Parser Generator

Program, that generates parsers from LL1 input grammar with synthesized attributes.

Reads input grammar with following structure:

• Header: module name + imports

  Example:

  module Parser (parse) where
  import Ast (Expression(..))

  * Ast usually represents attributes

• Tokens: token name = token value {regexp's usage} {constructor's name}

  Example:

  plus = "\+"         { const PlusT }     { PlusT }
  minus = "-"         { const MinusT }    { MinusT }
  star = "\*"         { const StarT }     { StarT }
  slash = "\\"        { const SlashT }    { SlashT }
  op = "\("           { const OpParT }    { OpParT }
  cl = "\)"           { const ClParT }    { ClParT }
  n = "[0-9]+"        { NumT . read }     { NumT $$ }
  ws = "[[:space:]]+" ;

• "%%" to separate tokens from grammar rules

• Grammar rules: fromTerminal -> [toTerminal] {attributes' actions}

  Example:

  E -> T E'           { $2 $1 }
  E' -> plus T E'     { $3 . flip Add $2 }
  E' -> minus T E'    { $3 . flip Sub $2 }
  E' ->               { id }
  T -> F T'           { $2 $1 }
  T' -> star F T'     { $3 . flip Mul $2 }
  T' -> slash F T'    { $3 . flip Div $2 }
  T' ->               { id }
  F -> n              { Num $1 }
  F -> op E cl        { $2 }
  F -> minus F        { Neg $2 }

• data Token definition

  Example:

   {
     data Token
       = OpParT
       | ClParT
       | PlusT
       | MinusT
       | StarT
       | SlashT
       | NumT Int
       deriving Eq
   }

  * ../assets/Common.hs defines parser's instances (common for each generated parser)