progress on a streaming interface

node.go

@@ -1,23 +1,22 @@
 package merkle
 
 import (
-	"crypto"
+	"crypto/sha1"
-	_ "crypto/sha1" // to satisfy our DefaultHash
 	"fmt"
 	"hash"
 )
 
 var (
-	// DefaultHash is for checksum of blocks and nodes
+	// DefaultHashMaker is for checksum of blocks and nodes
-	DefaultHash = crypto.SHA1
+	DefaultHashMaker = func() hash.Hash { return sha1.New() }
 )
 
 // HashMaker produces a new has for use in making checksums
 type HashMaker func() hash.Hash
 
-// NewNode returns a new Node with the DefaultHash for checksums
+// NewNode returns a new Node with the DefaultHashMaker for checksums
 func NewNode() *Node {
-	return NewNodeHash(DefaultHash.New)
+	return NewNodeHash(DefaultHashMaker)
 }
 
 // NewNodeHash returns a new Node using the provided crypto.Hash for checksums
@@ -25,13 +24,22 @@ func NewNodeHash(h HashMaker) *Node {
 	return &Node{hash: h}
 }
 
+// NewNodeHashBlock returns a new Node using the provided crypto.Hash, and calculates the block's checksum
+func NewNodeHashBlock(h HashMaker, b []byte) *Node {
+	n := &Node{hash: h}
+	h1 := n.hash()
+	h1.Write(b)
+	n.checksum = h1.Sum(nil)
+	return n
+}
+
 // Node is a fundamental part of the tree.
 type Node struct {
 	hash                HashMaker
 	checksum            []byte
 	Parent, Left, Right *Node
 
-	pos int // XXX maybe keep their order when it is a direct block's hash
+	//pos int // XXX maybe keep their order when it is a direct block's hash
 }
 
 // IsLeaf indicates this node is for specific block (and has no children)

node_test.go

@@ -9,7 +9,7 @@ import (
 func TestNodeSums(t *testing.T) {
 	var (
 		nodes            []*Node
-		h                = DefaultHash.New()
+		h                = DefaultHashMaker()
 		words            = `Who were expelled from the academy for crazy & publishing obscene odes on the windows of the skull`
 		expectedChecksum = "819fe8fed7a46900bd0613344c5ba2be336c74db"
 	)
@@ -19,7 +19,7 @@ func TestNodeSums(t *testing.T) {
 			t.Errorf("on word %q, encountered %s", word, err)
 		}
 		sum := h.Sum(nil)
-		nodes = append(nodes, &Node{checksum: sum, hash: DefaultHash.New})
+		nodes = append(nodes, &Node{checksum: sum, hash: DefaultHashMaker})
 	}
 
 	newNodes := nodes

stream.go

@@ -1,14 +1,18 @@
 package merkle
 
-import "hash"
+import (
+	"hash"
+	"log"
+)
 
 // NewHash provides a hash.Hash to generate a merkle.Tree checksum, given a
 // HashMaker for the checksums of the blocks written and the blockSize of each
 // block per node in the tree.
-func NewHash(hm HashMaker, merkleBlockSize int) hash.Hash {
+func NewHash(hm HashMaker, merkleBlockLength int) hash.Hash {
 	mh := new(merkleHash)
-	mh.blockSize = merkleBlockSize
+	mh.blockSize = merkleBlockLength
 	mh.hm = hm
+	mh.tree = &Tree{Nodes: []*Node{}, BlockLength: merkleBlockLength}
 	return mh
 }
 
@@ -19,8 +23,10 @@ func NewHash(hm HashMaker, merkleBlockSize int) hash.Hash {
 // TODO satisfy the hash.Hash interface
 type merkleHash struct {
 	blockSize        int
-	tree      Tree
+	tree             *Tree
 	hm               HashMaker
+	lastBlock        []byte // as needed, for Sum()
+	partialLastBlock bool
 }
 
 // XXX this will be tricky, as the last block can be less than the BlockSize.
@@ -32,15 +38,57 @@ type merkleHash struct {
 //
 // if that last block was complete, then no worries. start the next node.
 func (mh *merkleHash) Sum(b []byte) []byte {
-	return nil
+	// TODO check if len(mh.lastBlock) < blockSize
+	sum, err := mh.tree.Root().Checksum()
+	if err != nil {
+		log.Println(err)
+	}
+	return sum
 }
+
 func (mh *merkleHash) Write(b []byte) (int, error) {
-	return 0, nil
+	// basically we need to:
+	// * chunk these writes into blockSize
+	// * create Node of the sum
+	// * add the Node to the tree
+	// * stash remainder in the mh.lastBlock
+
+	var (
+		curBlock       = make([]byte, mh.blockSize)
+		numBytes   int = 0
+		numWritten int
+	)
+	if mh.lastBlock != nil && len(mh.lastBlock) > 0 {
+		numBytes = copy(curBlock, mh.lastBlock)
+		// not adding to numWritten, since these blocks were accounted for in a
+		// prior Write()
+	}
+
+	if numBytes > 0 {
+		copy(curBlock, b[:(mh.blockSize-numBytes)])
+		numWritten += (mh.blockSize - numBytes)
+		// TODO Node for curBlock
+		n := NewNodeHashBlock(mh.hm, curBlock)
+		_ = n
+	}
+
+	numBytes = len(b) - numBytes
+	for i := 0; i < numBytes/mh.blockSize; i++ {
+		// TODO Node for curBlock
+	}
+
+	// TODO stash (numBytes % mh.blockSize) in mh.lastBlock
+
+	// TODO if len(mh.lastBlock) < blockSize, then set that before returning
+	return numWritten, nil
 }
 
 func (mh *merkleHash) Reset() {
-	mh.Tree = Tree{}
+	mh.tree = &Tree{}
+	mh.lastBlock = nil
 }
 
+// likely not the best to pass this through and not use our own node block
+// size, but let's revisit this.
 func (mh *merkleHash) BlockSize() int { return mh.hm().BlockSize() }
 func (mh *merkleHash) Size() int      { return mh.hm().Size() }