crafting-interpreters-hs/src/Lox/Parser.hs

224 lines
7.9 KiB
Haskell
Raw Normal View History

2025-12-25 17:55:28 +03:00
module Lox.Parser (
2025-12-29 23:57:15 +03:00
ParserError (..),
2025-12-25 18:01:59 +03:00
parse
2025-12-25 17:55:28 +03:00
) where
import Control.Monad
import Control.Monad.State
2025-12-30 02:56:06 +03:00
import Data.Either
import Data.Maybe
2025-12-25 17:55:28 +03:00
import Lox.Scanner
import Lox.Expr
data ParserState = ParserState {tokens :: [Token]}
2025-12-30 01:27:01 +03:00
data ParserError = MismatchedParenthesesError
| ExpectedExpressionError
| ExpectedSemicolonError
2025-12-30 02:56:06 +03:00
| ExpectedVariableName
| InvalidAssignmentTarget
2025-12-30 01:27:01 +03:00
deriving Show
2025-12-30 02:56:06 +03:00
-- program → declaration* EOF ;
--
-- declaration → varDecl
-- | statement ;
2025-12-30 01:27:01 +03:00
--
-- statement → exprStmt
-- | printStmt ;
--
-- exprStmt → expression ";" ;
-- printStmt → "print" expression ";" ;
2025-12-30 02:56:06 +03:00
-- varDecl → "var" IDENTIFIER ( "=" expression )? ";" ;
2025-12-30 01:27:01 +03:00
2025-12-30 02:56:06 +03:00
-- expression → assignment ;
-- assignment → IDENTIFIER "=" assignment
-- | equality ;
2025-12-25 17:55:28 +03:00
-- equality → comparison ( ( "!=" | "==" ) comparison )* ;
-- comparison → term ( ( ">" | ">=" | "<" | "<=" ) term )* ;
-- term → factor ( ( "-" | "+" ) factor )* ;
-- factor → unary ( ( "/" | "*" ) unary )* ;
-- unary → ( "!" | "-" ) unary
-- | primary ;
-- primary → NUMBER | STRING | "true" | "false" | "nil"
2025-12-30 02:56:06 +03:00
-- | "(" expression ")" | IDENTIFIER;
2025-12-25 17:55:28 +03:00
2025-12-30 01:27:01 +03:00
parse :: [Token] -> Either ParserError [Stmt]
parse tokens = evalState program (ParserState {tokens=tokens})
program :: State ParserState (Either ParserError [Stmt])
program = do
atEnd <- isAtEnd
if atEnd then return $ Right [] else do
2025-12-30 02:56:06 +03:00
headMaybe <- declaration
2025-12-30 01:27:01 +03:00
case headMaybe of
Left err -> return $ Left err
Right head -> do
tailMaybe <- program
case tailMaybe of
Left err -> return $ Left err
Right tail -> return $ Right $ head : tail
2025-12-30 02:56:06 +03:00
declaration :: State ParserState (Either ParserError Stmt)
declaration = do
varMaybe <- matchToken [VAR]
case varMaybe of
Just _ -> varDeclaration
_ -> statement
varDeclaration :: State ParserState (Either ParserError Stmt)
varDeclaration = do
maybeName <- consume IDENTIFIER ExpectedVariableName
case maybeName of
Left err -> return $ Left err
Right name -> do
hasInit <- isJust <$> matchToken [EQUAL]
initMaybe <- if hasInit then expression else return $ Right $ LiteralExpr NullObject
semicolonMaybe <- consume SEMICOLON ExpectedSemicolonError
case (initMaybe, semicolonMaybe) of
(Left err, _) -> return $ Left err
(_, Left err) -> return $ Left err
(Right init, Right _) -> return $ Right $ VariableStmt name init
2025-12-30 01:27:01 +03:00
statement :: State ParserState (Either ParserError Stmt)
statement = do
printMaybe <- matchToken [PRINT]
case printMaybe of
Just _ -> printStatement
_ -> expressionStatement
printStatement :: State ParserState (Either ParserError Stmt)
printStatement = do
valueMaybe <- expression
semicolonMaybe <- consume SEMICOLON ExpectedSemicolonError
case (valueMaybe, semicolonMaybe) of
(Left err, _) -> return $ Left err
(_, Left err) -> return $ Left err
2025-12-30 02:56:06 +03:00
(Right value, Right _) -> return $ Right $ PrintStmt value
2025-12-30 01:27:01 +03:00
expressionStatement :: State ParserState (Either ParserError Stmt)
expressionStatement = do
valueMaybe <- expression
semicolonMaybe <- consume SEMICOLON ExpectedSemicolonError
case (valueMaybe, semicolonMaybe) of
(Left err, _) -> return $ Left err
(_, Left err) -> return $ Left err
2025-12-30 02:56:06 +03:00
(Right value, Right _) -> return $ Right $ ExpressionStmt value
2025-12-30 01:27:01 +03:00
2025-12-25 17:55:28 +03:00
2025-12-29 23:57:15 +03:00
expression :: State ParserState (Either ParserError Expr)
2025-12-30 02:56:06 +03:00
expression = assignment
assignment :: State ParserState (Either ParserError Expr)
assignment = do
maybeExpr <- equality
matchedEqual <- isJust <$> matchToken [EQUAL]
if matchedEqual then do
maybeValue <- assignment
case (maybeExpr, maybeValue) of
(Left err, _) -> return $ Left err
(_, Left err) -> return $ Left err
(Right (VariableExpr name), Right value) -> return $ Right $ AssignmentExpr name value
_ -> return $ Left InvalidAssignmentTarget
else return maybeExpr
