packages feed

sbv-14.1: Data/SBV/Utils/PrettyNum.hs

-----------------------------------------------------------------------------
-- |
-- Module    : Data.SBV.Utils.PrettyNum
-- Copyright : (c) Levent Erkok
-- License   : BSD3
-- Maintainer: erkokl@gmail.com
-- Stability : experimental
--
-- Number representations in hex/bin
-----------------------------------------------------------------------------

{-# LANGUAGE FlexibleInstances   #-}
{-# LANGUAGE OverloadedStrings   #-}
{-# LANGUAGE ScopedTypeVariables #-}

{-# OPTIONS_GHC -Wall -Werror #-}

module Data.SBV.Utils.PrettyNum (
        PrettyNum(..), readBin, shex, chex, shexI, sbin, sbinI
      , showCFloat, showCDouble, showHFloat, showHDouble, showBFloat, showFloatAtBase
      , showSMTFloat, showSMTDouble, smtRoundingMode, cvToSMTLib
      , showNegativeNumber
      ) where

import Data.Bits  ((.&.), countTrailingZeros, testBit)
import Data.Char  (intToDigit, ord, chr)
import Data.Int   (Int8, Int16, Int32, Int64)
import Data.List  (isPrefixOf)
import Data.Maybe (fromJust, fromMaybe, listToMaybe)
import Data.Ratio (numerator, denominator)
import Data.Word  (Word8, Word16, Word32, Word64)
import Data.Text  (Text)
import qualified Data.Text as T

import qualified Data.Set as Set

import Numeric (showIntAtBase, showHex, readInt, floatToDigits)
import qualified Numeric as N (showHFloat)

import Data.SBV.Core.Data
import Data.SBV.Core.Kind (smtType, showBaseKind)

import Data.SBV.Core.AlgReals    (algRealToSMTLib2)
import Data.SBV.Core.SizedFloats (fprToSMTLib2, bfToString)

import Data.SBV.Utils.Lib     (stringToQFS, showText)
import Data.SBV.Utils.Numeric (smtRoundingMode, floatToWord, doubleToWord)

-- | PrettyNum class captures printing of numbers in hex and binary formats; also supporting negative numbers.
class PrettyNum a where
  -- | Show a number in hexadecimal, starting with @0x@ and type.
  hexS :: a -> Text
  -- | Show a number in binary, starting with @0b@ and type.
  binS :: a -> Text
  -- | Show a number in hexadecimal, starting with @0x@ but no type.
  hexP :: a -> Text
  -- | Show a number in binary, starting with @0b@ but no type.
  binP :: a -> Text
  -- | Show a number in hex, without prefix, or types.
  hex :: a -> Text
  -- | Show a number in bin, without prefix, or types.
  bin :: a -> Text

-- Why not default methods? Because defaults need "Integral a" but Bool is not..
instance PrettyNum Bool where
  hexS = showText
  binS = showText
  hexP = showText
  binP = showText
  hex  = showText
  bin  = showText

instance PrettyNum String where
  hexS = showText
  binS = showText
  hexP = showText
  binP = showText
  hex  = showText
  bin  = showText

instance PrettyNum Word8 where
  hexS = shex True  True  (False, 8)
  binS = sbin True  True  (False, 8)

  hexP = shex False True  (False, 8)
  binP = sbin False True  (False, 8)

  hex  = shex False False (False, 8)
  bin  = sbin False False (False, 8)

instance PrettyNum Int8 where
  hexS = shex True  True  (True, 8)
  binS = sbin True  True  (True, 8)

  hexP = shex False True  (True, 8)
  binP = sbin False True  (True, 8)

  hex  = shex False False (True, 8)
  bin  = sbin False False (True, 8)

instance PrettyNum Word16 where
  hexS = shex True  True  (False, 16)
  binS = sbin True  True  (False, 16)

  hexP = shex False True  (False, 16)
  binP = sbin False True  (False, 16)

  hex  = shex False False (False, 16)
  bin  = sbin False False (False, 16)

instance PrettyNum Int16 where
  hexS = shex True  True  (True, 16)
  binS = sbin True  True  (True, 16)

  hexP = shex False True  (True, 16)
  binP = sbin False True  (True, 16)

  hex  = shex False False (True, 16)
  bin  = sbin False False (True, 16)

