Refactor ciphertext <-> plaintext offset translation functions

Move all the intelligence into the new file address_translation.go.
That the calculations were spread out too much became apparent when adding
the file header. This should make the code much easier to modify in the
future.

7 files changed, 113 insertions, 138 deletions

diff --git a/cryptfs/address_translation.go b/cryptfs/address_translation.go
new file mode 100644
index 0000000..dfc6ef9
--- /dev/null
+++ b/cryptfs/address_translation.go
@@ -0,0 +1,79 @@
+package cryptfs
+
+// CryptFS methods that translate offsets between ciphertext and plaintext
+
+// get the block number at plain-text offset
+func (be *CryptFS) PlainOffToBlockNo(plainOffset uint64) uint64 {
+	return plainOffset / be.plainBS
+}
+
+// get the block number at ciphter-text offset
+func (be *CryptFS) CipherOffToBlockNo(cipherOffset uint64) uint64 {
+	return (cipherOffset - HEADER_LEN) / be.cipherBS
+}
+
+// get ciphertext offset of block "blockNo"
+func (be *CryptFS) BlockNoToCipherOff(blockNo uint64) uint64 {
+	return HEADER_LEN + blockNo*be.cipherBS
+}
+
+// get plaintext offset of block "blockNo"
+func (be *CryptFS) BlockNoToPlainOff(blockNo uint64) uint64 {
+	return blockNo * be.plainBS
+}
+
+// PlainSize - calculate plaintext size from ciphertext size
+func (be *CryptFS) CipherSizeToPlainSize(cipherSize uint64) uint64 {
+
+	// Zero sized files stay zero-sized
+	if cipherSize == 0 {
+		return 0
+	}
+
+	// Block number at last byte
+	blockNo := be.CipherOffToBlockNo(cipherSize - 1)
+	blockCount := blockNo + 1
+
+	overhead := BLOCK_OVERHEAD*blockCount + HEADER_LEN
+
+	return cipherSize - overhead
+}
+
+// CipherSize - calculate ciphertext size from plaintext size
+func (be *CryptFS) PlainSizeToCipherSize(plainSize uint64) uint64 {
+
+	// Block number at last byte
+	blockNo := be.PlainOffToBlockNo(plainSize - 1)
+	blockCount := blockNo + 1
+
+	overhead := BLOCK_OVERHEAD*blockCount + HEADER_LEN
+
+	return plainSize + overhead
+}
+
+// Split a plaintext byte range into (possibly partial) blocks
+func (be *CryptFS) ExplodePlainRange(offset uint64, length uint64) []intraBlock {
+	var blocks []intraBlock
+	var nextBlock intraBlock
+	nextBlock.fs = be
+
+	for length > 0 {
+		nextBlock.BlockNo = be.PlainOffToBlockNo(offset)
+		nextBlock.Skip = offset - be.BlockNoToPlainOff(nextBlock.BlockNo)
+
+		// Minimum of remaining data and remaining space in the block
+		nextBlock.Length = MinUint64(length, be.plainBS-nextBlock.Skip)
+
+		blocks = append(blocks, nextBlock)
+		offset += nextBlock.Length
+		length -= nextBlock.Length
+	}
+	return blocks
+}
+
+func MinUint64(x uint64, y uint64) uint64 {
+	if x < y {
+		return x
+	}
+	return y
+}
diff --git a/cryptfs/config_file.go b/cryptfs/config_file.go
index 7e762ad..16a3eec 100644
--- a/cryptfs/config_file.go
+++ b/cryptfs/config_file.go
@@ -1,8 +1,8 @@
 package cryptfs
 
 import (
-	"fmt"
 	"encoding/json"
+	"fmt"
 	"io/ioutil"
 )
 import "os"
diff --git a/cryptfs/content_test.go b/cryptfs/content_test.go
index 4e1b447..37635f0 100644
--- a/cryptfs/content_test.go
+++ b/cryptfs/content_test.go
@@ -16,7 +16,7 @@ func TestSplitRange(t *testing.T) {
 		testRange{0, 10},
 		testRange{234, 6511},
 		testRange{65444, 54},
-		testRange{0, 1024*1024},
+		testRange{0, 1024 * 1024},
 		testRange{0, 65536},
 		testRange{6654, 8945})
 
