-- Character classification
{-# LANGUAGE CPP #-}
module GHC.Parser.CharClass
        ( is_ident      -- Char# -> Bool
        , is_symbol     -- Char# -> Bool
        , is_any        -- Char# -> Bool
        , is_space      -- Char# -> Bool
        , is_lower      -- Char# -> Bool
        , is_upper      -- Char# -> Bool
        , is_digit      -- Char# -> Bool
        , is_alphanum   -- Char# -> Bool

        , is_decdigit, is_hexdigit, is_octdigit, is_bindigit
        , hexDigit, octDecDigit
        ) where

#include "HsVersions.h"

import GHC.Prelude

import Data.Char        ( ord, chr )
import Data.Word
import GHC.Utils.Panic

-- Bit masks

cIdent, cSymbol, cAny, cSpace, cLower, cUpper, cDigit :: Word8
cIdent :: Word8
cIdent  =  Word8
1
cSymbol :: Word8
cSymbol =  Word8
2
cAny :: Word8
cAny    =  Word8
4
cSpace :: Word8
cSpace  =  Word8
8
cLower :: Word8
cLower  = Word8
16
cUpper :: Word8
cUpper  = Word8
32
cDigit :: Word8
cDigit  = Word8
64

-- | The predicates below look costly, but aren't, GHC+GCC do a great job
-- at the big case below.

{-# INLINABLE is_ctype #-}
is_ctype :: Word8 -> Char -> Bool
is_ctype :: Word8 -> Char -> Bool
is_ctype Word8
mask Char
c = (Char -> Word8
charType Char
c forall a. Bits a => a -> a -> a
.&. Word8
mask) forall a. Eq a => a -> a -> Bool
/= Word8
0

is_ident, is_symbol, is_any, is_space, is_lower, is_upper, is_digit,
    is_alphanum :: Char -> Bool
is_ident :: Char -> Bool
is_ident  = Word8 -> Char -> Bool
is_ctype Word8
cIdent
is_symbol :: Char -> Bool
is_symbol = Word8 -> Char -> Bool
is_ctype Word8
cSymbol
is_any :: Char -> Bool
is_any    = Word8 -> Char -> Bool
is_ctype Word8
cAny
is_space :: Char -> Bool
is_space  = Word8 -> Char -> Bool
is_ctype Word8
cSpace
is_lower :: Char -> Bool
is_lower  = Word8 -> Char -> Bool
is_ctype Word8
cLower
is_upper :: Char -> Bool
is_upper  = Word8 -> Char -> Bool
is_ctype Word8
cUpper
is_digit :: Char -> Bool
is_digit  = Word8 -> Char -> Bool
is_ctype Word8
cDigit
is_alphanum :: Char -> Bool
is_alphanum = Word8 -> Char -> Bool
is_ctype (Word8
cLowerforall a. Num a => a -> a -> a
+Word8
cUpperforall a. Num a => a -> a -> a
+Word8
cDigit)

-- Utils

hexDigit :: Char -> Int
hexDigit :: Char -> Int
hexDigit Char
c | Char -> Bool
is_decdigit Char
c = Char -> Int
ord Char
c forall a. Num a => a -> a -> a
- Char -> Int
ord Char
'0'
           | Bool
otherwise     = Char -> Int
ord (Char -> Char
to_lower Char
c) forall a. Num a => a -> a -> a
- Char -> Int
ord Char
'a' forall a. Num a => a -> a -> a
+ Int
10

octDecDigit :: Char -> Int
octDecDigit :: Char -> Int
octDecDigit Char
c = Char -> Int
ord Char
c forall a. Num a => a -> a -> a
- Char -> Int
ord Char
'0'

is_decdigit :: Char -> Bool
is_decdigit :: Char -> Bool
is_decdigit Char
c
        =  Char
c forall a. Ord a => a -> a -> Bool
>= Char
'0' Bool -> Bool -> Bool
&& Char
c forall a. Ord a => a -> a -> Bool
<= Char
'9'