instance PrettyNum Word32 where
  hexS = shex True  True  (False, 32)
  binS = sbin True  True  (False, 32)

  hexP = shex False True  (False, 32)
  binP = sbin False True  (False, 32)

  hex  = shex False False (False, 32)
  bin  = sbin False False (False, 32)

instance PrettyNum Int32 where
  hexS = shex True  True  (True, 32)
  binS = sbin True  True  (True, 32)

  hexP = shex False True  (True, 32)
  binP = sbin False True  (True, 32)

  hex  = shex False False (True, 32)
  bin  = sbin False False (True, 32)

instance PrettyNum Word64 where
  hexS = shex True  True  (False, 64)
  binS = sbin True  True  (False, 64)

  hexP = shex False True  (False, 64)
  binP = sbin False True  (False, 64)

  hex  = shex False False (False, 64)
  bin  = sbin False False (False, 64)

instance PrettyNum Int64 where
  hexS = shex True  True  (True, 64)
  binS = sbin True  True  (True, 64)

  hexP = shex False True  (True, 64)
  binP = sbin False True  (True, 64)

  hex  = shex False False (True, 64)
  bin  = sbin False False (True, 64)

instance PrettyNum Integer where
  hexS = shexI True  True
  binS = sbinI True  True

  hexP = shexI False True
  binP = sbinI False True

  hex  = shexI False False
  bin  = sbinI False False

shBKind :: HasKind a => a -> Text
shBKind a = " :: " <> showBaseKind (kindOf a)

instance PrettyNum CV where
  hexS = cvPretty True  True  True  True  (\f -> T.pack (N.showHFloat f "")) (\d -> T.pack (N.showHFloat d ""))
  binS = cvPretty False True  True  True  (\f -> T.pack (showBFloat f ""))   (\d -> T.pack (showBFloat d ""))
  hexP = cvPretty True  False True  False showText                           showText
  binP = cvPretty False False True  False showText                           showText
  hex  = cvPretty True  False False False showText                           showText
  bin  = cvPretty False False False False showText                           showText

-- | Factor out the common structure of PrettyNum CV methods
cvPretty :: Bool              -- ^ isHex (True) or isBin (False)
         -> Bool              -- ^ Show type suffix on integers
         -> Bool              -- ^ Show prefix (0x/0b) on integers
         -> Bool              -- ^ Show kind suffix on non-integer cases
         -> (Float -> Text)   -- ^ Float formatter
         -> (Double -> Text)  -- ^ Double formatter
         -> CV -> Text
cvPretty isHex shType shPre shKind fmtF fmtD cv
  | isADT           cv                         = showText cv <> knd
  | isBoolean       cv                         = (if isHex then hexS else binS) (cvToBool cv) <> knd
  | isFloat         cv, CFloat   f <- cvVal cv = fmtF f <> knd
  | isDouble        cv, CDouble  d <- cvVal cv = fmtD d <> knd
  | isFP            cv, CFP      f <- cvVal cv = T.pack (bfToString base shPre True f) <> knd
  | isReal          cv, CAlgReal r <- cvVal cv = showText r <> knd
  | isString        cv, CString  s <- cvVal cv = showText s <> knd
  | not (isBounded cv), CInteger i <- cvVal cv = intI i
  | CInteger i <- cvVal cv                     = intB (hasSign cv, intSizeOf cv) i
  | True                                       = error $ "PrettyNum: Received CV that can't be displayed: " ++ show cv
  where knd  = if shKind then shBKind cv else ""
        base = if isHex then 16 else 2
        intI = (if isHex then shexI else sbinI) shType shPre
        intB = (if isHex then shex  else sbin)  shType shPre

instance (SymVal a, PrettyNum a) => PrettyNum (SBV a) where
  hexS s = maybe (showText s) (hexS :: a -> Text) $ unliteral s
  binS s = maybe (showText s) (binS :: a -> Text) $ unliteral s

  hexP s = maybe (showText s) (hexP :: a -> Text) $ unliteral s
  binP s = maybe (showText s) (binP :: a -> Text) $ unliteral s

  hex  s = maybe (showText s) (hex  :: a -> Text) $ unliteral s
  bin  s = maybe (showText s) (bin  :: a -> Text) $ unliteral s

