add builtin

CHANGELOG.md

@@ -4,6 +4,7 @@
   * Performance improvements
   * Fix bug in how fields were split
   * `:` works on column literals
+  * Add `#*` (list length) builtin
 
 # 1.0.0.0
 

man/MANPAGE.md

@@ -54,6 +54,9 @@ Regular expressions follow Rust's regex library: https://docs.rs/regex/
 
 **#** Prefix operator: tally (count bytes in string)
 
+**#\*** Prefix operator: list length
+:   List a  ->  a
+
 **,** Ternary operator: zip with
 :   (a -> b -> c) -> Stream a -> Stream b -> Stream c
 

src/Jacinda/AST.hs

@@ -122,6 +122,7 @@ data BUn = Tally -- length of string field
          | Dedup
          | CatMaybes
          | Negate
+         | TallyList -- length of vector
          deriving (Eq)
 
 instance Pretty BUn where
@@ -139,6 +140,7 @@ instance Pretty BUn where
     pretty Dedup      = "~."
     pretty CatMaybes  = ".?"
     pretty Negate     = "-."
+    pretty TallyList  = "#*"
 
 -- ternary
 data BTer = ZipW
@@ -380,7 +382,7 @@ instance Eq (E a) where
     (==) (IParseCol _ i) (IParseCol _ j)        = i == j
     (==) (FParseCol _ i) (FParseCol _ j)        = i == j
     (==) (Field _ i) (Field _ j)                = i == j
-    (==) LastField{} LastField{}               = True
+    (==) LastField{} LastField{}                = True
     (==) AllColumn{} AllColumn{}                = True
     (==) AllField{} AllField{}                  = True
     (==) (EApp _ e0 e1) (EApp _ e0' e1')        = e0 == e0' && e1 == e1'

src/Jacinda/Backend/Normalize.hs

@@ -272,6 +272,11 @@ eNorm (EApp ty (UBuiltin ty' Tally) e) = do
     pure $ case eI of
         StrLit _ str -> IntLit tyI (fromIntegral $ BS.length str)
         _            -> EApp ty (UBuiltin ty' Tally) eI
+eNorm (EApp ty op@(UBuiltin _ TallyList) e) = do
+    eI <- eNorm e
+    pure $ case eI of
+        (Arr _ xs) -> mkI $ fromIntegral $ V.length xs
+        _          -> EApp ty op eI
 eNorm (EApp ty (EApp ty' op@(BBuiltin _ Lt) e) e') = do
     eI <- eNorm e
     eI' <- eNorm e'

src/Jacinda/Backend/TreeWalk.hs

@@ -267,12 +267,12 @@ eEval (ix, line, ctx) = go where
         let str = asStr (go e)
             re = asRegex (go e')
             bss = splitBy re str
-            in Arr undefined (StrLit undefined <$> bss)
+            in Arr undefined (mkStr <$> bss)
     go (EApp _ (EApp _ (BBuiltin _ Splitc) e) e') =
         let str = asStr (go e)
             c = the (asStr (go e'))
             bss = BS.split c str
-            in Arr undefined (StrLit undefined <$> V.fromList bss)
+            in Arr undefined (mkStr <$> V.fromList bss)
     go (EApp _ (EApp _ (EApp _ (TBuiltin _ Substr) e0) e1) e2) =
         let eI0 = asStr (go e0)
             eI1 = asInt (go e1)
@@ -352,6 +352,9 @@ eEval (ix, line, ctx) = go where
     go (Cond _ p e0 e1) =
         let p' = asBool (go p)
             in if p' then go e0 else go e1
+    go (EApp _ (UBuiltin _ TallyList) e) =
+        let xs = asArr (go e)
+            in mkI $ fromIntegral $ V.length xs
     go e = error ("Internal error: " ++ show e)
 
 -- just shove some big number into the renamer and hope it doesn't clash (bad,

src/Jacinda/Lexer.x

@@ -107,6 +107,7 @@ tokens :-
         ","                      { mkSym Comma }
         "."                      { mkSym Dot }
         "#"                      { mkSym TallyTok }
+        "#*"                     { mkSym LengthTok }
         "[:"                     { mkSym ConstTok }
         "!"                      { mkSym Exclamation }
         ":"                      { mkSym Colon }
@@ -274,6 +275,7 @@ data Sym = PlusTok
          | Comma
          | Dot
          | TallyTok
+         | LengthTok
          | ConstTok
          | LBracePercent
          | LBraceBar
@@ -322,6 +324,7 @@ instance Pretty Sym where
     pretty Comma         = ","
     pretty Dot           = "."
     pretty TallyTok      = "#"
+    pretty LengthTok     = "#*"
     pretty Quot          = "\""
     pretty Caret         = "^"
     pretty ConstTok      = "[:"

src/Jacinda/Parser.y

@@ -53,6 +53,7 @@ import Prettyprinter (Pretty (pretty), (<+>))
     lbracePercent { TokSym $$ LBracePercent }
     lbraceBar { TokSym $$ LBraceBar }
     tally { TokSym $$ TallyTok }
+    tallyL { TokSym $$ LengthTok }
     const { TokSym $$ ConstTok }
     filter { TokSym $$ FilterTok }
     exclamation { TokSym $$ Exclamation }
@@ -257,6 +258,7 @@ E :: { E AlexPosn }
   | lanchor sepBy(E, dot) rparen { Anchor $1 (reverse $2) }
   | E E { EApp (eLoc $1) $1 $2 }
   | tally { UBuiltin $1 Tally }
+  | tallyL { UBuiltin $1 TallyList }
   | const { UBuiltin $1 Const }
   | exclamation { UBuiltin $1 Not }
   | lsqbracket E rsqbracket { Dfn $1 $2 }

src/Jacinda/Ty.hs

@@ -457,6 +457,10 @@ tyE0 (UBuiltin _ IParse)     = pure $ UBuiltin (tyArr tyStr tyI) IParse
 tyE0 (UBuiltin _ FParse)     = pure $ UBuiltin (tyArr tyStr tyF) FParse
 tyE0 (UBuiltin _ Floor)      = pure $ UBuiltin (tyArr tyF tyI) Floor
 tyE0 (UBuiltin _ Ceiling)    = pure $ UBuiltin (tyArr tyF tyI) Ceiling
+tyE0 (UBuiltin _ TallyList) = do
+    a <- dummyName "a"
+    let a' = var a
+    pure $ UBuiltin (tyArr a' tyI) TallyList
 tyE0 (UBuiltin l Negate) = do
     a <- dummyName "a"
     modify (mapClassVars (addC a (IsNum, l)))

test/Spec.hs

@@ -42,7 +42,8 @@ main = defaultMain $
         , testCase "typechecks dfn" (tyFile "test/examples/ab.jac")
         , testCase "parses parens" (tyFile "examples/lib.jac")
         , testCase "typechecks/parses correctly" (tyFile "test/examples/line.jac")
-        , testCase "split eval" (evalTo "[x+' '+y]|'' split '01-23-1987' /-/" " 01 23 1987")
+        , testCase "split eval" (evalTo "[x+' '+y]|> split '01-23-1987' /-/" "01 23 1987")
+        , testCase "length eval" (evalTo "#*split '01-23-1987' /-/" "3")
         , testCase "captureE" (evalTo "'01-23-1987' ~* 3 /(\\d{2})-(\\d{2})-(\\d{4})/" "Some 1987")
         , testCase "if...then...else" (evalTo "if #t then 0 else 1" "0")
         ]