is_hexdigit :: Char -> Bool
is_hexdigit :: Char -> Bool
is_hexdigit Char
c
        =  Char -> Bool
is_decdigit Char
c
        Bool -> Bool -> Bool
|| (Char
c forall a. Ord a => a -> a -> Bool
>= Char
'a' Bool -> Bool -> Bool
&& Char
c forall a. Ord a => a -> a -> Bool
<= Char
'f')
        Bool -> Bool -> Bool
|| (Char
c forall a. Ord a => a -> a -> Bool
>= Char
'A' Bool -> Bool -> Bool
&& Char
c forall a. Ord a => a -> a -> Bool
<= Char
'F')

is_octdigit :: Char -> Bool
is_octdigit :: Char -> Bool
is_octdigit Char
c = Char
c forall a. Ord a => a -> a -> Bool
>= Char
'0' Bool -> Bool -> Bool
&& Char
c forall a. Ord a => a -> a -> Bool
<= Char
'7'

is_bindigit :: Char -> Bool
is_bindigit :: Char -> Bool
is_bindigit Char
c = Char
c forall a. Eq a => a -> a -> Bool
== Char
'0' Bool -> Bool -> Bool
|| Char
c forall a. Eq a => a -> a -> Bool
== Char
'1'

to_lower :: Char -> Char
to_lower :: Char -> Char
to_lower Char
c
  | Char
c forall a. Ord a => a -> a -> Bool
>=  Char
'A' Bool -> Bool -> Bool
&& Char
c forall a. Ord a => a -> a -> Bool
<= Char
'Z' = Int -> Char
chr (Char -> Int
ord Char
c forall a. Num a => a -> a -> a
- (Char -> Int
ord Char
'A' forall a. Num a => a -> a -> a
- Char -> Int
ord Char
'a'))
  | Bool
otherwise = Char
c

