more work on well scoped

src/Lib/Check.idr

@@ -1,10 +1,95 @@
 module Lib.Check
 
+import Control.Monad.Error.Interface
+import Control.Monad.Identity
 import Lib.Parser.Impl
+import Data.Vect
+import Data.String
 import Lib.TT
+import Syntax
 
-
-record Cxt where
-  env : List Val
-
+record Context (n : Nat) (f : Nat)  where
+  -- review this
+  env : Env f n -- Vect n (Val f)
+  types : List (String, Val f)
   pos : SourcePos
+
+
+extend : Context n f -> Val f -> Context (S n) f
+extend ctx ty = { env := ty :: ctx.env } ctx
+
+-- cribbed this, it avoids MonadError String m => everywhere
+parameters {0 m : Type -> Type} {auto _ : MonadError String m}
+
+  infer : {f : Nat} -> Context n f -> Raw -> m (Tm f n, Val f)
+  -- I think I'm hand-waving n here, probably need it in Context
+  check : {f : Nat} -> Context n f -> Raw -> Val f -> m (Tm f n)
+
+  check ctx (RLam _ _ _) ty = ?ch_rhs
+  check ctx tm ty = do
+    (tm', ty') <- infer ctx tm
+    if quote _ ty /= quote _ ty' then
+        throwError "type mismatch"
+      else pure tm'
+
+
+  infer ctx (RVar nm) = go 0 ctx.types
+    where
+      go : Nat -> List (String, Val f) -> m (Tm f n, Val f)
+      go i [] = throwError "\{show nm} not in scope"
+      -- REVIEW Local or Bnd (ezoo does not have both)
+      go i ((x, ty) :: xs) = if x == nm then pure $ (Bnd ?i_not_fin, ty) 
+        else go (i + 1) xs
+
+      
+    -- need environment of name -> type..
+  infer ctx (RApp t u icit) = do 
+    -- icit will be used for insertion, lets get this working first...
+    (t, tty) <- infer ctx t
+    case tty of
+      (VPi str icit' a b) => do
+        u <- check ctx u a
+        
+        -- is zero right here?
+        -- I have ctx.env here and TypeTheory has []
+        -- maybe because I'm not substing?
+        pure (App t u, b 0 (eval ctx.env t))
+      _ => throwError "Expected Pi type"
+      
+    -- FIXME ctx.env?
+    -- vtya <- nf ctx.env tma
+    
+  infer ctx RU = pure (U, VU) -- YOLO
+  infer ctx (RPi nm icit ty ty2) = do
+    ty' <- check ctx ty VU
+    let vty' := eval ctx.env ty'
+    -- gallais and the source paper have subst here.  They're using Tm rather
+    -- than raw. Lets look at the zoo.
+    -- maybe run through zoo2 well typed...
+    -- it just binds vty' into the environment and evaluates 
+    -- Kovacs is sticking the type vty' and the value Var l into the context
+    -- and letting the Ix pick up the Var l from the context
+    -- gallais/paper are subst the Var l into the Tm.
+    -- yaffle just pushes to the environment, but it's a list of binders
+    -- so types, names, no values
+    ty2' <- check (extend ctx vty') ty2 VU
+    let nm := fromMaybe "_" nm
+    pure (Pi nm icit ty' ty2', VU)
+  infer ctx (RLet str tm tm1 tm2) = ?rhs_5
+  infer ctx (RSrcPos x tm) = infer ({pos := x} ctx) tm
+  infer ctx (RAnn tm rty) = do
+    ty <- check ctx rty VU
+    let vty = eval ctx.env ty 
+    tm <- check ctx tm vty
+    pure (tm, vty)
+
+  infer ctx (RLam str icit tm) = throwError "can't infer lambda"
+
+  infer ctx _ = ?later
+  -- I don't have types for these yet...
+  -- infer ctx (RLit (LString str)) = ?rhs_10
+  -- infer ctx (RLit (LInt i)) = ?rhs_11
+  -- infer ctx (RLit (LBool x)) = ?rhs_12
+  -- infer ctx RHole = ?todo_meta2
+  -- infer ctx (RParseError str) = ?todo_insert_meta
+  -- infer ctx (RCase tm xs) = ?rhs_9