module Text.ParserCombinators.PolyLazy ( -- * The Parser datatype. -- $parser Parser(P) -- datatype, instance of: Functor, Monad , runParser -- :: Parser t a -> [t] -> (a, [t]) , failBad -- :: String -> Parser t a , commit -- :: Parser t a -> Parser t a -- * Combinators: -- ** Primitives , next -- :: Parser t t , satisfy -- :: (t->Bool) -> Parser t t , apply -- :: Parser t (a->b) -> Parser t a -> Parser t b , discard -- :: Parser t a -> Parser t b -> Parser t a -- ** Error-handling , adjustErr -- :: Parser t a -> (String->String) -> Parser t a , adjustErrBad-- :: Parser t a -> (String->String) -> Parser t a , indent -- :: Int -> String -> String -- ** Choices , onFail -- :: Parser t a -> Parser t a -> Parser t a , oneOf -- :: Show t => [Parser t a] -> Parser t a , oneOf' -- :: [(String,Parser t a)] -> Parser t a , optional -- :: Parser t a -> Parser t (Maybe a) -- ** Sequences , exactly -- :: Int -> Parser t a -> Parser t [a] , many -- :: Parser t a -> Parser t [a] , many1 -- :: Parser t a -> Parser t [a] , sepBy -- :: Parser t a -> Parser t sep -> Parser t [a] , sepBy1 -- :: Parser t a -> Parser t sep -> Parser t [a] , bracketSep -- :: Parser t bra -> Parser t sep -> Parser t ket -- -> Parser t a -> Parser t [a] , bracket -- :: Parser t bra -> Parser t ket -> Parser t a -- -> Parser t a , manyFinally -- :: Parser t a -> Parser t z -> Parser t [a] -- ** Re-parsing , reparse -- :: [t] -> Parser t () ) where #if __GLASGOW_HASKELL__ import Control.Exception hiding (bracket) throwE :: String -> a throwE msg = throw (ErrorCall msg) #else throwE :: String -> a throwE msg = error msg #endif -- $parser -- When applied, these parsers do not return explicit failure. -- An exception is -- raised instead. This allows partial results to be returned -- before a full parse is complete. -- One of the key ways to ensure that your parser is properly lazy, -- is to parse the initial portion of text returning a function, then -- use the @apply@ combinator to build the final value. -- | The @Parser@ datatype is a fairly generic parsing monad with error -- reporting. It can be used for arbitrary token types, not just -- String input. (If you require a running state, use module PolyStateLazy -- instead.) newtype Parser t a = P ([t] -> (Either String a, [t])) -- A return type like Either, that distinguishes not only between -- right and wrong answers, but also had gradations of wrongness. -- Not used in this library. !!!!!!!!!!!!!!!!!!!!!!!!!!! type EitherE a b = Either (Bool,a) b -- | Apply a parser to an input token sequence. The parser cannot return -- an error value explicitly, so errors raise an exception. Thus, results -- can be partial (lazily constructed, but containing undefined). runParser :: Parser t a -> [t] -> (a, [t]) runParser (P p) = (\ (e,ts)-> (case e of {Left m->throwE m; Right x->x}, ts) ) . p instance Functor (Parser t) where fmap f (P p) = P (\ts-> case p ts of (Left msg, ts') -> (Left msg, ts') (Right x, ts') -> (Right (f x), ts')) instance Monad (Parser t) where return x = P (\ts-> (Right x, ts)) (P f) >>= g = P (\ts-> case f ts of (Left msg, ts') -> (Left msg, ts') (Right x, ts') -> let (P g') = g x in g' ts') fail e = P (\ts-> (Left e, ts)) -- | Simple failure can be corrected, but when a simple fail is not strong -- enough, use failBad for emphasis. It guarantees parsing will -- terminate with an exception. failBad :: String -> Parser t a failBad msg = P (\ts-> (throwE msg, ts)) -- | Commit is a way of raising the severity of any errors found within -- its argument. Used in the middle of a parser definition, it means that -- any operations prior to commitment fail softly, but after commitment, -- they fail hard. commit :: Parser t a -> Parser t a commit (P p) = P (\ts-> case p ts of (Left e, ts') -> (throwE e, ts') right -> right ) -- Combinators -- | One token next :: Parser t t next = P (\ts-> case ts of [] -> (Left "Ran out of input (EOF)", []) (t:ts') -> (Right t, ts') ) -- | One token satifying a predicate satisfy :: (t->Bool) -> Parser t t satisfy p = do{ x <- next ; if p x then return x else fail "Parse.satisfy: failed" } infixl 3 `apply` -- | Apply a parsed function to a parsed value apply :: Parser t (a->b) -> Parser t a -> Parser t b --pf `apply` px = do { f <- pf; x <- px; return (f x) } -- Needs to be lazier! Must not force the argument value too early. (P pf) `apply` (P px) = P (\ts-> case pf ts of (Left msg, ts') -> (Left msg, ts') (Right f, ts') -> let (x',ts'') = px ts' x = case x' of { Right x -> x; Left e -> throwE e } in (Right (f x), ts'') ) infixl 3 `discard` -- | @x `discard` y@ parses both x and y, but discards the result of y discard :: Parser t a -> Parser t b -> Parser t a --px `discard` py = do { x <- px; _ <- py; return x } -- Needs to be lazier! Do not force the discarded portion too early. px `discard` py = do { x <- px; return (\_-> x) `apply` py } -- | @p `adjustErr` f@ applies the transformation @f@ to any error message -- generated in @p@, having no effect if @p@ succeeds. adjustErr :: Parser t a -> (String->String) -> Parser t a (P p) `adjustErr` f = P (\ts-> case p ts