charType :: Char -> Word8
charType :: Char -> Word8
charType Char
c = case Char
c of
   Char
'\0'   -> Word8
0                             -- \000
   Char
'\1'   -> Word8
0                             -- \001
   Char
'\2'   -> Word8
0                             -- \002
   Char
'\3'   -> Word8
0                             -- \003
   Char
'\4'   -> Word8
0                             -- \004
   Char
'\5'   -> Word8
0                             -- \005
   Char
'\6'   -> Word8
0                             -- \006
   Char
'\7'   -> Word8
0                             -- \007
   Char
'\8'   -> Word8
0                             -- \010
   Char
'\9'   -> Word8
cSpace                        -- \t  (not allowed in strings, so !cAny)
   Char
'\10'  -> Word8
cSpace                        -- \n  (ditto)
   Char
'\11'  -> Word8
cSpace                        -- \v  (ditto)
   Char
'\12'  -> Word8
cSpace                        -- \f  (ditto)
   Char
'\13'  -> Word8
cSpace                        --  ^M (ditto)
   Char
'\14'  -> Word8
0                             -- \016
   Char
'\15'  -> Word8
0                             -- \017
   Char
'\16'  -> Word8
0                             -- \020
   Char
'\17'  -> Word8
0                             -- \021
   Char
'\18'  -> Word8
0                             -- \022
   Char
'\19'  -> Word8
0                             -- \023
   Char
'\20'  -> Word8
0                             -- \024
   Char
'\21'  -> Word8
0                             -- \025
   Char
'\22'  -> Word8
0                             -- \026
   Char
'\23'  -> Word8
0                             -- \027
   Char
'\24'  -> Word8
0                             -- \030
   Char
'\25'  -> Word8
0                             -- \031
   Char
'\26'  -> Word8
0                             -- \032
   Char
'\27'  -> Word8
0                             -- \033
   Char
'\28'  -> Word8
0                             -- \034
   Char
'\29'  -> Word8
0                             -- \035
   Char
'\30'  -> Word8
0                             -- \036
   Char
'\31'  -> Word8
0                             -- \037
   Char
'\32'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSpace               --
   Char
'\33'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              -- !
   Char
'\34'  -> Word8
cAny                          -- "
   Char
'\35'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              --  #
   Char
'\36'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              --  $
   Char
'\37'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              -- %
   Char
'\38'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              -- &
   Char
'\39'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent               -- '
   Char
'\40'  -> Word8
cAny                          -- (
   Char
'\41'  -> Word8
cAny                          -- )
   Char
'\42'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              --  *
   Char
'\43'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              -- +
   Char
'\44'  -> Word8
cAny                          -- ,
   Char
'\45'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              -- -
   Char
'\46'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              -- .
   Char
'\47'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              --  /
   Char
'\48'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cDigit   -- 0
   Char
'\49'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cDigit   -- 1
   Char
'\50'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cDigit   -- 2
   Char
'\51'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cDigit   -- 3
   Char
'\52'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cDigit   -- 4
   Char
'\53'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cDigit   -- 5
   Char
'\54'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cDigit   -- 6
   Char
'\55'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cDigit   -- 7
   Char
'\56'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cDigit   -- 8
   Char
'\57'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cDigit   -- 9
   Char
'\58'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              -- :
   Char
'\59'  -> Word8
cAny                          -- ;
   Char
'\60'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              -- <
   Char
'\61'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              -- =
   Char
'\62'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              -- >
   Char
'\63'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              -- ?
   Char
'\64'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              -- @
   Char
'\65'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- A
   Char
'\66'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- B
   Char
'\67'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- C
   Char
'\68'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- D
   Char
'\69'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- E
   Char
'\70'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- F
   Char
'\71'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- G
   Char
'\72'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- H
   Char
'\73'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- I
   Char
'\74'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- J
   Char
'\75'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- K
   Char
'\76'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- L
   Char
'\77'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- M
   Char
'\78'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- N
   Char
'\79'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- O
   Char
'\80'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- P
   Char
'\81'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- Q
   Char
'\82'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- R
   Char
'\83'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- S
   Char
'\84'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- T
   Char
'\85'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- U
   Char
'\86'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- V
   Char
'\87'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- W
   Char
'\88'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- X
   Char
'\89'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- Y
   Char
'\90'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cUpper   -- Z
   Char
'\91'  -> Word8
cAny                          -- [
   Char
'\92'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              -- backslash
   Char
'\93'  -> Word8
cAny                          -- ]
   Char
'\94'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              --  ^
   Char
'\95'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- _
   Char
'\96'  -> Word8
cAny                          -- `
   Char
'\97'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- a
   Char
'\98'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- b
   Char
'\99'  -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- c
   Char
'\100' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- d
   Char
'\101' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- e
   Char
'\102' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- f
   Char
'\103' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- g
   Char
'\104' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- h
   Char
'\105' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- i
   Char
'\106' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- j
   Char
'\107' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- k
   Char
'\108' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- l
   Char
'\109' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- m
   Char
'\110' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- n
   Char
'\111' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- o
   Char
'\112' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- p
   Char
'\113' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- q
   Char
'\114' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- r
   Char
'\115' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- s
   Char
'\116' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- t
   Char
'\117' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- u
   Char
'\118' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- v
   Char
'\119' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- w
   Char
'\120' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- x
   Char
'\121' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- y
   Char
'\122' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cIdent  forall a. Bits a => a -> a -> a
.|. Word8
cLower   -- z
   Char
'\123' -> Word8
cAny                          -- {
   Char
'\124' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              --  |
   Char
'\125' -> Word8
cAny                          -- }
   Char
'\126' -> Word8
cAny forall a. Bits a => a -> a -> a
.|. Word8
cSymbol              -- ~
   Char
'\127' -> Word8
0                             -- \177
   Char
_ -> forall a. String -> a
panic (String
"charType: " forall a. [a] -> [a] -> [a]
++ forall a. Show a => a -> String
show Char
c)