Hi, Can anyone explain how to turn the code from infix to Prefix. I want to include a let statement below. I not sure of how prefix works. expr :: Parser Expr expr = buildExpressionParser table factor <?> "expression" table :: [[Operator Char st Expr]] table = [[op "*" Mul AssocLeft, op "/" Div AssocLeft] ,[op "+" Add AssocLeft, op "-" Sub AssocLeft] ] where op s f assoc = Infix (do{ string s; return f}) assoc factor :: Parser Expr factor = do{ char '(' ; x <- expr ; char ')' ; return x } <|> number <?> "simple expression" number :: Parser Expr number = do{ ds <- many1 alphaNum ; return (Val $ read ds) } John
Am Sonntag 10 Januar 2010 20:33:40 schrieb John Moore:
Hi, Can anyone explain how to turn the code from infix to Prefix. I want to include a let statement below. I not sure of how prefix works.
Can you elaborate? I don't understand what it is you want to do. Do you want to parse expressions in prefix notation (aka Polish notation), such as +(3,4) (or, more Lispy, (+ 3 4))? That'd be simple, but you wouldn't use buldExpressionParser for that (in Polish notation, neither associativity nor precedence have a meaning). Or do you want to add a parser for let-expressions to expr?
expr :: Parser Expr expr = buildExpressionParser table factor <?> "expression" table :: [[Operator Char st Expr]] table = [[op "*" Mul AssocLeft, op "/" Div AssocLeft] ,[op "+" Add AssocLeft, op "-" Sub AssocLeft] ] where op s f assoc = Infix (do{ string s; return f}) assoc factor :: Parser Expr factor = do{ char '(' ; x <- expr ; char ')' ; return x } <|> number <?> "simple expression" number :: Parser Expr number = do{ ds <- many1 alphaNum ; return (Val $ read ds) }
John
2010/1/10 John Moore
Can anyone explain how to turn the code from infix to Prefix. I want to include a let statement below. I not sure of how prefix works.
Hi John For a let expression you want to extend the 'topmost' expr parser to include a case for "let" rather than try to accommodate it in the 'table' parser. Putting a extra 'in' keyword probably makes parsing simpler, i.e. " let var = ... in ...", should you want to add other syntax later (especially function application). Once you start adding keywords you then start to be concerned about whitespace and lexing - Parsec has a couple of modules for this - Language and Token. In a nutshell where the parser previously used the 'string' parser it should use the 'symbol' parser as symbol consumes trailing whitespace. The Token module is usually imported qualified and you then redefine the functions you want to use with a particular language definition (in this case emptyDef) - the bottom of the file shows this idiom for the 'symbol', 'identifier' and 'integer' parsers. Best wishes Stephen module ExprSyn where import Text.ParserCombinators.Parsec import Text.ParserCombinators.Parsec.Expr -- new imports ... import qualified Text.ParserCombinators.Parsec.Token as P import Text.ParserCombinators.Parsec.Language (emptyDef) runExpr :: String -> IO () runExpr str = case runParser expr () "nofile" str of Left err -> putStrLn "Error:" >> print err Right val -> print val demo1 = runExpr "1+1" demo2 = runExpr "let x = 5+3 in x+7" type Variable = String data Expr = Mul Expr Expr | Div Expr Expr | Add Expr Expr | Sub Expr Expr | Val Integer | Var String | LetIn Variable Expr Expr deriving (Eq,Show) expr :: Parser Expr expr = letInExpr <|> buildExpressionParser table factor <?> "expression" letInExpr :: Parser Expr letInExpr = do { symbol "let" ; var <- identifier ; symbol "=" ; bind <- expr ; symbol "in" ; body <- expr ; return (LetIn var bind body) } table :: [[Operator Char st Expr]] table = [[op "*" Mul AssocLeft, op "/" Div AssocLeft] ,[op "+" Add AssocLeft, op "-" Sub AssocLeft] ] where op s f assoc = Infix (do{ symbol s; return f}) assoc factor :: Parser Expr factor = do{ char '(' ; x <- expr ; char ')' ; return x } <|> number <|> variable <?> "simple expression" number :: Parser Expr number = do{ n <- integer ; return (Val $ fromIntegral n) } <?> "number" variable :: Parser Expr variable = do{ cs <- identifier ; return (Var cs) } <?> "number" --- -- Bit of work to get 'symbol' and 'identifier' parsers working -- both are useful as they consume trailing white space -- Lexical analysis baseLex :: P.TokenParser st baseLex = P.makeTokenParser emptyDef symbol :: String -> CharParser st String symbol = P.symbol baseLex identifier :: CharParser st String identifier = P.identifier baseLex integer :: CharParser st Integer integer = P.integer baseLex
participants (3)
-
Daniel Fischer -
John Moore -
Stephen Tetley