Compare commits

..

3 Commits

Author SHA1 Message Date
Patrick Aldis
d28638e286 Current progress 2026-03-03 11:05:11 +00:00
Patrick Aldis
902fb4e0b5 Refactor parser to decorate with void 2026-03-03 10:29:22 +00:00
Patrick Aldis
e7340f9a4d add pretty-simple 2026-03-03 10:29:01 +00:00
2 changed files with 73 additions and 53 deletions

View File

@ -1,18 +1,18 @@
{-# LANGUAGE MultilineStrings #-} {-# LANGUAGE MultilineStrings #-}
module Datalog.Parser module Datalog.Parser (
( parseTerm parseTerm,
, parseAtom parseAtom,
, parseRule parseRule,
, parseQuery parseQuery,
, parseProgram parseProgram,
) )
where where
import Datalog.Syntax
import Data.Text (Text) import Data.Text (Text)
import Data.Text qualified as T import Data.Text qualified as T
import Data.Void (Void) import Data.Void (Void)
import Datalog.Syntax
import Text.Megaparsec import Text.Megaparsec
import Text.Megaparsec.Char import Text.Megaparsec.Char
import Text.Megaparsec.Char.Lexer qualified as L import Text.Megaparsec.Char.Lexer qualified as L
@ -27,10 +27,11 @@ symbol :: (MonadParsec e Text m) => Text -> m Text
symbol = L.symbol whitespace symbol = L.symbol whitespace
whitespace :: (MonadParsec e Text m) => m () whitespace :: (MonadParsec e Text m) => m ()
whitespace = L.space whitespace =
space1 L.space
(L.skipLineComment "--") space1
(L.skipBlockComment "{-" "-}") (L.skipLineComment "--")
(L.skipBlockComment "{-" "-}")
parens :: (MonadParsec e Text m) => m a -> m a parens :: (MonadParsec e Text m) => m a -> m a
parens = between (symbol "(") (symbol ")") parens = between (symbol "(") (symbol ")")
@ -39,63 +40,72 @@ comma, period :: (MonadParsec e Text m) => m ()
comma = () <$ symbol "," comma = () <$ symbol ","
period = () <$ symbol "." period = () <$ symbol "."
parseCon :: (MonadParsec e Text m) => m (Term' SrcLoc) parseCon :: (MonadParsec e Text m) => m Term
parseCon = Con dummy . ConId . T.pack <$> lexeme (liftA2 (:) (numberChar <|> lowerChar) (many alphaNumChar)) parseCon = Con () . ConId . T.pack <$> lexeme (liftA2 (:) (numberChar <|> lowerChar) (many alphaNumChar))
parseVar :: (MonadParsec e Text m) => m (Term' SrcLoc) parseVar :: (MonadParsec e Text m) => m Term
parseVar = Var dummy . VarId . T.pack <$> lexeme (liftA2 (:) upperChar (many alphaNumChar)) parseVar = Var () . VarId . T.pack <$> lexeme (liftA2 (:) upperChar (many alphaNumChar))
parseTerm :: Parser (Term' SrcLoc) parseTerm :: Parser Term
parseTerm = parseVar <|> parseCon parseTerm = parseVar <|> parseCon
parseAtom :: Parser (Atom' SrcLoc) parseAtom :: Parser (Atom' SrcLoc)
parseAtom = do parseAtom = annotateSrcLoc $ do
p <- getSourcePos rel <- RelId . T.pack <$> lexeme (liftA2 (:) lowerChar (many alphaNumChar))
rel <- RelId . T.pack <$> lexeme (liftA2 (:) lowerChar (many alphaNumChar)) args <- parens (parseTerm `sepBy` comma)
args <- parens (parseTerm `sepBy` comma) return (Atom () rel args)
return (Atom SrcLoc{start = p} rel args)
parseQuery :: Parser [Atom' SrcLoc] parseQuery :: Parser [Atom' SrcLoc]
parseQuery = parseAtom `sepBy` comma parseQuery = parseAtom `sepBy` comma
parseFact :: Parser (Rule' SrcLoc) parseFact :: Parser (Rule' SrcLoc)
parseFact = do parseFact = annotateSrcLoc $ do
p <- getSourcePos headAtom <- parseAtom
headAtom <- parseAtom period
period return (Rule NoLoc headAtom [])
return (Rule SrcLoc{start = p} headAtom [])
parseRule :: Parser (Rule' SrcLoc) parseRule :: Parser (Rule' SrcLoc)
parseRule = try parseFact <|> do parseRule =
p <- getSourcePos annotateSrcLoc $
headAtom <- parseAtom <* symbol ":-" try parseFact <|> do
bodyAtoms <- parseQuery headAtom <- parseAtom <* symbol ":-"
period bodyAtoms <- parseQuery
return (Rule SrcLoc{start = p} headAtom bodyAtoms) period
return (Rule NoLoc headAtom bodyAtoms)
parseProgram :: Parser (Program' SrcLoc) parseProgram :: Parser (Program' SrcLoc)
parseProgram = Program dummy <$> many parseRule parseProgram = annotateSrcLoc $ Program NoLoc <$> many parseRule
data SrcLoc = SrcLoc annotateSrcLoc :: (Functor f) => Parser (f a) -> Parser (f SrcLoc)
{ start :: SourcePos annotateSrcLoc p = do
-- , end :: (Word, Word) s <- getSourcePos
} deriving Show res <- p
dummy = SrcLoc{start = initialPos "dummy-file"} f <- getSourcePos
pure (SrcLoc s f <$ res)
data SrcLoc
= SrcLoc
{ start :: SourcePos
, end :: SourcePos
}
| NoLoc
deriving (Show)
test = do test = do
let r = runParser parseProgram "???" prog let r = runParser parseProgram "???" prog
pPrint @IO r pPrint @IO r
prog = """ prog =
odd(X,Y) :- r(X,Y). """
odd(X,Y) :- even(X,Z), r(Z,Y). odd(X,Y) :- r(X,Y).
even(X,Y) :- odd(X,Z), r(Z,Y). odd(X,Y) :- even(X,Z), r(Z,Y).
even(X,Y) :- odd(X,Z), r(Z,Y).
r(0,1). r(0,1).
r(1,2). r(1,2).
r(2,3). r(2,3).
r(3,4). r(3,4).
r(4,5). r(4,5).
r(X,Y) :- r(Y,X). r(X,Y) :- r(Y,X).
""" """

View File

@ -48,6 +48,16 @@ type Program = Program' ()
data Program' a = Program a [Rule' a] data Program' a = Program a [Rule' a]
deriving (Eq, Ord, Show) deriving (Eq, Ord, Show)
class Decorable t where
decorateNode :: a -> t a -> t a
instance (Decorable Program') where decorateNode x (Program _ rs) = Program x rs
instance (Decorable Rule') where decorateNode x (Rule _ a as) = Rule x a as
instance (Decorable Atom') where decorateNode x (Atom _ relId ts) = Atom x relId ts
instance (Decorable Term') where
decorateNode x (Var _ varId) = Var x varId
decorateNode x (Con _ conId) = Con x conId
class HasConstants a where class HasConstants a where
constants :: a -> Set ConId constants :: a -> Set ConId