@@ -24,8 +24,8 @@ func TestSplitRange(t *testing.T) {
 	f := NewCryptFS(key, true)
 
 	for _, r := range ranges {
-		parts := f.SplitRange(r.offset, r.length)
-		var lastBlockNo uint64 = 1<<63
+		parts := f.ExplodePlainRange(r.offset, r.length)
+		var lastBlockNo uint64 = 1 << 63
 		for _, p := range parts {
 			if p.BlockNo == lastBlockNo {
 				t.Errorf("Duplicate block number %d", p.BlockNo)
@@ -51,11 +51,15 @@ func TestCiphertextRange(t *testing.T) {
 	f := NewCryptFS(key, true)
 
 	for _, r := range ranges {
-		alignedOffset, alignedLength, skipBytes := f.CiphertextRange(r.offset, r.length)
+
+		blocks := f.ExplodePlainRange(r.offset, r.length)
+		alignedOffset, alignedLength := blocks[0].JointCiphertextRange(blocks)
+		skipBytes := blocks[0].Skip
+
 		if alignedLength < r.length {
 			t.Errorf("alignedLength=%s is smaller than length=%d", alignedLength, r.length)
 		}
-		if (alignedOffset - HEADER_LEN)%f.cipherBS != 0 {
+		if (alignedOffset-HEADER_LEN)%f.cipherBS != 0 {
 			t.Errorf("alignedOffset=%d is not aligned", alignedOffset)
 		}
 		if r.offset%f.plainBS != 0 && skipBytes == 0 {
@@ -68,19 +72,19 @@ func TestBlockNo(t *testing.T) {
 	key := make([]byte, KEY_LEN)
 	f := NewCryptFS(key, true)
 
-	b := f.BlockNoCipherOff(788)
+	b := f.CipherOffToBlockNo(788)
 	if b != 0 {
 		t.Errorf("actual: %d", b)
 	}
-	b = f.BlockNoCipherOff(HEADER_LEN + f.CipherBS())
+	b = f.CipherOffToBlockNo(HEADER_LEN + f.cipherBS)
 	if b != 1 {
 		t.Errorf("actual: %d", b)
 	}
-	b = f.BlockNoPlainOff(788)
+	b = f.PlainOffToBlockNo(788)
 	if b != 0 {
 		t.Errorf("actual: %d", b)
 	}
-	b = f.BlockNoPlainOff(f.PlainBS())
+	b = f.PlainOffToBlockNo(f.plainBS)
 	if b != 1 {
 		t.Errorf("actual: %d", b)
 	}
diff --git a/cryptfs/cryptfs.go b/cryptfs/cryptfs.go
index 0593214..9fe492d 100644
--- a/cryptfs/cryptfs.go
+++ b/cryptfs/cryptfs.go
@@ -13,7 +13,7 @@ const (
 	KEY_LEN         = 32 // AES-256
 	NONCE_LEN       = 12
 	AUTH_TAG_LEN    = 16
-	BLOCK_OVERHEAD = NONCE_LEN + AUTH_TAG_LEN
+	BLOCK_OVERHEAD  = NONCE_LEN + AUTH_TAG_LEN
 )
 
 type CryptFS struct {
@@ -61,8 +61,3 @@ func NewCryptFS(key []byte, useOpenssl bool) *CryptFS {
 func (be *CryptFS) PlainBS() uint64 {
 	return be.plainBS
 }
-
-// Get ciphertext block size
-func (be *CryptFS) CipherBS() uint64 {
-	return be.cipherBS
-}
diff --git a/cryptfs/cryptfs_content.go b/cryptfs/cryptfs_content.go
index 03253d3..d74570f 100644
--- a/cryptfs/cryptfs_content.go
+++ b/cryptfs/cryptfs_content.go
@@ -12,11 +12,6 @@ import (
 	"os"
 )
 