2025-12-25 17:55:28 +03:00
2025-12-29 23:57:15 +03:00
equality :: State ParserState (Either ParserError Expr)
2025-12-25 17:55:28 +03:00
equality = do
2025-12-29 23:57:15 +03:00
exprMaybe <- comparison
case exprMaybe of
Left err -> return $ Left err
Right expr -> fmap (mergeExpressionMaybe expr) <$> matchTail [BANG_EQUAL, EQUAL_EQUAL] comparison
2025-12-25 17:55:28 +03:00
2025-12-29 23:57:15 +03:00
comparison :: State ParserState (Either ParserError Expr)
2025-12-25 17:55:28 +03:00
comparison = do
2025-12-29 23:57:15 +03:00
exprMaybe <- term
case exprMaybe of
Left err -> return $ Left err
Right expr -> fmap (mergeExpressionMaybe expr) <$> matchTail [GREATER, GREATER_EQUAL, LESS, LESS_EQUAL] term
2025-12-25 17:55:28 +03:00
2025-12-29 23:57:15 +03:00
term :: State ParserState (Either ParserError Expr)
2025-12-25 17:55:28 +03:00
term = do
2025-12-29 23:57:15 +03:00
exprMaybe <- factor
case exprMaybe of
Left err -> return $ Left err
Right expr -> fmap (mergeExpressionMaybe expr) <$> matchTail [MINUS, PLUS] factor
2025-12-25 17:55:28 +03:00
2025-12-29 23:57:15 +03:00
factor :: State ParserState (Either ParserError Expr)
2025-12-25 17:55:28 +03:00
factor = do
2025-12-29 23:57:15 +03:00
exprMaybe <- unary
case exprMaybe of
Left err -> return $ Left err
Right expr -> fmap (mergeExpressionMaybe expr) <$> matchTail [SLASH, STAR] unary
2025-12-25 17:55:28 +03:00
2025-12-29 23:57:15 +03:00
unary :: State ParserState (Either ParserError Expr)
2025-12-25 17:55:28 +03:00
unary = do
maybeOperator <- matchToken [BANG, MINUS]
case maybeOperator of
Nothing -> primary
2025-12-29 23:57:15 +03:00
Just op -> do
exprMaybe <- unary
case exprMaybe of
Left err -> return $ Left err
2025-12-30 02:56:06 +03:00
Right expr -> return $ Right $ UnaryExpr op expr
2025-12-25 17:55:28 +03:00
2025-12-29 23:57:15 +03:00
primary :: State ParserState (Either ParserError Expr)
2025-12-25 17:55:28 +03:00
primary = do
token <- advance
2025-12-25 18:48:12 +03:00
case tokenType token of
2025-12-30 02:56:06 +03:00
FALSE -> return $ Right $ LiteralExpr $ BoolObject False
TRUE -> return $ Right $ LiteralExpr $ BoolObject True
NIL -> return $ Right $ LiteralExpr NullObject
NUMBER -> return $ Right $ LiteralExpr $ tokenObject token
STRING -> return $ Right $ LiteralExpr $ tokenObject token
2025-12-25 17:55:28 +03:00
LEFT_PAREN -> do
2025-12-29 23:57:15 +03:00
exprMaybe <- expression
case exprMaybe of
Left err -> return $ Left err
Right expr -> do
consume RIGHT_PAREN MismatchedParenthesesError
2025-12-30 02:56:06 +03:00
return $ Right $ GroupingExpr expr
IDENTIFIER -> return $ Right $ VariableExpr token
2025-12-29 23:57:15 +03:00
_ -> return $ Left ExpectedExpressionError
matchTail :: [TokenType] -> State ParserState (Either ParserError Expr) -> State ParserState (Either ParserError (Maybe (Token, Expr)))
2025-12-25 17:55:28 +03:00
matchTail tokenTypes f = do
maybeOperator <- matchToken tokenTypes
case maybeOperator of
2025-12-29 23:57:15 +03:00
Nothing -> return $ Right Nothing
2025-12-25 17:55:28 +03:00
Just op -> do
2025-12-29 23:57:15 +03:00
exprMaybe <- comparison
restMaybe <- matchTail tokenTypes f
case (exprMaybe, restMaybe) of
(Left err, _) -> return $ Left err
(_, Left err) -> return $ Left err
(Right expr, Right rest) -> return $ Right $ Just (op, mergeExpressionMaybe expr rest)
2025-12-25 17:55:28 +03:00
mergeExpressionMaybe :: Expr -> Maybe (Token, Expr) -> Expr
mergeExpressionMaybe expr Nothing = expr
2025-12-30 02:56:06 +03:00
mergeExpressionMaybe left (Just (op, right)) = BinaryExpr left op right
2025-12-25 17:55:28 +03:00
matchToken :: [TokenType] -> State ParserState (Maybe Token)
matchToken [] = return Nothing
matchToken (t:ts) = do
isMatch <- check t
if isMatch then Just <$> advance else matchToken ts
check :: TokenType -> State ParserState Bool
check t = do
atEnd <- isAtEnd
2025-12-25 18:48:12 +03:00
if atEnd then return False else (== t) . tokenType <$> peek
2025-12-25 17:55:28 +03:00
2025-12-29 23:57:15 +03:00
consume :: TokenType -> ParserError -> State ParserState (Either ParserError Token)
consume t err = do
2025-12-25 17:55:28 +03:00
isOk <- check t
2025-12-29 23:57:15 +03:00
if isOk then Right <$> advance else return $ Left err
2025-12-25 17:55:28 +03:00
advance :: State ParserState Token
advance = state (\s@ParserState {tokens=(t:ts)} -> (t, s {tokens = ts}))
peek :: State ParserState Token
peek = gets (head . tokens)
isAtEnd :: State ParserState Bool
2025-12-25 18:48:12 +03:00
isAtEnd = (== EOF) . tokenType <$> peek