-- | Show as a hexadecimal value. First bool controls whether type info is printed
-- while the second boolean controls whether 0x prefix is printed. The tuple is
-- the signedness and the bit-length of the input. The length of the string
-- will /not/ depend on the value, but rather the bit-length.
shex :: (Show a, Integral a) => Bool -> Bool -> (Bool, Int) -> a -> Text
shex shType shPre (signed, size) a
 | a < 0
 = "-" <> pre <> T.pack (pad l (s16 (abs (fromIntegral a :: Integer)))) <> t
 | True
 = pre <> T.pack (pad l (s16 a)) <> t
 where t | shType = " :: " <> (if signed then "Int" else "Word") <> showText size
         | True   = T.empty
       pre | shPre = "0x"
           | True  = T.empty
       l = (size + 3) `div` 4

-- | Show as hexadecimal, but for C programs. We have to be careful about
-- printing min-bounds, since C does some funky casting, possibly losing
-- the sign bit. In those cases, we use the defined constants in <stdint.h>.
-- We also properly append the necessary suffixes as needed.
chex :: (Show a, Integral a) => Bool -> Bool -> (Bool, Int) -> a -> Text
chex shType shPre (signed, size) a
   | Just s <- (signed, size, fromIntegral a) `lookup` specials
   = T.pack s
   | True
   = shex shType shPre (signed, size) a <> T.pack suffix
  where specials :: [((Bool, Int, Integer), String)]
        specials = [ ((True,  8, fromIntegral (minBound :: Int8)),  "INT8_MIN" )
                   , ((True, 16, fromIntegral (minBound :: Int16)), "INT16_MIN")
                   , ((True, 32, fromIntegral (minBound :: Int32)), "INT32_MIN")
                   , ((True, 64, fromIntegral (minBound :: Int64)), "INT64_MIN")
                   ]
        suffix = case (signed, size) of
                   (False, 16) -> "U"

                   (False, 32) -> "UL"
                   (True,  32) -> "L"

                   (False, 64) -> "ULL"
                   (True,  64) -> "LL"

                   _           -> ""

-- | Show as a hexadecimal value, integer version. Almost the same as shex above
-- except we don't have a bit-length so the length of the string will depend
-- on the actual value.
shexI :: Bool -> Bool -> Integer -> Text
shexI shType shPre a
 | a < 0
 = "-" <> pre <> T.pack (s16 (abs a)) <> t
 | True
 = pre <> T.pack (s16 a) <> t
 where t | shType = " :: Integer"
         | True   = T.empty
       pre | shPre = "0x"
           | True  = T.empty

-- | Similar to 'shex'; except in binary.
sbin :: (Show a, Integral a) => Bool -> Bool -> (Bool, Int) -> a -> Text
sbin shType shPre (signed,size) a
 | a < 0
 = "-" <> pre <> T.pack (pad size (s2 (abs (fromIntegral a :: Integer)))) <> t
 | True
 = pre <> T.pack (pad size (s2 a)) <> t
 where t | shType = " :: " <> (if signed then "Int" else "Word") <> showText size
         | True   = T.empty
       pre | shPre = "0b"
           | True  = T.empty

-- | Similar to 'shexI'; except in binary.
sbinI :: Bool -> Bool -> Integer -> Text
sbinI shType shPre a
 | a < 0
 = "-" <> pre <> T.pack (s2 (abs a)) <> t
 | True
 = pre <> T.pack (s2 a) <> t
 where t | shType = " :: Integer"
         | True   = T.empty
       pre | shPre = "0b"
           | True  = T.empty

-- | Pad a string to a given length. If the string is longer, then we don't drop anything.
pad :: Int -> String -> String
pad l s = replicate (l - length s) '0' ++ s

-- | Binary printer
s2 :: (Show a, Integral a) => a -> String
s2  v = showIntAtBase 2 dig v "" where dig = fromJust . flip lookup [(0, '0'), (1, '1')]

-- | Hex printer
s16 :: (Show a, Integral a) => a -> String
s16 v = showHex v ""

-- | A more convenient interface for reading binary numbers, also supports negative numbers
readBin :: Num a => String -> a
readBin ('-':s) = -(readBin s)
readBin s = case readInt 2 isDigit cvt s' of
              [(a, "")] -> a
              _         -> error $ "SBV.readBin: Cannot read a binary number from: " ++ show s
  where cvt c = ord c - ord '0'
        isDigit = (`elem` ("01" :: String))
        s' | "0b" `isPrefixOf` s = drop 2 s
           | True                = s