-const (
-	// A block of 4124 zero bytes has this md5
-	ZeroBlockMd5 = "64331af89bd15a987b39855338336237"
-)
-
 // md5sum - debug helper, return md5 hex string
 func md5sum(buf []byte) string {
 	rawHash := md5.Sum(buf)
@@ -110,106 +105,6 @@ func (be *CryptFS) EncryptBlock(plaintext []byte, blockNo uint64, fileId []byte)
 	return ciphertext
 }
 
-// Split a plaintext byte range into (possibly partial) blocks
-func (be *CryptFS) SplitRange(offset uint64, length uint64) []intraBlock {
-	var b intraBlock
-	var parts []intraBlock
-
-	b.fs = be
-
-	for length > 0 {
-		b.BlockNo = offset / be.plainBS
-		b.Skip = offset % be.plainBS
-		// Minimum of remaining data and remaining space in the block
-		b.Length = be.minu64(length, be.plainBS-b.Skip)
-		parts = append(parts, b)
-		offset += b.Length
-		length -= b.Length
-	}
-	return parts
-}
-
-// PlainSize - calculate plaintext size from ciphertext size
-func (be *CryptFS) PlainSize(size uint64) uint64 {
-
-	// Zero sized files stay zero-sized
-	if size == 0 {
-		return 0
-	}
-
-	// Account for header
-	size -= HEADER_LEN
-
-	overhead := be.cipherBS - be.plainBS
-	nBlocks := (size + be.cipherBS - 1) / be.cipherBS
-	if nBlocks*overhead > size {
-		Warn.Printf("PlainSize: Negative size, returning 0 instead\n")
-		return 0
-	}
-	size -= nBlocks * overhead
-
-	return size
-}
-
-// CipherSize - calculate ciphertext size from plaintext size
-func (be *CryptFS) CipherSize(size uint64) uint64 {
-	overhead := be.cipherBS - be.plainBS
-	nBlocks := (size + be.plainBS - 1) / be.plainBS
-	size += nBlocks * overhead
-
-	return size
-}
-
-func (be *CryptFS) minu64(x uint64, y uint64) uint64 {
-	if x < y {
-		return x
-	}
-	return y
-}
-
-// CiphertextRange - Get byte range in backing ciphertext corresponding
-// to plaintext range. Returns a range aligned to ciphertext blocks.
-func (be *CryptFS) CiphertextRange(offset uint64, length uint64) (alignedOffset uint64, alignedLength uint64, skipBytes int) {
-	// Decrypting the ciphertext will yield too many plaintext bytes. Skip this number
-	// of bytes from the front.
-	skip := offset % be.plainBS
-
-	firstBlockNo := offset / be.plainBS
-	lastBlockNo := (offset + length - 1) / be.plainBS
-
-	alignedOffset = HEADER_LEN + firstBlockNo * be.cipherBS
-	alignedLength = (lastBlockNo - firstBlockNo + 1) * be.cipherBS
-
-	skipBytes = int(skip)
-	return alignedOffset, alignedLength, skipBytes
-}
-
-// Get the byte range in the ciphertext corresponding to blocks
-// (full blocks!)
-func (be *CryptFS) JoinCiphertextRange(blocks []intraBlock) (uint64, uint64) {
-
-	offset, _ := blocks[0].CiphertextRange()
-	last := blocks[len(blocks)-1]
-	length := (last.BlockNo - blocks[0].BlockNo + 1) * be.cipherBS
-
-	return offset, length
-}
-
-// Crop plaintext that correspons to complete cipher blocks down to what is
-// requested according to "iblocks"
-func (be *CryptFS) CropPlaintext(plaintext []byte, blocks []intraBlock) []byte {
-	offset := blocks[0].Skip
-	last := blocks[len(blocks)-1]
-	length := (last.BlockNo - blocks[0].BlockNo + 1) * be.plainBS
-	var cropped []byte
-	if offset+length > uint64(len(plaintext)) {
-		cropped = plaintext[offset:]
-	} else {
-		cropped = plaintext[offset : offset+length]
-	}
-	return cropped
-}
-
 // MergeBlocks - Merge newData into oldData at offset
 // New block may be bigger than both newData and oldData
 func (be *CryptFS) MergeBlocks(oldData []byte, newData []byte, offset int) []byte {
@@ -230,13 +125,3 @@ func (be *CryptFS) MergeBlocks(oldData []byte, newData []byte, offset int) []byt
 	}
 	return out[0:outLen]
 }
-
-// Get the block number at plain-text offset
-func (be *CryptFS) BlockNoPlainOff(plainOffset uint64) uint64 {
-	return plainOffset / be.plainBS
-}
-
-// Get the block number at ciphter-text offset
-func (be *CryptFS) BlockNoCipherOff(cipherOffset uint64) uint64 {
-	return (cipherOffset - HEADER_LEN) / be.cipherBS
-}
diff --git a/cryptfs/file_header.go b/cryptfs/file_header.go
index 3fd7266..e16cbab 100644
--- a/cryptfs/file_header.go
+++ b/cryptfs/file_header.go
@@ -10,15 +10,15 @@ import (
 )
 
