implementation module parse
+import GenPrint
import StdString
import StdTuple
import StdList
from StdFunc import const, o
import Data.Either
-import Data.Functor
import Data.Maybe
+import Data.Functor
+import Data.Tuple
import Control.Monad
import Control.Applicative
import Data.Func
+import StdMisc
from Data.List import intercalate, replicate, instance Functor []
from Text import class Text(concat), instance Text String
-import GenPrint
import yard
import lex
+import AST
-parser :: LexerOutput -> ParserOutput
-parser (Left e) = Left e
-parser (Right r) = fst $ runParser parseProgram r
+parser :: [Token] -> ParserOutput
+parser ts = case runParser parseProgram ts of
+ (Right ast, [(p, t):xs]) = Left $ PositionalError p.line p.col (
+ "Unable to parse from: " +++ printToString t)
+ x = fst x
parseProgram :: Parser Token AST
-parseProgram = liftM2 AST (many parseVarDecl) (some parseFunDecl)
+parseProgram = many parseLetDecl >>= \fds1->
+ some parseFunDecl >>= \fds2->
+ pure $ AST (fds1++fds2)
+
+parseLetDecl :: Parser Token FunDecl
+parseLetDecl = peekPos >>= \p->
+ satTok LetToken >>|
+ parseFunType >>= \mt->
+ parseIdent >>= \f->
+ satTok AssignmentToken >>|
+ parseExpr >>= \e->
+ satTok SColonToken >>|
+ pure (FunDecl p f [] (Just mt) [] [ReturnStmt $ Just e])
parseFunDecl :: Parser Token FunDecl
-parseFunDecl = liftM5 FunDecl
- (parseIdent <* satTok BraceOpenToken)
- (parseSepList CommaToken parseIdent <* satTok BraceCloseToken)
- (optional parseFunType <* satTok CBraceOpenToken)
- (many parseVarDecl)
- (many parseStmt <* satTok CBraceCloseToken)
-
-parseStmt :: Parser Token Stmt
-parseStmt = parseIfStmt <|> parseWhileStmt <|>
+parseFunDecl = liftM6 FunDecl
+ (peekPos)
+ (parseIdent)
+ (parseBBraces $ parseSepList CommaToken parseIdent)
+ (optional (satTok DoubleColonToken *> parseFunType))
+ (satTok CBraceOpenToken *> many parseVarDecl)
+ (flatten <$> (many parseStmt <* satTok CBraceCloseToken))
+
+parseStmt :: Parser Token [Stmt]
+parseStmt = (parseIfStmt <|> parseWhileStmt <|>
parseSColon parseAssStmt <|> parseSColon parseReturnStmt <|>
- (liftM FunStmt (parseSColon parseFunCall))
+ (parseSColon parseFunCall
+ >>= \(ident, args, fs)->pure $ FunStmt ident args fs))
+ >>= \stmt->case stmt of
+ FunStmt "print" args fs = pure $ map (\a->FunStmt "print" [a] []) args
+ s = pure [s]
where
parseSColon :: (Parser Token a) -> Parser Token a
parseSColon p = p <* satTok SColonToken
satTok ReturnToken *> liftM ReturnStmt (optional parseExpr)
parseAssStmt :: Parser Token Stmt
- parseAssStmt =
- liftM2 AssStmt (parseVarDef <* satTok AssignmentToken) parseExpr
+ parseAssStmt =
+ AssStmt <$> (parseVarDef <* satTok AssignmentToken) <*> parseExpr
parseIfStmt :: Parser Token Stmt
parseIfStmt = liftM3 IfStmt
(satTok IfToken *> parseBBraces parseExpr)
(parseBlock <|> parseOneLine)
- (liftM (fromMaybe [])
+ (liftM (fromMaybe [])