-- | A version of show for floats that generates correct C literals for nan/infinite. NB. Requires "math.h" to be included.
showCFloat :: Float -> String
showCFloat f
   | isNaN f             = "((float) NAN)"
   | isInfinite f, f < 0 = "((float) (-INFINITY))"
   | isInfinite f        = "((float) INFINITY)"
   | True                = N.showHFloat f $ "F /* " ++ show f ++ "F */"

-- | A version of show for doubles that generates correct C literals for nan/infinite. NB. Requires "math.h" to be included.
showCDouble :: Double -> String
showCDouble d
   | isNaN d             = "((double) NAN)"
   | isInfinite d, d < 0 = "((double) (-INFINITY))"
   | isInfinite d        = "((double) INFINITY)"
   | True                = N.showHFloat d " /* " ++ show d ++ " */"

-- | A version of show for floats that generates correct Haskell literals for nan/infinite
showHFloat :: Float -> String
showHFloat f
   | isNaN f             = "((0/0) :: Float)"
   | isInfinite f, f < 0 = "((-1/0) :: Float)"
   | isInfinite f        = "((1/0) :: Float)"
   | True                = show f

-- | A version of show for doubles that generates correct Haskell literals for nan/infinite
showHDouble :: Double -> String
showHDouble d
   | isNaN d             = "((0/0) :: Double)"
   | isInfinite d, d < 0 = "((-1/0) :: Double)"
   | isInfinite d        = "((1/0) :: Double)"
   | True                = show d

-- | A version of show for floats that generates correct SMTLib literals using the rounding mode
showSMTFloat :: Float -> Text
showSMTFloat f
   | isNaN f             = as "NaN"
   | isInfinite f, f < 0 = as "-oo"
   | isInfinite f        = as "+oo"
   | isNegativeZero f    = as "-zero"
   | f == 0              = as "+zero"
   | True                = let w   = floatToWord f
                               b i = if w `testBit` i then '1' else '0'
                               s   = T.pack [b 31]
                               e   = T.pack [b i | i <- [30, 29 .. 23]]
                               m   = T.pack [b i | i <- [22, 21 ..  0]]
                           in "(fp #b" <> s <> " #b" <> e <> " #b" <> m <> ")"
   where as s = "(_ " <> s <> " 8 24)"

-- | A version of show for doubles that generates correct SMTLib literals using the rounding mode
showSMTDouble :: Double -> Text
showSMTDouble d
   | isNaN d             = as "NaN"
   | isInfinite d, d < 0 = as "-oo"
   | isInfinite d        = as "+oo"
   | isNegativeZero d    = as "-zero"
   | d == 0              = as "+zero"
   | True                = let w   = doubleToWord d
                               b i = if w `testBit` i then '1' else '0'
                               s   = T.pack [b 63]
                               e   = T.pack [b i | i <- [62, 61 .. 52]]
                               m   = T.pack [b i | i <- [51, 50 ..  0]]
                           in "(fp #b" <> s <> " #b" <> e <> " #b" <> m <> ")"
   where as s = "(_ " <> s <> " 11 53)"

-- | Show an SBV rational as an SMTLib value. This is used for faithful rationals.
showSMTRational :: Rational -> Text
showSMTRational r = "(SBV.Rational " <> showNegativeNumber (numerator r) <> " " <> showNegativeNumber (denominator r) <> ")"

