flag.Parse()
+	fileName := strconv.Itoa(*fixVolumeId)
+	if *fixVolumeCollection != "" {
+		fileName = *fixVolumeCollection + "_" + fileName
+	}
+	indexFile, err := os.OpenFile(path.Join(*fixVolumePath, fileName+".idx"), os.O_RDONLY, 0644)
+	if err != nil {
+		glog.Fatalf("Read Volume Index [ERROR] %s\n", err)
+	}
+	defer indexFile.Close()
+	datFile, err := os.OpenFile(path.Join(*fixVolumePath, fileName+".dat"), os.O_RDONLY, 0644)
+	if err != nil {
+		glog.Fatalf("Read Volume Data [ERROR] %s\n", err)
+	}
+	defer datFile.Close()
+
+	newDatFile, err := os.Create(path.Join(*fixVolumePath, fileName+".dat_fixed"))
+	if err != nil {
+		glog.Fatalf("Write New Volume Data [ERROR] %s\n", err)
+	}
+	defer newDatFile.Close()
+
+	header := make([]byte, storage.SuperBlockSize)
+	datFile.Read(header)
+	newDatFile.Write(header)
+
+	iterateEntries(datFile, indexFile, func(n *storage.Needle, offset int64) {
+		fmt.Printf("file id=%d name=%s size=%d dataSize=%d\n", n.Id, string(n.Name), n.Size, n.DataSize)
+		s, e := n.Append(newDatFile, storage.Version2)
+		fmt.Printf("size %d error %v\n", s, e)
+	})
+
+}
+
+func iterateEntries(datFile, idxFile *os.File, visitNeedle func(n *storage.Needle, offset int64)) {
+	// start to read index file
+	var readerOffset int64
+	bytes := make([]byte, 16)
+	count, _ := idxFile.ReadAt(bytes, readerOffset)
+	readerOffset += int64(count)
+
+	// start to read dat file
+	offset := int64(storage.SuperBlockSize)
+	version := storage.Version2
+	n, rest, err := storage.ReadNeedleHeader(datFile, version, offset)
+	if err != nil {
+		fmt.Printf("cannot read needle header: %v", err)
+		return
+	}
+	fmt.Printf("Needle %+v, rest %d\n", n, rest)
+	for n != nil && count > 0 {
+		// parse index file entry
+		key := util.BytesToUint64(bytes[0:8])
+		offsetFromIndex := util.BytesToUint32(bytes[8:12])
+		sizeFromIndex := util.BytesToUint32(bytes[12:16])
+		count, _ = idxFile.ReadAt(bytes, readerOffset)
+		readerOffset += int64(count)
+
+		if offsetFromIndex != 0 && offset != int64(offsetFromIndex)*8 {
+			//t := offset
+			offset = int64(offsetFromIndex) * 8
+			//fmt.Printf("Offset change %d => %d\n", t, offset)
+		}
+
+		fmt.Printf("key: %d offsetFromIndex %d n.Size %d sizeFromIndex:%d\n", key, offsetFromIndex, n.Size, sizeFromIndex)
+
+		padding := storage.NeedlePaddingSize - ((sizeFromIndex + storage.NeedleHeaderSize + storage.NeedleChecksumSize) % storage.NeedlePaddingSize)
+		rest = sizeFromIndex + storage.NeedleChecksumSize + padding
+
+		func() {
+			defer func() {
+				if r := recover(); r != nil {
+					fmt.Println("Recovered in f", r)
+				}
+			}()
+			if err = n.ReadNeedleBody(datFile, version, offset+int64(storage.NeedleHeaderSize), rest); err != nil {
+				fmt.Printf("cannot read needle body: offset %d body %d %v\n", offset, rest, err)
+			}
+		}()
+
+		if n.Size <= n.DataSize {
+			continue
+		}
+		visitNeedle(n, offset)
+
+		offset += int64(storage.NeedleHeaderSize) + int64(rest)
+		//fmt.Printf("==> new entry offset %d\n", offset)
+		if n, rest, err = storage.ReadNeedleHeader(datFile, version, offset); err != nil {
+			if err == io.EOF {
+				return
+			}
+
+			fmt.Printf("cannot read needle header: %v\n", err)
+			return
+		}
+		//fmt.Printf("new entry needle size:%d rest:%d\n", n.Size, rest)
+	}
+
+}