From 5a057a8dedd1fa284e04bc2e7780e74d4600fdeb Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?P=C3=A9ter=20Szil=C3=A1gyi?= <peterke@gmail.com>
Date: Wed, 6 Jan 2016 12:11:56 +0200
Subject: core/state, trie: surface iterator entry hashes

---
 trie/iterator.go      | 25 +++++++++++++++----------
 trie/iterator_test.go | 32 +++++++++++++++++++++++++++++++-
 trie/sync_test.go     |  3 +--
 3 files changed, 47 insertions(+), 13 deletions(-)

(limited to 'trie')

diff --git a/trie/iterator.go b/trie/iterator.go
index e79de2e4e..12ca96bd1 100644
--- a/trie/iterator.go
+++ b/trie/iterator.go
@@ -20,6 +20,7 @@ import (
 	"bytes"
 	"fmt"
 
+	"github.com/ethereum/go-ethereum/common"
 	"github.com/ethereum/go-ethereum/logger"
 	"github.com/ethereum/go-ethereum/logger/glog"
 )
@@ -152,8 +153,9 @@ func (self *Iterator) key(node interface{}) []byte {
 // nodeIteratorState represents the iteration state at one particular node of the
 // trie, which can be resumed at a later invocation.
 type nodeIteratorState struct {
-	node  node // Trie node being iterated
-	child int  // Child to be processed next
+	hash  common.Hash // Hash of the node being iterated (nil if not standalone)
+	node  node        // Trie node being iterated
+	child int         // Child to be processed next
 }
 
 // NodeIterator is an iterator to traverse the trie post-order.
@@ -161,9 +163,10 @@ type NodeIterator struct {
 	trie  *Trie                // Trie being iterated
 	stack []*nodeIteratorState // Hierarchy of trie nodes persisting the iteration state
 
-	Node     node   // Current node being iterated (internal representation)
-	Leaf     bool   // Flag whether the current node is a value (data) node
-	LeafBlob []byte // Data blob contained within a leaf (otherwise nil)
+	Hash     common.Hash // Hash of the current node being iterated (nil if not standalone)
+	Node     node        // Current node being iterated (internal representation)
+	Leaf     bool        // Flag whether the current node is a value (data) node
+	LeafBlob []byte      // Data blob contained within a leaf (otherwise nil)
 }
 
 // NewNodeIterator creates an post-order trie iterator.
@@ -221,18 +224,18 @@ func (it *NodeIterator) step() {
 			}
 			parent.child++
 			it.stack = append(it.stack, &nodeIteratorState{node: node.Val, child: -1})
-		} else if node, ok := parent.node.(hashNode); ok {
+		} else if hash, ok := parent.node.(hashNode); ok {
 			// Hash node, resolve the hash child from the database, then the node itself
 			if parent.child >= 0 {
 				break
 			}
 			parent.child++
 
-			node, err := it.trie.resolveHash(node, nil, nil)
+			node, err := it.trie.resolveHash(hash, nil, nil)
 			if err != nil {
 				panic(err)
 			}
-			it.stack = append(it.stack, &nodeIteratorState{node: node, child: -1})
+			it.stack = append(it.stack, &nodeIteratorState{hash: common.BytesToHash(hash), node: node, child: -1})
 		} else {
 			break
 		}
@@ -246,14 +249,16 @@ func (it *NodeIterator) step() {
 // The method returns whether there are any more data left for inspection.
 func (it *NodeIterator) retrieve() bool {
 	// Clear out any previously set values
-	it.Node, it.Leaf, it.LeafBlob = nil, false, nil
+	it.Hash, it.Node, it.Leaf, it.LeafBlob = common.Hash{}, nil, false, nil
 
 	// If the iteration's done, return no available data
 	if it.trie == nil {
 		return false
 	}
 	// Otherwise retrieve the current node and resolve leaf accessors
-	it.Node = it.stack[len(it.stack)-1].node
+	state := it.stack[len(it.stack)-1]
+
+	it.Hash, it.Node = state.hash, state.node
 	if value, ok := it.Node.(valueNode); ok {
 		it.Leaf, it.LeafBlob = true, []byte(value)
 	}
diff --git a/trie/iterator_test.go b/trie/iterator_test.go
index fdc60b412..dc8276116 100644
--- a/trie/iterator_test.go
+++ b/trie/iterator_test.go
@@ -16,7 +16,12 @@
 
 package trie
 
-import "testing"
+import (
+	"testing"
+
+	"github.com/ethereum/go-ethereum/common"
+	"github.com/ethereum/go-ethereum/ethdb"
+)
 
 func TestIterator(t *testing.T) {
 	trie := newEmpty()
@@ -47,3 +52,28 @@ func TestIterator(t *testing.T) {
 		}
 	}
 }
+
+// Tests that the node iterator indeed walks over the entire database contents.
+func TestNodeIteratorCoverage(t *testing.T) {
+	// Create some arbitrary test trie to iterate
+	db, trie, _ := makeTestTrie()
+
+	// Gather all the node hashes found by the iterator
+	hashes := make(map[common.Hash]struct{})
+	for it := NewNodeIterator(trie); it.Next(); {
+		if it.Hash != (common.Hash{}) {
+			hashes[it.Hash] = struct{}{}
+		}
+	}
+	// Cross check the hashes and the database itself
+	for hash, _ := range hashes {
+		if _, err := db.Get(hash.Bytes()); err != nil {
+			t.Errorf("failed to retrieve reported node %x: %v", hash, err)
+		}
+	}
+	for _, key := range db.(*ethdb.MemDatabase).Keys() {
+		if _, ok := hashes[common.BytesToHash(key)]; !ok {
+			t.Errorf("state entry not reported %x", key)
+		}
+	}
+}
diff --git a/trie/sync_test.go b/trie/sync_test.go
index 94d6edd68..ce098d5a0 100644
--- a/trie/sync_test.go
+++ b/trie/sync_test.go
@@ -96,8 +96,7 @@ func checkTrieConsistency(db Database, root common.Hash) (failure error) {
 	if err != nil {
 		return
 	}
-	it := NewNodeIterator(trie)
-	for it.Next() {
+	for it := NewNodeIterator(trie); it.Next(); {
 	}
 	return nil
 }
-- 
cgit v1.2.3