(optional (satTok ElseToken *> (parseBlock<|> parseOneLine))))
-
parseWhileStmt :: Parser Token Stmt
- parseWhileStmt = satTok WhileToken *>
- liftM2 WhileStmt (parseBBraces parseExpr) parseBlock
+ parseWhileStmt = satTok WhileToken *> (WhileStmt <$>
+ (parseBBraces parseExpr) <*> (parseBlock <|> parseOneLine))
parseBlock :: Parser Token [Stmt]
- parseBlock = parseBCBraces (many parseStmt)
+ parseBlock = parseBCBraces (flatten <$> many parseStmt)
parseOneLine :: Parser Token [Stmt]
- //first pure makes singleton list from the statement
- parseOneLine = liftM pure parseStmt
+ parseOneLine = parseStmt
-parseFunType :: Parser Token FunType
-parseFunType = satTok DoubleColonToken *>
- (parseInOutType <|> (liftM (FunType []) parseVoidOrType))
+parseFunType :: Parser Token Type
+parseFunType = parseFT >>= \t -> case t of
+ t1 ->> t2 = pure t
+ simpleT = pure $ FuncType simpleT
where
- parseInOutType :: Parser Token FunType
- parseInOutType = liftM2 FunType
- (some parseType <* satTok ArrowToken) parseVoidOrType
-
- parseVoidOrType :: Parser Token (Maybe Type)
- parseVoidOrType = (satTok VoidToken *> pure Nothing) <|>
- (liftM Just parseType) <|> pure Nothing
+ parseFT :: Parser Token Type
+ parseFT = (liftM2 (->>) (parseSF <* satTok ArrowToken) (parseFT)) <|>
+ parseSF
+ parseSF :: Parser Token Type
+ parseSF = parseBBraces parseFT <|> parseType
parseVarDecl :: Parser Token VarDecl
-parseVarDecl = liftM3 VarDecl
- (parseType <|> trans1 VarToken VarType )
+parseVarDecl = liftM4 VarDecl
+ peekPos
+ ((parseType >>= \t->pure $ Just t)<|> trans1 VarToken Nothing)
(parseIdent <* satTok AssignmentToken)
(parseExpr <* satTok SColonToken)
parseType :: Parser Token Type
-parseType =
+parseType =
trans1 IntTypeToken IntType <|>
trans1 CharTypeToken CharType <|>
trans1 BoolTypeToken BoolType <|>
- (liftM ListType (parseBSqBraces parseType)) <|>
- (liftM TupleType (parseTuple parseType)) <|>
- (liftM IdType parseIdent)
+ trans1 VoidToken VoidType <|>
+ (ListType <$> (parseBSqBraces parseType)) <|>
+ (TupleType <$> (parseTuple parseType)) <|>
+ (IdType <$> parseIdent)
parseExpr :: Parser Token Expr
-parseExpr = //Operators in order of binding strength
+parseExpr = parseValueExpr <|> parseLambda
+
+parseValueExpr :: Parser Token Expr
+parseValueExpr = //Operators in order of binding strength
parseOpR (trans1 ColonToken BiCons) $
parseOpR (trans1 PipesToken BiOr) $
parseOpR (trans1 AmpersandsToken BiAnd) $
trans1 PercentToken BiMod) $ parseBasicExpr
where
parseOpR :: (Parser Token Op2) (Parser Token Expr) -> Parser Token Expr
- parseOpR ops prev = prev >>= \e1->optional (
+ parseOpR ops prev = peekPos >>= \pos-> prev >>= \e1->optional (
ops >>= \op->parseOpR ops prev >>= \e->pure (op, e)
- ) >>= \moe->pure $ maybe e1 (\(op,e2)->Op2Expr e1 op e2) moe
+ ) >>= \moe->pure $ maybe e1 (\(op,e2)->Op2Expr pos e1 op e2) moe
