Lex.x

-- -*- haskell -*- File generated by the BNF Converter (bnfc 2.9.5).

-- Lexer definition for use with Alex 3
{
{-# OPTIONS -fno-warn-incomplete-patterns #-}
{-# OPTIONS_GHC -w #-}

{-# LANGUAGE PatternSynonyms #-}

module Language.EO.Phi.Syntax.Lex where

import Prelude

import qualified Data.Bits
import Data.Char     (ord)
import Data.Function (on)
import Data.Word     (Word8)
}

-- Predefined character classes

$c = [A-Z\192-\221] # [\215]  -- capital isolatin1 letter (215 = \times) FIXME
$s = [a-z\222-\255] # [\247]  -- small   isolatin1 letter (247 = \div  ) FIXME
$l = [$c $s]         -- letter
$d = [0-9]           -- digit
$i = [$l $d _ ']     -- identifier character
$u = [. \n]          -- universal: any character

-- Symbols and non-identifier-like reserved words

@rsyms = \Φ | \ξ | \Δ | \λ | \φ | \ρ | \{ | \⟦ | \⟧ | \} | \( | \) | \. | \Φ \̇ | \⊥ | \[ | \↦ | \] | \⌈ | \, | \⌉ | \* | \∅ | \⤍ | \~

:-

-- Line comment "//"
"//" [.]* ;

-- Block comment "/*" "*/"
\/ \* [$u # \*]* \* ([$u # [\* \/]] [$u # \*]* \* | \*)* \/ ;

-- Whitespace (skipped)
$white+ ;

-- Symbols
@rsyms
    { tok (eitherResIdent TV) }

-- token Bytes
\- \- | [0 1 2 3 4 5 6 7 8 9 A B C D E F][0 1 2 3 4 5 6 7 8 9 A B C D E F]\- | [0 1 2 3 4 5 6 7 8 9 A B C D E F][0 1 2 3 4 5 6 7 8 9 A B C D E F](\- [0 1 2 3 4 5 6 7 8 9 A B C D E F][0 1 2 3 4 5 6 7 8 9 A B C D E F]) +
    { tok (eitherResIdent T_Bytes) }

-- token Function
$c [$u # [\t \n \r \  \! \' \( \) \, \- \. \: \; \? \[ \] \{ \| \} \⟦ \⟧]] *
    { tok (eitherResIdent T_Function) }

-- token LabelId
$s [$u # [\t \n \r \  \! \' \( \) \, \. \: \; \? \[ \] \{ \| \} \⟦ \⟧]] *
    { tok (eitherResIdent T_LabelId) }

-- token AlphaIndex
α 0 | α [$d # 0]$d *
    { tok (eitherResIdent T_AlphaIndex) }

-- token LabelMetaId
\! τ [$u # [\t \n \r \  \! \' \( \) \, \- \. \: \; \? \[ \] \{ \| \} \⟦ \⟧]] *
    { tok (eitherResIdent T_LabelMetaId) }

-- token TailMetaId
\! t [$u # [\t \n \r \  \! \' \( \) \, \- \. \: \; \? \[ \] \{ \| \} \⟦ \⟧]] *
    { tok (eitherResIdent T_TailMetaId) }

-- token BindingsMetaId
\! B [$u # [\t \n \r \  \! \' \( \) \, \- \. \: \; \? \[ \] \{ \| \} \⟦ \⟧]] *
    { tok (eitherResIdent T_BindingsMetaId) }

-- token ObjectMetaId
\! b [$u # [\t \n \r \  \! \' \( \) \, \- \. \: \; \? \[ \] \{ \| \} \⟦ \⟧]] *
    { tok (eitherResIdent T_ObjectMetaId) }

-- token BytesMetaId
\! y [$u # [\t \n \r \  \! \' \( \) \, \- \. \: \; \? \[ \] \{ \| \} \⟦ \⟧]] *
    { tok (eitherResIdent T_BytesMetaId) }

-- token MetaFunctionName
\@ [$u # [\t \n \r \  \! \' \( \) \, \- \. \: \; \? \[ \] \{ \| \} \⟦ \⟧]] *
    { tok (eitherResIdent T_MetaFunctionName) }

-- token IntegerSigned
\- ? $d +
    { tok (eitherResIdent T_IntegerSigned) }

-- token DoubleSigned
\- ? $d + \. $d + (e \- ? $d +)?
    { tok (eitherResIdent T_DoubleSigned) }

-- token StringRaw
\" ([$u # [\" \\]] | \\ [\" \\ f n r t u]) * \"
    { tok (eitherResIdent T_StringRaw) }

