emptyCid: workaround for blocks with all 0s

there was a different CID for something declared an empty block
(used for padding), and the block with all 0s. This becomes a
problem after decode, where the docoder returns a block with all 0s.

This is a workaround, not a real solution.

Signed-off-by: Csaba Kiraly <csaba.kiraly@gmail.com>

codex/blocktype.nim

@@ -9,6 +9,7 @@
 
 import std/tables
 import std/sugar
+import std/sequtils
 export tables
 
 import pkg/upraises
@@ -88,7 +89,9 @@ proc `$`*(b: Block): string =
   result &= "cid: " & $b.cid
   result &= "\ndata: " & string.fromBytes(b.data)
 
-func new*(
+proc emptyCid*(version: CidVersion, hcodec: MultiCodec, dcodec: MultiCodec): ?!Cid
+
+proc new*(
     T: type Block,
     data: openArray[byte] = [],
     version = CIDv1,
@@ -98,9 +101,13 @@ func new*(
   ## creates a new block for both storage and network IO
   ##
 
-  let
-    hash = ? MultiHash.digest($mcodec, data).mapFailure
-    cid = ? Cid.init(version, codec, hash).mapFailure
+  let cid =
+    if all(data, proc (x: byte): bool = x == 0):
+      ? emptyCid(version, mcodec, codec)
+    else:
+      let
+        hash = ? MultiHash.digest($mcodec, data).mapFailure
+      ? Cid.init(version, codec, hash).mapFailure
 
   # TODO: If the hash is `>=` to the data,
   # use the Cid as a container!