parseOpL :: (Parser Token Op2) (Parser Token Expr) -> Parser Token Expr
- parseOpL ops prev = prev >>= \e1->many (
- ops >>= \op->prev >>= \e->pure (op, e)
- ) >>= \moe->foldM (\e->(\(op,e2)->pure $ Op2Expr e op e2)) e1 moe
+ parseOpL ops prev = peekPos >>= \pos-> prev >>= \e1->many (
+ ops >>= \op->prev >>= \e->pure (op, e))
+ >>= \moe->foldM (\e->(\(op,e2)->pure $ Op2Expr pos e op e2)) e1 moe
parseBasicExpr :: Parser Token Expr
- parseBasicExpr =
- (liftM TupleExpr (parseTuple parseExpr)) <|>
- (liftM FunExpr parseFunCall) <|>
+ parseBasicExpr = peekPos >>= \pos ->
+ (trans2 (StringToken []) (\(StringToken cs)->makeStrExpr pos cs)) <|>
+ (TupleExpr pos <$> (parseTuple parseExpr)) <|>
parseBBraces parseExpr <|>
- trans1 EmptyListToken EmptyListExpr <|>
- trans1 TrueToken (BoolExpr True) <|>
- trans1 FalseToken (BoolExpr False) <|>
- trans2 (NumberToken zero) (\(NumberToken i)->IntExpr i) <|>
- trans2 (CharToken zero) (\(CharToken c)->CharExpr c) <|>
- (liftM2 Op1Expr parseOp1 parseExpr) <|>
- (liftM VarExpr parseVarDef)
-
-parseFunCall :: Parser Token FunCall
-parseFunCall = liftM2 FunCall
- parseIdent (parseBBraces $ parseSepList CommaToken parseExpr)
+ parseListLiteral <|>
+ trans1 EmptyListToken (EmptyListExpr pos) <|>
+ trans1 TrueToken (BoolExpr pos True) <|>
+ trans1 FalseToken (BoolExpr pos False) <|>
+ trans2 (NumberToken zero) (\(NumberToken i)->IntExpr pos i) <|>
+ trans2 (CharToken zero) (\(CharToken c)->CharExpr pos c) <|>
+ (Op1Expr pos <$> parseOp1 <*> parseExpr) <|>
+ (parseFunCall >>= \(ident, args, fs)->
+ pure $ FunExpr pos ident args fs) <|>
+ (VarExpr pos <$> parseVarDef)
+
+ parseListLiteral :: Parser Token Expr
+ parseListLiteral = peekPos >>= \p->
+ satTok SquareOpenToken >>|
+ many (parseExpr <* satTok CommaToken) >>= \es->
+ parseExpr >>= \e->
+ foldM (\res e-> pure $ Op2Expr p e BiCons res)
+ (EmptyListExpr p)
+ [e : reverse es] >>= \res->
+ satTok SquareCloseToken >>|
+ pure res
+
+parseLambda :: Parser Token Expr
+parseLambda = LambdaExpr <$> peekPos
+ <*> (satTok BackslashToken *> many parseIdent)
+ <*> (satTok ArrowToken *> parseExpr)
+
+makeStrExpr :: Pos [Char] -> Expr
+makeStrExpr p [] = EmptyListExpr p
+makeStrExpr p [x:xs] = Op2Expr p (CharExpr zero x) BiCons (makeStrExpr p xs)
+
+parseFunCall :: Parser Token (String, [Expr], [FieldSelector])
+parseFunCall = liftM3 (\x y z->(x, y, z))
+ parseIdent
+ (parseBBraces $ parseSepList CommaToken parseExpr)
+ parseFieldSelectors
parseVarDef :: Parser Token VarDef
-parseVarDef = liftM2 VarDef
- parseIdent
- (many (satTok DotToken *> (
- (parseIdent >>= (\i.if (i == "hd") (pure FieldHd) empty)) <|>
- (parseIdent >>= \i.if (i == "tl") (pure FieldTl) empty) <|>
- (parseIdent >>= \i.if (i == "fst") (pure FieldFst) empty) <|>
- (parseIdent >>= \i.if (i == "snd") (pure FieldSnd) empty))))
+parseVarDef = liftM2 VarDef parseIdent parseFieldSelectors
+