 const (
-	HEADER_CURRENT_VERSION  = 1 // Current on-disk-format version
-	HEADER_VERSION_LEN = 2 // uint16
-	HEADER_ID_LEN = 16 // 128 bit random file id
-	HEADER_LEN = HEADER_VERSION_LEN + HEADER_ID_LEN // Total header length
+	HEADER_CURRENT_VERSION = 1                                  // Current on-disk-format version
+	HEADER_VERSION_LEN     = 2                                  // uint16
+	HEADER_ID_LEN          = 16                                 // 128 bit random file id
+	HEADER_LEN             = HEADER_VERSION_LEN + HEADER_ID_LEN // Total header length
 )
 
 type FileHeader struct {
 	Version uint16
-	Id []byte
+	Id      []byte
 }
 
 // Pack - serialize fileHeader object
diff --git a/cryptfs/intrablock.go b/cryptfs/intrablock.go
index c83976c..faff471 100644
--- a/cryptfs/intrablock.go
+++ b/cryptfs/intrablock.go
@@ -19,13 +19,13 @@ func (ib *intraBlock) IsPartial() bool {
 // CiphertextRange - get byte range in ciphertext file corresponding to BlockNo
 // (complete block)
 func (ib *intraBlock) CiphertextRange() (offset uint64, length uint64) {
-	return HEADER_LEN + ib.BlockNo * ib.fs.cipherBS, ib.fs.cipherBS
+	return ib.fs.BlockNoToCipherOff(ib.BlockNo), ib.fs.cipherBS
 }
 
 // PlaintextRange - get byte range in plaintext corresponding to BlockNo
 // (complete block)
 func (ib *intraBlock) PlaintextRange() (offset uint64, length uint64) {
-	return ib.BlockNo * ib.fs.plainBS, ib.fs.plainBS
+	return ib.fs.BlockNoToPlainOff(ib.BlockNo), ib.fs.plainBS
 }
 
 // CropBlock - crop a potentially larger plaintext block down to the relevant part
@@ -37,3 +37,15 @@ func (ib *intraBlock) CropBlock(d []byte) []byte {
 	}
 	return d[ib.Skip:lenWant]
 }
+
+// Ciphertext range corresponding to the sum of all "blocks" (complete blocks)
+func (ib *intraBlock) JointCiphertextRange(blocks []intraBlock) (offset uint64, length uint64) {
+	firstBlock := blocks[0]
+	lastBlock := blocks[len(blocks)-1]
+
+	offset = ib.fs.BlockNoToCipherOff(firstBlock.BlockNo)
+	offsetLast := ib.fs.BlockNoToCipherOff(lastBlock.BlockNo)
+	length = offsetLast + ib.fs.cipherBS - offset
+
+	return offset, length
+}