-- | Convert a CV to an SMTLib2 compliant value
cvToSMTLib :: CV -> Text
cvToSMTLib x
  | isBoolean       x, CInteger  w      <- cvVal x = if w == 0 then "false" else "true"
  | isRoundingMode  x, CADT (s, [])     <- cvVal x = roundModeConvert s
  | isReal          x, CAlgReal  r      <- cvVal x = T.pack (algRealToSMTLib2 r)
  | isRational      x, CRational r      <- cvVal x = showSMTRational r
  | isFloat         x, CFloat    f      <- cvVal x = showSMTFloat  f
  | isDouble        x, CDouble   d      <- cvVal x = showSMTDouble d
  | isFP            x, CFP       f      <- cvVal x = T.pack (fprToSMTLib2 f)
  | not (isBounded x), CInteger  w      <- cvVal x = if w >= 0 then showText w else "(- " <> showText (abs w) <> ")"
  | not (hasSign x)  , CInteger  w      <- cvVal x = smtLibHex (intSizeOf x) w
  -- signed numbers (with 2's complement representation) is problematic
  -- since there's no way to put a bvneg over a positive number to get minBound..
  -- Hence, we punt and use binary notation in that particular case
  | hasSign x        , CInteger  w      <- cvVal x = if w == negate (2 ^ intSizeOf x)
                                                     then mkMinBound (intSizeOf x)
                                                     else negIf (w < 0) $ smtLibHex (intSizeOf x) (abs w)
  | isChar x         , CChar c          <- cvVal x = "(_ char " <> smtLibHex 8 (fromIntegral (ord c)) <> ")"
  | isString x       , CString s        <- cvVal x = "\"" <> T.pack (stringToQFS s) <> "\""
  | isList x         , CList xs         <- cvVal x = smtLibSeq (kindOf x) xs
  | isSet x          , CSet s           <- cvVal x = smtLibSet (kindOf x) s
  | isTuple x        , CTuple xs        <- cvVal x = smtLibTup (kindOf x) xs

  -- Arrays become sequence of stores
  | isArray x        , CArray ac       <- cvVal x  = smtLibArray (kindOf x) ac

  -- ADTs
  | isADT x          , CADT c          <- cvVal x = smtLibADT (cvKind x) c

  | True = error $ "SBV.cvtCV: Impossible happened: Kind/Value disagreement on: " ++ show (kindOf x, x)
  where roundModeConvert s = fromMaybe (T.pack s) (listToMaybe [smtRoundingMode m | m <- [minBound .. maxBound] :: [RoundingMode], show m == s])
        -- Carefully code hex numbers, SMTLib is picky about lengths of hex constants. For the time
        -- being, SBV only supports sizes that are multiples of 4, but the below code is more robust
        -- in case of future extensions to support arbitrary sizes.
        smtLibHex :: Int -> Integer -> Text
        smtLibHex 1  v = "#b" <> showText v
        smtLibHex sz v
          | sz `mod` 4 == 0 = "#x" <> T.pack (pad (sz `div` 4) (showHex v ""))
          | True            = "#b" <> T.pack (pad sz (showBin v ""))
           where showBin = showIntAtBase 2 intToDigit
        negIf :: Bool -> Text -> Text
        negIf True  a = "(bvneg " <> a <> ")"
        negIf False a = a

        smtLibSeq :: Kind -> [CVal] -> Text
        smtLibSeq k          [] = "(as seq.empty " <> smtType k <> ")"
        smtLibSeq (KList ek) xs = let mkSeq  [e]   = e
                                      mkSeq  es    = "(seq.++ " <> T.unwords es <> ")"
                                      mkUnit inner = "(seq.unit " <> inner <> ")"
                                  in mkSeq (mkUnit . cvToSMTLib . CV ek <$> xs)
        smtLibSeq k _ = error $ "SBV.cvToSMTLib: Impossible case (smtLibSeq), received kind: " ++ show k

        smtLibSet :: Kind -> RCSet CVal -> Text
        smtLibSet k set = case set of
                            RegularSet    rs -> Set.foldr' (modify "true")  (start "false") rs
                            ComplementSet rs -> Set.foldr' (modify "false") (start "true")  rs
          where ke = case k of
                       KSet ek -> ek
                       _       -> error $ "SBV.cvToSMTLib: Impossible case (smtLibSet), received kind: " ++ show k

                start def = "((as const " <> smtType k <> ") " <> def <> ")"

                modify how e s = "(store " <> s <> " " <> cvToSMTLib (CV ke e) <> " " <> how <> ")"

        smtLibTup :: Kind -> [CVal] -> Text
        smtLibTup (KTuple []) _  = "mkSBVTuple0"
        smtLibTup (KTuple ks) xs = "(mkSBVTuple" <> showText (length ks) <> " " <> T.unwords (zipWith (\ek e -> cvToSMTLib (CV ek e)) ks xs) <> ")"
        smtLibTup k           _  = error $ "SBV.cvToSMTLib: Impossible case (smtLibTup), received kind: " ++ show k