+parseFieldSelectors :: Parser Token [FieldSelector]
+parseFieldSelectors = many (satTok DotToken *>
+ parseIdent >>= \i->case i of
+ "hd" = pure FieldHd
+ "tl" = pure FieldTl
+ "fst" = pure FieldFst
+ "snd" = pure FieldSnd
+ _ = empty)
parseOp1 :: Parser Token Op1
parseOp1 = trans1 DashToken UnMinus <|> trans1 ExclamationToken UnNegation
parseBSqBraces p = satTok SquareOpenToken *> p <* satTok SquareCloseToken
parseTuple :: (Parser Token a) -> Parser Token (a, a)
-parseTuple p = satTok BraceOpenToken *>
- (liftM2 (\a->(\b->(a,b))) (p <* satTok CommaToken) p)
+parseTuple p = satTok BraceOpenToken *>
+ (liftM2 tuple (p <* satTok CommaToken) p)
<* satTok BraceCloseToken
+trans :: TokenValue (TokenValue -> a) -> Parser Token (Pos, a)
+trans t f = (\(pos,token)->(pos, f token)) <$> satTok t
+
trans2 :: TokenValue (TokenValue -> a) -> Parser Token a
-trans2 t f = liftM (\{token}->f token) $ satTok t
+trans2 t f = snd <$> trans t f
trans1 :: TokenValue a -> Parser Token a
trans1 t r = trans2 t $ const r
+peekPos :: Parser Token Pos
+peekPos = fst <$> peek
+
derive gPrint TokenValue
derive gEq TokenValue
satTok :: TokenValue -> Parser Token Token
-satTok t = top >>= \tok=:{line,column,token} -> if (eq t token)
- (return tok) (fail <?> PositionalError line column ("ParseError: Unexpected token: " +++ printToString token))
- where
- eq (IdentToken _) (IdentToken _) = True
- eq (NumberToken _) (NumberToken _) = True
- eq (CharToken _) (CharToken _) = True
- eq x y = gEq {|*|} x y
+satTok t = top >>= \tok=:({line,col},token) -> if (tokEq t token)
+ (pure tok) (fail <?> PositionalError line col
+ ("ParseError: Unexpected token: " +++ printToString token
+ +++ "\nExpected: " +++ printToString t))
+
+tokEq (IdentToken _) (IdentToken _) = True
+tokEq (NumberToken _) (NumberToken _) = True
+tokEq (CharToken _) (CharToken _) = True
+tokEq (StringToken _) (StringToken _) = True
+tokEq x y = gEq {|*|} x y
parseSepList :: TokenValue (Parser Token a) -> Parser Token [a]
-parseSepList sep p =
- (liftM2 (\es->(\e->reverse [e:es])) (some (p <* satTok sep)) p) <|>
- (liftM pure p) <|> pure empty
+parseSepList sep p = pSL` sep p <|> pure []
+ where
+ pSL` sep p = (p <* satTok sep >>= \v1->parseSepList sep p >>= \vs->pure [v1:vs]) <|> (p >>= \v->pure [v])
parseIdent :: Parser Token String
-parseIdent = trans2 (IdentToken "") (\(IdentToken e).toString e)
-
-instance toString AST where
- toString (AST v f) = concat (
- ["\n":printersperse "\n" v] ++
- ["\n":printersperse "\n" f])
-
-class print a :: a -> [String]
-
-printersperse :: String [a] -> [String] | print a
-printersperse i j = intercalate [i] (map print j)
-
-instance print FunDecl where
- print (FunDecl i as t vs ss) =
- ["\n", i, " (":printersperse "," as] ++
- [")"] ++ maybe [] (\tt->[" :: ":print tt]) t ++
- ["{\n\t":printersperse "\n\t" vs] ++
- ["\n":printStatements ss 1] ++ ["}\n"]
-
