Begin MBR parsing

Author: Philipp Heckel

.gitignore

@@ -1 +1,2 @@
 .idea
+build

Makefile

@@ -1,4 +1,7 @@
-all:
+all: clean
+	mkdir -p build
 	protoc --go_out=. internal/*.proto
-	go build
-	
+	go build -o build/fsdup
+
+clean:
+	rm -rf build

fsdup.go

@@ -6,6 +6,11 @@ import (
 	"os"
 )
 
+// TODO [HIGH] Support for MBR/GPT partition tables
+// TODO [HIGH] Support for any file
+// TODO [MED] Make manifest creation the last step (or part of all steps)
+// TODO [LOW] Sparsify all runs automatically
+
 func exit(code int, message string) {
 	fmt.Println(message)
 	os.Exit(code)

index.go

@@ -1,12 +1,38 @@
 package main
 
 import (
+	"bytes"
 	"github.com/golang/protobuf/proto"
+	"heckel.io/fsdup/internal"
+	"io"
 	"io/ioutil"
 	"log"
 	"os"
 )
 
+type fileType int
+
+const (
+	typeNtfs fileType = iota + 1
+	typeMbrDisk
+	typeUnknown
+)
+
+const (
+	probeTypeBufferLength = 512
+
+	mbrLength                         = 512
+	mbrSectorSize                     = 512
+	mbrSignatureOffset                = 510
+	mbrSignatureLength                = 2
+	mbrSignatureMagic                 = 0xaa55 // 0x55aa as little endian
+	mbrEntryCount                     = 4
+	mbrEntryFirstOffset               = 446
+	mbrEntryLength                    = 16
+	mbrFirstSectorRelativeOffset      = 8
+	mbrEntryFirstSectorRelativeLength = 4
+)
+
 func index(inputFile string, manifestFile string, offset int64, nowrite bool, exact bool) error {
 	file, err := os.Open(inputFile)
 	if err != nil {
@@ -15,13 +41,20 @@ func index(inputFile string, manifestFile string, offset int64, nowrite bool, ex
 
 	defer file.Close()
 
-	// Determine file type (partition, NTFS, other)
-	// ...
+	manifest := &internal.ManifestV1{}
 
-	ntfs := NewNtfsDeduper(file, offset, nowrite, exact)
-	manifest, err := ntfs.Dedup()
+	fileType, err := probeType(file, offset)
 	if err != nil {
-		log.Fatalln("Failed to dedup:", err)
+		return err
+	}
+
+	switch fileType {
+	case typeNtfs:
+		manifest, err = indexNtfs(file, offset, nowrite, exact)
+	case typeMbrDisk:
+		manifest, err = indexMbrDisk(file, offset, nowrite, exact)
+	default:
+		manifest, err = indexOther(file, offset, nowrite)
 	}
 
 	if debug {
@@ -38,4 +71,74 @@ func index(inputFile string, manifestFile string, offset int64, nowrite bool, ex
 	}
 
 	return nil
-}
+}
+
+
+func probeType(reader io.ReaderAt, offset int64) (fileType, error) {
+	buffer := make([]byte, probeTypeBufferLength)
+	_, err := reader.ReadAt(buffer, offset)
+	if err != nil {
+		return -1, err
+	}
+
+	// Detect NTFS (note: this also has an MBR signature!)
+	if bytes.Compare([]byte(ntfsBootMagic), buffer[ntfsBootMagicOffset:ntfsBootMagicOffset+len(ntfsBootMagic)]) == 0 {
+		return typeNtfs, nil
+	}
+
+	// Detect MBR
+	if mbrSignatureMagic == parseUintLE(buffer, mbrSignatureOffset, mbrSignatureLength) {
+		return typeMbrDisk, nil
+	}
+
+	return typeUnknown, nil
+}
+
+func indexMbrDisk(reader io.ReaderAt, offset int64, nowrite bool, exact bool) (*internal.ManifestV1, error) {
+	println("i am a disk")
+
+	buffer := make([]byte, mbrLength)
+	_, err := reader.ReadAt(buffer, offset)
+	if err != nil {
+		return nil, err
+	}
+
+	for i := int64(0); i < mbrEntryCount; i++ {
+		entryOffset := mbrEntryFirstOffset + i * mbrEntryLength
+
+		partitionFirstSector := parseUintLE(buffer, entryOffset+mbrFirstSectorRelativeOffset, mbrEntryFirstSectorRelativeLength)
+		partitionOffset := offset + partitionFirstSector*mbrSectorSize
+		Debugf("Reading MBR entry at %d, partition begins at sector %d, offset %d\n",
+			entryOffset, partitionFirstSector, partitionOffset)
+
+		if partitionOffset == 0 {
+			continue
+		}
+
+		partitionType, err := probeType(reader, partitionOffset)
+		if err != nil {
+			continue
+		}
+
+		if partitionType == typeNtfs {
+			Debugf("NTFS partition found at offset %d\n", partitionOffset)
+			return indexNtfs(reader, partitionOffset, nowrite, exact)
+		}
+	}
+
+	return nil, nil
+}
+
+func indexNtfs(reader io.ReaderAt, offset int64, nowrite bool, exact bool) (*internal.ManifestV1, error) {
+	ntfs := NewNtfsDeduper(reader, offset, nowrite, exact)
+	manifest, err := ntfs.Dedup()
+	if err != nil {
+		return nil, err
+	}
+
+	return manifest, nil
+}
+
+func indexOther(file *os.File, offset int64, nowrite bool) (*internal.ManifestV1, error) {
+	return nil, nil
+}

manifest.go

@@ -6,6 +6,10 @@ import (
 	"io/ioutil"
 )
 
+type diskManifest struct {
+
+}
+
 func readManifestFromFile(manifestFile string) (*internal.ManifestV1, error) {
 	in, err := ioutil.ReadFile(manifestFile)
 	if err != nil {

partition.go

@@ -0,0 +1,2 @@
+package main
+