+-- |Yet another parser combinator. This is mostly a subset of
+-- "Text.ParserCombinators.Parsec" but there are some differences:
+--
+-- * This parser works on 'Data.ByteString.Base.LazyByteString'
+-- instead of 'Prelude.String'.
+--
+-- * Backtracking is the only possible behavior so there is no \"try\"
+-- action.
+--
+-- * On success, the remaining string is returned as well as the
+-- parser result.
+--
+-- * You can choose whether to treat reaching EOF (trying to eat one
+-- more letter at the end of string) a fatal error or to treat it a
+-- normal failure. If a fatal error occurs, the entire parsing
+-- process immediately fails without trying any backtracks. The
+-- default behavior is to treat EOF fatal.
+--
+-- In general, you don't have to use this module directly.
module Network.HTTP.Lucu.Parser
- ( Parser(..)
- , parse -- Parser a -> ByteString -> Maybe (a, ByteString)
- , anyChar -- Parser Char
- , satisfy -- (Char -> Bool) -> Parser Char
- , char -- Char -> Parser Char
- , string -- String -> Parser String
- , (<|>) -- Parser a -> Parser a -> Parser a
- , oneOf -- [Char] -> Parser Char
- , digit -- Parser Char
- , notFollowedBy -- Parser a -> Parser ()
- , many -- Parser a -> Parser [a]
- , many1 -- Parser a -> Parser [a]
- , manyTill -- Parser a -> Parser end -> Parser [a]
- , many1Till -- Parser a -> Parser end -> Parser [a]
- , sp -- Parser Char
- , crlf -- Parser String
+ ( Parser
+ , ParserResult(..)
+
+ , failP
+
+ , parse
+ , parseStr
+
+ , anyChar
+ , eof
+ , allowEOF
+ , satisfy
+ , char
+ , string
+ , (<|>)
+ , choice
+ , oneOf
+ , digit
+ , hexDigit
+ , notFollowedBy
+ , many
+ , many1
+ , count
+ , option
+ , sepBy
+ , sepBy1
+
+ , sp
+ , ht
+ , crlf
)
where
-import Control.Monad
-import Control.Monad.State
-import qualified Data.ByteString.Lazy.Char8 as B
-import Data.ByteString.Lazy.Char8 (ByteString)
+import Control.Monad.State.Strict
+import qualified Data.ByteString.Lazy as Lazy (ByteString)
+import qualified Data.ByteString.Lazy.Char8 as B hiding (ByteString)
+import qualified Data.Foldable as Fold
+import qualified Data.Sequence as Seq
+import Data.Sequence (Seq, (|>))
-data Parser a = Parser {
- runParser :: State ByteString (Maybe a)
+-- |@'Parser' a@ is obviously a parser which parses and returns @a@.
+newtype Parser a = Parser {
+ runParser :: State ParserState (ParserResult a)
}
+data ParserState
+ = PST {
+ pstInput :: Lazy.ByteString
+ , pstIsEOFFatal :: !Bool
+ }
+ deriving (Eq, Show)
+
+
+data ParserResult a = Success !a
+ | IllegalInput -- 受理出來ない入力があった
+ | ReachedEOF -- 限界を越えて讀まうとした
+ deriving (Eq, Show)
+
+
-- (>>=) :: Parser a -> (a -> Parser b) -> Parser b
instance Monad Parser where
- p >>= f = Parser $ do saved <- get -- 失敗した時の爲に状態を保存
- result <- runParser p
- case result of
- Just a -> runParser (f a)
- Nothing -> do put saved -- 状態を復歸
- return Nothing
- return = Parser . return . Just
- fail _ = Parser $ return Nothing
+ p >>= f = Parser $! do saved <- get -- 失敗した時の爲に状態を保存
+ result <- runParser p
+ case result of
+ Success a -> runParser (f a)
+ IllegalInput -> do put saved -- 状態を復歸
+ return IllegalInput
+ ReachedEOF -> do put saved -- 状態を復歸
+ return ReachedEOF
+ return x = x `seq` Parser $! return $! Success x
+ fail _ = Parser $! return $! IllegalInput
+
+-- |@'failP'@ is just a synonym for @'Prelude.fail'
+-- 'Prelude.undefined'@.
+failP :: Parser a
+failP = fail undefined
+
+-- |@'parse' p bstr@ parses @bstr@ with @p@ and returns @(# result,
+-- remaining #)@.
+parse :: Parser a -> Lazy.ByteString -> (# ParserResult a, Lazy.ByteString #)
+parse p input -- input は lazy である必要有り。
+ = p `seq`
+ let (result, state') = runState (runParser p) (PST input True)
+ in
+ result `seq` (# result, pstInput state' #) -- pstInput state' も lazy である必要有り。
+
+-- |@'parseStr' p str@ packs @str@ and parses it.
+parseStr :: Parser a -> String -> (# ParserResult a, Lazy.ByteString #)
+parseStr p input
+ = p `seq` -- input は lazy である必要有り。
+ parse p (B.pack input)
+
+anyChar :: Parser Char
+anyChar = Parser $!
+ do state@(PST input _) <- get
+ if B.null input then
+ return ReachedEOF
+ else
+ do put $! state { pstInput = B.tail input }
+ return (Success $! B.head input)
-parse :: Parser a -> ByteString -> Maybe (a, ByteString)
-parse p input = case runState (runParser p) input of
- (Just a , input') -> Just (a, input')
- (Nothing, _ ) -> Nothing
+eof :: Parser ()
+eof = Parser $!
+ do PST input _ <- get
+ if B.null input then
+ return $! Success ()
+ else
+ return IllegalInput
-anyChar :: Parser Char
-anyChar = Parser $ do input <- get
- if B.null input then
- return Nothing
- else
- do let c = B.head input
- put (B.tail input)
- return (Just c)
+-- |@'allowEOF' p@ makes @p@ treat reaching EOF a normal failure.