-- Keywords and Ident
$l $i*
    { tok (eitherResIdent TV) }

-- String
\" ([$u # [\" \\ \n]] | (\\ (\" | \\ | \' | n | t | r | f)))* \"
    { tok (TL . unescapeInitTail) }

-- Integer
$d+
    { tok TI }

-- Double
$d+ \. $d+ (e (\-)? $d+)?
    { tok TD }

{
-- | Create a token with position.
tok :: (String -> Tok) -> (Posn -> String -> Token)
tok f p = PT p . f

-- | Token without position.
data Tok
  = TK {-# UNPACK #-} !TokSymbol  -- ^ Reserved word or symbol.
  | TL !String                    -- ^ String literal.
  | TI !String                    -- ^ Integer literal.
  | TV !String                    -- ^ Identifier.
  | TD !String                    -- ^ Float literal.
  | TC !String                    -- ^ Character literal.
  | T_Bytes !String
  | T_Function !String
  | T_LabelId !String
  | T_AlphaIndex !String
  | T_LabelMetaId !String
  | T_TailMetaId !String
  | T_BindingsMetaId !String
  | T_ObjectMetaId !String
  | T_BytesMetaId !String
  | T_MetaFunctionName !String
  | T_IntegerSigned !String
  | T_DoubleSigned !String
  | T_StringRaw !String
  deriving (Eq, Show, Ord)

-- | Smart constructor for 'Tok' for the sake of backwards compatibility.
pattern TS :: String -> Int -> Tok
pattern TS t i = TK (TokSymbol t i)

-- | Keyword or symbol tokens have a unique ID.
data TokSymbol = TokSymbol
  { tsText :: String
      -- ^ Keyword or symbol text.
  , tsID   :: !Int
      -- ^ Unique ID.
  } deriving (Show)

-- | Keyword/symbol equality is determined by the unique ID.
instance Eq  TokSymbol where (==)    = (==)    `on` tsID

-- | Keyword/symbol ordering is determined by the unique ID.
instance Ord TokSymbol where compare = compare `on` tsID

-- | Token with position.
data Token
  = PT  Posn Tok
  | Err Posn
  deriving (Eq, Show, Ord)

-- | Pretty print a position.
printPosn :: Posn -> String
printPosn (Pn _ l c) = "line " ++ show l ++ ", column " ++ show c

-- | Pretty print the position of the first token in the list.
tokenPos :: [Token] -> String
tokenPos (t:_) = printPosn (tokenPosn t)
tokenPos []    = "end of file"

-- | Get the position of a token.
tokenPosn :: Token -> Posn
tokenPosn (PT p _) = p
tokenPosn (Err p)  = p

-- | Get line and column of a token.
tokenLineCol :: Token -> (Int, Int)
tokenLineCol = posLineCol . tokenPosn

-- | Get line and column of a position.
posLineCol :: Posn -> (Int, Int)
posLineCol (Pn _ l c) = (l,c)

-- | Convert a token into "position token" form.
mkPosToken :: Token -> ((Int, Int), String)
mkPosToken t = (tokenLineCol t, tokenText t)

-- | Convert a token to its text.
tokenText :: Token -> String
tokenText t = case t of
  PT _ (TS s _) -> s
  PT _ (TL s)   -> show s
  PT _ (TI s)   -> s
  PT _ (TV s)   -> s
  PT _ (TD s)   -> s
  PT _ (TC s)   -> s
  Err _         -> "#error"
  PT _ (T_Bytes s) -> s
  PT _ (T_Function s) -> s
  PT _ (T_LabelId s) -> s
  PT _ (T_AlphaIndex s) -> s
  PT _ (T_LabelMetaId s) -> s
  PT _ (T_TailMetaId s) -> s
  PT _ (T_BindingsMetaId s) -> s
  PT _ (T_ObjectMetaId s) -> s
  PT _ (T_BytesMetaId s) -> s
  PT _ (T_MetaFunctionName s) -> s
  PT _ (T_IntegerSigned s) -> s
  PT _ (T_DoubleSigned s) -> s
  PT _ (T_StringRaw s) -> s

-- | Convert a token to a string.
prToken :: Token -> String
prToken t = tokenText t

-- | Finite map from text to token organized as binary search tree.
data BTree
  = N -- ^ Nil (leaf).
  | B String Tok BTree BTree
      -- ^ Binary node.
  deriving (Show)