-printStatements :: [Stmt] Int -> [String]
-printStatements [] i = []
-printStatements [s:ss] i = (case s of
- (IfStmt b thens elses) = indent i ["if (":print b] ++ [")"] ++
- printCodeBlock thens i ++
- indent i ["else ":printCodeBlock elses i] ++ ["\n"]
- (WhileStmt b dos) = indent i ["while (":print b] ++
- [")":printCodeBlock dos i]
- (AssStmt vardef val) =
- indent i $ print vardef ++ ["=":print val] ++ [";\n"]
- (FunStmt fc) = indent i $ print fc ++ [";\n"]
- (ReturnStmt me) = indent i ["return ":maybe [""] print me] ++ [";\n"]
- ) ++ printStatements ss i
- where
- printCodeBlock :: [Stmt] Int -> [String]
- printCodeBlock [] _ = ["{}\n"]
- printCodeBlock [x] i = ["\n":printStatements [x] (i+1)]
- printCodeBlock x i =
- ["{\n":printStatements x (i+1)] ++ indent i ["}\n"]
-
- indent :: Int [String] -> [String]
- indent i rest = replicate i "\t" ++ rest
-
-instance print FunType where
- print (FunType at rt) = printersperse " " at ++
- [if (isEmpty at) "" "->":maybe ["Void"] print rt]
-
-instance print VarDecl where
- print (VarDecl t i e) = print t ++ [" ":i:"=":print e] ++ [";"]
-
-instance print Type where
- print (TupleType (t1, t2)) = ["(":print t1] ++ [",":print t2] ++ [")"]
- print (ListType t) = ["[":print t] ++ ["]"]
- print (IdType s) = print s
- print IntType = print "Int"
- print BoolType = print "Bool"
- print CharType = print "Char"
- print VarType = print "var"
-
-instance print String where
- print s = [s]
-
-instance print FieldSelector where
- print FieldHd = print "hd"
- print FieldTl = print "tl"
- print FieldSnd = print "snd"
- print FieldFst = print "fst"
-
-instance print VarDef where
- print (VarDef i fs) = printersperse "." [i:flatten $ map print fs]
-
-instance print FunCall where
- print (FunCall i args) = [i,"(":printersperse "," args] ++ [")"]
-
-instance print Expr where
- print (VarExpr vd) = print vd
- print (Op2Expr e1 o e2) = ["(":print e1] ++ [" ",case o of
- BiPlus = "+"; BiMinus = "-"; BiTimes = "*"; BiDivide = "/"
- BiMod = "%"; BiEquals = "="; BiLesser = "<"; BiGreater = ">"
- BiLesserEq = "<="; BiGreaterEq = ">="; BiUnEqual = "!=";
- BiAnd = "&&"; BiOr = "||"; BiCons = ":"
- ," ":print e2] ++ [")"]
- print (Op1Expr o e) = ["(",case o of
- UnNegation = "!"; UnMinus = "-"
- :print e] ++ [")"]
- print (IntExpr i) = [toString i]
- print (CharExpr c) = ["\'", case c of
- '\b' = "\\b"; '\f' = "\\f"; '\n' = "\\n"
- '\r' = "\\r"; '\t' = "\\t"; '\v' = "\\v"
- c = if (c == toChar 7) "\\a" (toString c)
- ,"\'"]
- print (BoolExpr b) = [toString b]
- print (FunExpr fc) = print fc
- print EmptyListExpr = ["[]"]
- print (TupleExpr (e1, e2)) = ["(":print e1] ++ [",":print e2] ++ [")"]
+parseIdent = trans2 (IdentToken "") (\(IdentToken e)->toString e)
+
+//liftM only goes to liftM5
+liftM6 f m1 m2 m3 m4 m5 m6 = f <$> m1 <*> m2 <*> m3 <*> m4 <*> m5 <*> m6
repositories / cc1516.git / blobdiff