Updated TreeHash to spec - added padding

eth2/utils/ssz/src/impl_tree_hash.rs

@@ -3,49 +3,49 @@ use super::{merkle_hash, ssz_encode, TreeHash};
 use hashing::hash;
 
 impl TreeHash for u8 {
-    fn hash_tree_root(&self) -> Vec<u8> {
+    fn hash_tree_root_internal(&self) -> Vec<u8> {
         ssz_encode(self)
     }
 }
 
 impl TreeHash for u16 {
-    fn hash_tree_root(&self) -> Vec<u8> {
+    fn hash_tree_root_internal(&self) -> Vec<u8> {
         ssz_encode(self)
     }
 }
 
 impl TreeHash for u32 {
-    fn hash_tree_root(&self) -> Vec<u8> {
+    fn hash_tree_root_internal(&self) -> Vec<u8> {
         ssz_encode(self)
     }
 }
 
 impl TreeHash for u64 {
-    fn hash_tree_root(&self) -> Vec<u8> {
+    fn hash_tree_root_internal(&self) -> Vec<u8> {
         ssz_encode(self)
     }
 }
 
 impl TreeHash for usize {
-    fn hash_tree_root(&self) -> Vec<u8> {
+    fn hash_tree_root_internal(&self) -> Vec<u8> {
         ssz_encode(self)
     }
 }
 
 impl TreeHash for Address {
-    fn hash_tree_root(&self) -> Vec<u8> {
+    fn hash_tree_root_internal(&self) -> Vec<u8> {
         ssz_encode(self)
     }
 }
 
 impl TreeHash for H256 {
-    fn hash_tree_root(&self) -> Vec<u8> {
+    fn hash_tree_root_internal(&self) -> Vec<u8> {
         ssz_encode(self)
     }
 }
 
 impl TreeHash for [u8] {
-    fn hash_tree_root(&self) -> Vec<u8> {
+    fn hash_tree_root_internal(&self) -> Vec<u8> {
         if self.len() > 32 {
             return hash(&self);
         }
@@ -57,12 +57,12 @@ impl<T> TreeHash for Vec<T>
 where
     T: TreeHash,
 {
-    /// Returns the merkle_hash of a list of hash_tree_root values created
+    /// Returns the merkle_hash of a list of hash_tree_root_internal values created
     /// from the given list.
     /// Note: A byte vector, Vec<u8>, must be converted to a slice (as_slice())
     ///       to be handled properly (i.e. hashed) as byte array.
-    fn hash_tree_root(&self) -> Vec<u8> {
-        let mut tree_hashes = self.iter().map(|x| x.hash_tree_root()).collect();
+    fn hash_tree_root_internal(&self) -> Vec<u8> {
+        let mut tree_hashes = self.iter().map(|x| x.hash_tree_root_internal()).collect();
         merkle_hash(&mut tree_hashes)
     }
 }
@@ -73,7 +73,7 @@ mod tests {
 
     #[test]
     fn test_impl_tree_hash_vec() {
-        let result = vec![1u32, 2, 3, 4, 5, 6, 7].hash_tree_root();
+        let result = vec![1u32, 2, 3, 4, 5, 6, 7].hash_tree_root_internal();
         assert_eq!(result.len(), 32);
     }
 }

eth2/utils/ssz/src/tree_hash.rs

@@ -4,7 +4,14 @@ const SSZ_CHUNK_SIZE: usize = 128;
 const HASHSIZE: usize = 32;
 
 pub trait TreeHash {
-    fn hash_tree_root(&self) -> Vec<u8>;
+    fn hash_tree_root_internal(&self) -> Vec<u8>;
+    fn hash_tree_root(&self) -> Vec<u8> {
+        let mut result = self.hash_tree_root_internal();
+        if result.len() < HASHSIZE {
+            zpad(&mut result, HASHSIZE);
+        }
+        result
+    }
 }
 
 /// Returns a 32 byte hash of 'list' - a vector of byte vectors.
@@ -14,7 +21,8 @@ pub fn merkle_hash(list: &mut Vec<Vec<u8>>) -> Vec<u8> {
     let (mut chunk_size, mut chunkz) = list_to_blob(list);
 
     // get data_len as bytes. It will hashed will the merkle root
-    let datalen = list.len().to_le_bytes();
+    let mut datalen = list.len().to_le_bytes().to_vec();
+    zpad(&mut datalen, 32);
 
     // Tree-hash
     while chunkz.len() > HASHSIZE {
@@ -36,33 +44,63 @@ pub fn merkle_hash(list: &mut Vec<Vec<u8>>) -> Vec<u8> {
         chunkz = new_chunkz;
     }
 
-    chunkz.append(&mut datalen.to_vec());
+    chunkz.append(&mut datalen);
     hash(&chunkz)
 }
 
 fn list_to_blob(list: &mut Vec<Vec<u8>>) -> (usize, Vec<u8>) {
-    let chunk_size = if list.is_empty() {
+    let chunk_size = if list.is_empty() || list[0].len() < SSZ_CHUNK_SIZE {
         SSZ_CHUNK_SIZE
-    } else if list[0].len() < SSZ_CHUNK_SIZE {
-        let items_per_chunk = SSZ_CHUNK_SIZE / list[0].len();
-        items_per_chunk * list[0].len()
     } else {
         list[0].len()
     };
 
-    let mut data = Vec::new();
+    let items_per_chunk = SSZ_CHUNK_SIZE / list[0].len();
+    let chunk_count = list.len() / items_per_chunk;
+
+    let mut chunkz = Vec::new();
     if list.is_empty() {
         // handle and empty list
-        data.append(&mut vec![0; SSZ_CHUNK_SIZE]);
-    } else {
+        chunkz.append(&mut vec![0; SSZ_CHUNK_SIZE]);
+    } else if list[0].len() <= SSZ_CHUNK_SIZE {
         // just create a blob here; we'll divide into
         // chunked slices when we merklize
-        data.reserve(list[0].len() * list.len());
+        let mut chunk = Vec::with_capacity(chunk_size);
+        let mut item_count_in_chunk = 0;
+        chunkz.reserve(chunk_count * chunk_size);
         for item in list.iter_mut() {
-            data.append(item);
+            item_count_in_chunk += 1;
+            chunk.append(item);
+
+            // completed chunk?
+            if item_count_in_chunk == items_per_chunk {
+                zpad(&mut chunk, chunk_size);
+                chunkz.append(&mut chunk);
+                item_count_in_chunk = 0;
+            }
+        }
+
+        // left-over uncompleted chunk?
+        if item_count_in_chunk != 0 {
+            zpad(&mut chunk, chunk_size);
+            chunkz.append(&mut chunk);
+        }
+    } else {
+        // chunks larger than SSZ_CHUNK_SIZE
+        chunkz.reserve(chunk_count * chunk_size);
+        for item in list.iter_mut() {
+            chunkz.append(item);
         }
     }
-    (chunk_size, data)
+
+    (chunk_size, chunkz)
+}
+
+/// right pads with zeros making 'bytes' 'size' in length
+fn zpad(bytes: &mut Vec<u8>, size: usize) {
+    if bytes.len() < size {
+        bytes.resize(size, 0);
+    }
 }
 
 #[cfg(test)]