        -- Remember that in an ArrayModel we keep a history; i.e., the earlier elements are written later. So, we reverse the assocs
        smtLibArray :: Kind -> ArrayModel CVal CVal -> Text
        smtLibArray k@(KArray k1 k2) (ArrayModel assocs def) = mkStoreChain k k1 k2 (reverse assocs) def
        smtLibArray k              _                         = error $ "SBV.cvToSMTLib: Impossible case (smtLibArray), received non-matching kind: " ++ show k

        mkStoreChain k k1 k2 writes def = walk writes base
          where base = "((as const " <> smtType k <> ") " <> cvToSMTLib (CV k2 def) <> ")"

                walk []                  sofar = sofar
                walk ((key, val) : rest) sofar = walk rest (store key val sofar)

                store key val sofar = "(store " <> sofar <> " " <> cvToSMTLib (CV k1 key) <> " " <> cvToSMTLib (CV k2 val) <> ")"

        -- anomaly at the 2's complement min value! Have to use binary notation here
        -- as there is no positive value we can provide to make the bvneg work.. (see above)
        mkMinBound :: Int -> Text
        mkMinBound i = "#b1" <> T.replicate (i-1) "0"

        -- ADTs
        smtLibADT :: Kind -> (String,  [(Kind, CVal)]) -> Text
        smtLibADT knd (c, [])  = ascribe c knd
        smtLibADT knd (c, kvs) = "(" <> ascribe c knd <> " " <> T.unwords (map (\(k, v) -> cvToSMTLib (CV  k v)) kvs) <> ")"
        ascribe nm k = "(as " <> T.pack nm <> " " <> smtType k <> ")"

-- | Show a float as a binary
showBFloat :: (Show a, RealFloat a) => a -> ShowS
showBFloat = showFloatAtBase 2

-- | Like Haskell's showHFloat, but uses arbitrary base instead.
-- Note that the exponent is always written in decimal. Let the exponent value be d:
--    If base=10, then we use @e@ to denote the exponent; meaning 10^d
--    If base is a power of 2, then we use @p@ to denote the exponent; meaning 2^d
--    Otherwise, we use @ to denote the exponent, and it means base^d
showFloatAtBase :: (Show a, RealFloat a) => Int -> a -> ShowS
showFloatAtBase base input
  | base < 2 = error $ "showFloatAtBase: Received invalid base (must be >= 2): " ++ show base
  | True     = showString $ fmt input
  where fmt x
         | isNaN x                   = "NaN"
         | isInfinite x              = (if x < 0 then "-" else "") ++ "Infinity"
         | x < 0 || isNegativeZero x = '-' : cvt (-x)
         | True                      = cvt x

        basePow2 = base .&. (base-1) == 0
        lg2Base  = countTrailingZeros base  -- only used when basePow2 is true

        prefix = case base of
                   2  -> "0b"
                   8  -> "0o"
                   10 -> ""
                   16 -> "0x"
                   x  -> "0<" ++ show x ++ ">"

        powChar
          | base == 10 = 'e'
          | basePow2   = 'p'
          | True       = '@'

        -- why r-1? Because we're shifting the fraction by 1 digit; does reducing the exponent by 1
        f2d x = case floatToDigits (fromIntegral base) x of
                  ([],   e) -> (0, [], e - 1)
                  (d:ds, e) -> (d, ds, e - 1)

        cvt x
         | x == 0 = prefix ++ '0' : powChar : "+0"
         | True   = prefix ++ toDigit d ++ frac ds ++ pow
         where (d, ds, e)  = f2d x
               pow
                | base == 10 = powChar : shSigned e
                | basePow2   = powChar : shSigned (e * lg2Base)
                | True       = powChar : shSigned e

               shSigned v
                | v < 0      =       show v
                | True       = '+' : show v

        -- Given digits, show them except if they're all 0 then drop
        frac digits
         | all (== 0) digits = ""
         | True              = "." ++ concatMap toDigit digits

        toDigit v | v <= 15 = [intToDigit v]
                  | v <  36 = [chr (ord 'a' + v - 10)]
                  | True    = '<' : show v ++ ">"

-- | When we show a negative number in SMTLib, we must properly parenthesize.
showNegativeNumber :: (Show a, Num a, Ord a) => a -> Text
showNegativeNumber i
  | i < 0 = "(- " <> showText (-i) <> ")"
  | True  = showText i