Fix sub_transactions generator

Author: Soupstraw

eras/dijkstra/impl/cddl/lib/Cardano/Ledger/Dijkstra/HuddleSpec.hs

@@ -47,6 +47,7 @@ import Cardano.Ledger.Huddle.Gen qualified as Gen
 import Codec.CBOR.Cuddle.CDDL.CBORGenerator (CBORGen, liftAntiGen, withAntiGen)
 import Codec.CBOR.Term (Term (..))
 import Control.Monad (zipWithM)
+import Data.Maybe (mapMaybe)
 import Data.Proxy (Proxy (..))
 import Data.Text ()
 import Data.Text qualified as T
@@ -101,10 +102,30 @@ subTransactionsRule pname p =
   withCBORGen generate $
     pname =.= huddleRule1 @"nonempty_oset" p (huddleRule @"sub_transaction" p)
   where
+    -- The Haskell representation is @OMap TxId (Tx SubTx era)@: dedup is by
+    -- body hash, so generated sub_transactions must have distinct bodies, not
+    -- just distinct full @[body, witness, aux]@ tuples.
     generate = do
-      -- Limit the number of subtransactions generated to max 3, since they are quite large
       nElems <- Gen.sized $ \sz -> choose (1, max 1 (min sz 3))
-      S <$> generateMaybeTaggedSet nElems (scale (`div` 2) $ genRule @"sub_transaction" @era)
+      let subTxGen = scale (`div` 2) $ genRule @"sub_transaction" @era
+      txs <- uniqueByBody nElems subTxGen
+      elemsArr <- genArrayTerm txs
+      tagged <- Gen.arbitrary
+      pure $ S $ if tagged then TTagged 258 elemsArr else elemsArr
+    uniqueByBody :: Int -> CBORGen Term -> CBORGen [Term]
+    uniqueByBody n gen = loop [] n
+      where
+        triesPerElement = 20 :: Int
+        loop acc 0 = pure acc
+        loop acc k = attempt triesPerElement acc k
+        attempt 0 acc _ = pure acc
+        attempt tries acc k = do
+          tx <- gen
+          case bodyOf tx of
+            Just b | b `notElem` mapMaybe bodyOf acc -> loop (tx : acc) (k - 1)
+            _ -> attempt (tries - 1) acc k
+        bodyOf (TList (b : _)) = Just b
+        bodyOf _ = Nothing
 
 subTransactionRule ::
   forall era.