of (Left msg, ts') -> (Left (f msg), ts') right -> right ) -- | @adjustErrBad@ is just like @adjustErr@ except it also raises the -- severity of the error. adjustErrBad :: Parser t a -> (String->String) -> Parser t a p `adjustErrBad` f = commit (p `adjustErr` f) infixl 6 `onFail` -- not sure about precedence 6? -- | @p `onFail` q@ means parse p unless p fails in which case parse q instead. -- Can be chained together to give multiple attempts to parse something. -- (Note that q could itself be a failing parser, e.g. to change the error -- message from that defined in p to something different.) -- However, a *severe* failure in p cannot be ignored. onFail :: Parser t a -> Parser t a -> Parser t a (P p) `onFail` (P q) = P (\ts-> case p ts of (Left _, _) -> q ts right -> right ) -- | Parse the first alternative in the list that succeeds. oneOf :: [Parser t a] -> Parser t a oneOf [] = do { n <- next ; fail ("failed to parse any of the possible choices") } --oneOf :: Show t => [Parser t a] -> Parser t a --oneOf [] = do { n <- next -- ; fail ("failed to parse any of the possible choices" -- ++"\n next token is "++show n) -- } oneOf (p:ps) = p `onFail` oneOf ps -- | Parse the first alternative that succeeds, but if none succeed, -- report only the severe errors, and if none of those, then report -- all the soft errors. oneOf' :: [(String, Parser t a)] -> Parser t a oneOf' ps = accum [] ps where accum errs [] = case errs of [] -> failBad ("internal failure in parser (oneOf'):\n" ++indent 2 (show (map fst ps))) [(_,e)] -> fail e es -> fail ("one of the following failures occurred:\n" ++indent 2 (concatMap showErr (reverse es))) accum errs ((e,P p):ps) = P (\ts-> case p ts of (Left err,_) -> let (P p) = accum ((e,err):errs) ps in p ts right -> right ) showErr (name,err) = name++":\n"++indent 2 err -- | Helper for formatting error messages: indents all lines by a fixed amount. indent :: Int -> String -> String indent n = unlines . map (replicate n ' ' ++) . lines -- | 'optional' indicates whether the parser succeeded through the Maybe type. optional :: Parser t a -> Parser t (Maybe a) optional p = fmap Just p `onFail` return Nothing -- | 'exactly n p' parses a precise number of items, n, using the parser -- p, in sequence. exactly :: Int -> Parser t a -> Parser t [a] exactly 0 p = return [] exactly n p = return (:) `apply` p `apply` exactly (n-1) p -- | 'many p' parses a list of elements with individual parser p. -- Cannot fail, since an empty list is a valid return value. many :: Parser t a -> Parser t [a] many p = many1 p `onFail` return [] where many1 p = do { x <- p ; return (x:) `apply` many p } -- | Parse a non-empty list of items. many1 :: Parser t a -> Parser t [a] many1 p = do { x <- p `adjustErr` (("In a sequence:\n"++). indent 2) ; return (x:) `apply` many p } -- `adjustErr` ("When looking for a non-empty sequence:\n\t"++) -- | Parse a list of items separated by discarded junk. sepBy :: Parser t a -> Parser t sep -> Parser t [a] sepBy p sep = do sepBy1 p sep `onFail` return [] -- | Parse a non-empty list of items separated by discarded junk. sepBy1 :: Parser t a -> Parser t sep -> Parser t [a] sepBy1 p sep = do { x <- p ; return (x:) `apply` many (do {sep; p}) } `adjustErr` ("When looking for a non-empty sequence with separators:\n\t"++) -- | Parse a list of items, discarding the start, end, and separator -- items. bracketSep :: Parser t bra -> Parser t sep -> Parser t ket -> Parser t a -> Parser t [a] bracketSep open sep close p = do { open; close; return [] } `onFail` do { open `adjustErr` ("Missing opening bracket:\n\t"++) ; x <- p `adjustErr` ("After first bracket in a group:\n\t"++) ; return (x:) `apply` manyFinally (do {sep; p}) (close `adjustErrBad` ("When looking for closing bracket:\n\t"++)) } -- | Parse a bracketed item, discarding the brackets. bracket :: Parser t bra -> Parser t ket -> Parser t a -> Parser t a bracket open close p = do do { open `adjustErr` ("Missing opening bracket:\n\t"++) ; p `discard` (close `adjustErrBad` ("Missing closing bracket:\n\t"++)) } -- | 'manyFinally e t' parses a possibly-empty sequence of e's, -- terminated by a t. Any parse failures could be due either to -- a badly-formed terminator or a badly-formed element, so raise -- both possible errors. manyFinally :: Parser t a -> Parser t z -> Parser t [a] manyFinally pp@(P p) pt@(P t) = P (\ts -> case p ts of (Left e, _) -> case t ts of (Right _, ts') -> (Right [], ts') (Left e, ts') -> (Left e, ts') (Right x, ts') -> let (tail,ts'') = runParser (manyFinally pp pt) ts' in (Right (x:tail), ts'') ) ------------------------------------------------------------------------ -- | Push some tokens back onto the front of the input stream and reparse. -- This is useful e.g. for recursively expanding macros. When the -- user-parser recognises a macro use, it can lookup the macro -- expansion from the parse state, lex it, and then stuff the -- lexed expansion back down into the parser. reparse :: [t] -> Parser t () reparse ts = P (\inp-> (Right (), ts++inp)) ------------------------------------------------------------------------