+allowEOF :: Parser a -> Parser a
+allowEOF f = f `seq`
+ Parser $! do saved@(PST _ isEOFFatal) <- get
+ put $! saved { pstIsEOFFatal = False }
+
+ result <- runParser f
+
+ state <- get
+ put $! state { pstIsEOFFatal = isEOFFatal }
+
+ return result
satisfy :: (Char -> Bool) -> Parser Char
-satisfy f = do c <- anyChar
- unless (f c) (fail "")
- return c
+satisfy !f
+ = do c <- anyChar
+ if f c then
+ return c
+ else
+ failP
char :: Char -> Parser Char
-char c = satisfy (== c)
+char !c = satisfy (== c)
string :: String -> Parser String
-string str = do mapM_ char str
- return str
+string !str = str `seq`
+ do mapM_ char str
+ return str
infixr 0 <|>
+-- |This is the backtracking alternation. There is no non-backtracking
+-- equivalent.
(<|>) :: Parser a -> Parser a -> Parser a
-f <|> g = Parser $ do saved <- get -- 状態を保存
- result <- runParser f
- case result of
- Just a -> return (Just a)
- Nothing -> do put saved -- 状態を復歸
- runParser g
+f <|> g
+ = f `seq` g `seq`
+ Parser $! do saved <- get -- 状態を保存
+ result <- runParser f
+ case result of
+ Success a -> return $! Success a
+ IllegalInput -> do put saved -- 状態を復歸
+ runParser g
+ ReachedEOF -> if pstIsEOFFatal saved then
+ do put saved
+ return ReachedEOF
+ else
+ do put saved
+ runParser g
+
+
+choice :: [Parser a] -> Parser a
+choice = foldl (<|>) failP
oneOf :: [Char] -> Parser Char
-oneOf = foldl (<|>) (fail "") . map char
+oneOf = foldl (<|>) failP . map char
notFollowedBy :: Parser a -> Parser ()
-notFollowedBy p = p >>= fail "" <|> return ()
+notFollowedBy !p
+ = Parser $! do saved <- get -- 状態を保存
+ result <- runParser p
+ case result of
+ Success _ -> do put saved -- 状態を復歸
+ return IllegalInput
+ IllegalInput -> do put saved -- 状態を復歸
+ return $! Success ()
+ ReachedEOF -> do put saved -- 状態を復歸
+ return $! Success ()
digit :: Parser Char
-digit = oneOf "0123456789"
+digit = do c <- anyChar
+ if c >= '0' && c <= '9' then
+ return c
+ else
+ failP
+
+
+hexDigit :: Parser Char
+hexDigit = do c <- anyChar
+ if (c >= '0' && c <= '9') ||
+ (c >= 'a' && c <= 'f') ||
+ (c >= 'A' && c <= 'F') then
+ return c
+ else
+ failP
+
+
+many :: forall a. Parser a -> Parser [a]
+many !p = Parser $!
+ do state <- get
+ let (# result, state' #) = many' state Seq.empty
+ put state'
+ return result
+ where
+ many' :: ParserState -> Seq a -> (# ParserResult [a], ParserState #)
+ many' !st !soFar
+ = case runState (runParser p) st of
+ (Success a, st') -> many' st' (soFar |> a)
+ (IllegalInput, _) -> (# Success (Fold.toList soFar), st #)
+ (ReachedEOF , _) -> if pstIsEOFFatal st then
+ (# ReachedEOF, st #)
+ else
+ (# Success (Fold.toList soFar), st #)
-many :: Parser a -> Parser [a]
-many p = do x <- p
- xs <- many p
- return (x:xs)
- <|>
- return []
+many1 :: Parser a -> Parser [a]
+many1 !p = do x <- p
+ xs <- many p
+ return (x:xs)
-many1 :: Parser a -> Parser [a]
-many1 p = do ret <- many p
- case ret of
- [] -> fail ""
- xs -> return xs
+count :: Int -> Parser a -> Parser [a]
+count !n !p = Parser $! count' n p Seq.empty
+
+-- This implementation is rather ugly but we need to make it
+-- tail-recursive to avoid stack overflow.
+count' :: Int -> Parser a -> Seq a -> State ParserState (ParserResult [a])
+count' 0 _ !soFar = return $! Success $! Fold.toList soFar
+count' !n !p !soFar = do saved <- get
+ result <- runParser p
+ case result of
+ Success a -> count' (n-1) p (soFar |> a)
+ IllegalInput -> do put saved
+ return IllegalInput
+ ReachedEOF -> do put saved
+ return ReachedEOF
-manyTill :: Parser a -> Parser end -> Parser [a]
-manyTill p end = many $ do x <- p
- end
- return x
+-- def may be a _|_
+option :: a -> Parser a -> Parser a
+option def p = p `seq`
+ p <|> return def
-many1Till :: Parser a -> Parser end -> Parser [a]
-many1Till p end = many1 $ do x <- p
- end
- return x
+sepBy :: Parser a -> Parser sep -> Parser [a]
+sepBy p sep = p `seq` sep `seq`
+ sepBy1 p sep <|> return []
+
+
+sepBy1 :: Parser a -> Parser sep -> Parser [a]
+sepBy1 p sep = p `seq` sep `seq`
+ do x <- p
+ xs <- many $! sep >> p
+ return (x:xs)
sp :: Parser Char
sp = char ' '
+ht :: Parser Char
+ht = char '\t'
+
+
crlf :: Parser String
crlf = string "\x0d\x0a"
projects / Lucu.git / blobdiff