-- | Convert potential keyword into token or use fallback conversion.
eitherResIdent :: (String -> Tok) -> String -> Tok
eitherResIdent tv s = treeFind resWords
  where
  treeFind N = tv s
  treeFind (B a t left right) =
    case compare s a of
      LT -> treeFind left
      GT -> treeFind right
      EQ -> t

-- | The keywords and symbols of the language organized as binary search tree.
resWords :: BTree
resWords =
  b "\934\775" 13
    (b "]" 7
       (b "," 4
          (b ")" 2 (b "(" 1 N N) (b "*" 3 N N)) (b "[" 6 (b "." 5 N N) N))
       (b "~" 10
          (b "}" 9 (b "{" 8 N N) N) (b "\934" 12 (b "\916" 11 N N) N)))
    (b "\8869" 20
       (b "\966" 17
          (b "\958" 15 (b "\955" 14 N N) (b "\961" 16 N N))
          (b "\8709" 19 (b "\8614" 18 N N) N))
       (b "\10214" 23
          (b "\8969" 22 (b "\8968" 21 N N) N)
          (b "\10509" 25 (b "\10215" 24 N N) N)))
  where
  b s n = B bs (TS bs n)
    where
    bs = s

-- | Unquote string literal.
unescapeInitTail :: String -> String
unescapeInitTail = id . unesc . tail . id
  where
  unesc s = case s of
    '\\':c:cs | elem c ['\"', '\\', '\''] -> c : unesc cs
    '\\':'n':cs  -> '\n' : unesc cs
    '\\':'t':cs  -> '\t' : unesc cs
    '\\':'r':cs  -> '\r' : unesc cs
    '\\':'f':cs  -> '\f' : unesc cs
    '"':[]       -> []
    c:cs         -> c : unesc cs
    _            -> []

-------------------------------------------------------------------
-- Alex wrapper code.
-- A modified "posn" wrapper.
-------------------------------------------------------------------

data Posn = Pn !Int !Int !Int
  deriving (Eq, Show, Ord)

alexStartPos :: Posn
alexStartPos = Pn 0 1 1

alexMove :: Posn -> Char -> Posn
alexMove (Pn a l c) '\t' = Pn (a+1)  l     (((c+7) `div` 8)*8+1)
alexMove (Pn a l c) '\n' = Pn (a+1) (l+1)   1
alexMove (Pn a l c) _    = Pn (a+1)  l     (c+1)

type Byte = Word8

type AlexInput = (Posn,     -- current position,
                  Char,     -- previous char
                  [Byte],   -- pending bytes on the current char
                  String)   -- current input string

tokens :: String -> [Token]
tokens str = go (alexStartPos, '\n', [], str)
    where
      go :: AlexInput -> [Token]
      go inp@(pos, _, _, str) =
               case alexScan inp 0 of
                AlexEOF                   -> []
                AlexError (pos, _, _, _)  -> [Err pos]
                AlexSkip  inp' len        -> go inp'
                AlexToken inp' len act    -> act pos (take len str) : (go inp')

alexGetByte :: AlexInput -> Maybe (Byte,AlexInput)
alexGetByte (p, c, (b:bs), s) = Just (b, (p, c, bs, s))
alexGetByte (p, _, [], s) =
  case s of
    []  -> Nothing
    (c:s) ->
             let p'     = alexMove p c
                 (b:bs) = utf8Encode c
              in p' `seq` Just (b, (p', c, bs, s))

alexInputPrevChar :: AlexInput -> Char
alexInputPrevChar (p, c, bs, s) = c

-- | Encode a Haskell String to a list of Word8 values, in UTF8 format.
utf8Encode :: Char -> [Word8]
utf8Encode = map fromIntegral . go . ord
  where
  go oc
   | oc <= 0x7f       = [oc]

   | oc <= 0x7ff      = [ 0xc0 + (oc `Data.Bits.shiftR` 6)
                        , 0x80 + oc Data.Bits..&. 0x3f
                        ]

   | oc <= 0xffff     = [ 0xe0 + (oc `Data.Bits.shiftR` 12)
                        , 0x80 + ((oc `Data.Bits.shiftR` 6) Data.Bits..&. 0x3f)
                        , 0x80 + oc Data.Bits..&. 0x3f
                        ]
   | otherwise        = [ 0xf0 + (oc `Data.Bits.shiftR` 18)
                        , 0x80 + ((oc `Data.Bits.shiftR` 12) Data.Bits..&. 0x3f)
                        , 0x80 + ((oc `Data.Bits.shiftR` 6) Data.Bits..&. 0x3f)
                        , 0x80 + oc Data.Bits..&. 0x3f
                        ]
}