X-Git-Url: http://git.cielonegro.org/gitweb.cgi?p=Lucu.git;a=blobdiff_plain;f=Network%2FHTTP%2FLucu%2FParser.hs;h=660f550a9353cc687a56ae0e6e422ecc41a7e3f1;hp=f2e4be337abdc375503ec4761385a2222a6fbedb;hb=19043d7;hpb=ba87ca171f0a058f501655ffa6f6e845f2f800a6

diff --git a/Network/HTTP/Lucu/Parser.hs b/Network/HTTP/Lucu/Parser.hs
index f2e4be3..660f550 100644
--- a/Network/HTTP/Lucu/Parser.hs
+++ b/Network/HTTP/Lucu/Parser.hs
@@ -1,260 +1,57 @@
--- |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.
+{-# LANGUAGE
+    UnicodeSyntax
+  #-}
+-- |This is an auxiliary parser utilities. You usually don't have to
+-- use this module directly.
 module Network.HTTP.Lucu.Parser
-    ( Parser
-    , ParserResult(..)
-
-    , failP
-
-    , parse
-    , parseStr
-
-    , anyChar
-    , eof
-    , allowEOF
-    , satisfy
-    , char
-    , string
-    , (<|>)
-    , oneOf
-    , digit
-    , hexDigit
-    , notFollowedBy
-    , many
-    , many1
-    , count
-    , option
-    , sepBy
-    , sepBy1
-
-    , sp
-    , ht
-    , crlf
+    ( atMost
+    , finishOff
+    , skipManyTill
+    , skipWhile1
+    , skipSpace1
+    , isAlphaNum
     )
     where
-
-import           Control.Monad.State.Strict
-import           Data.ByteString.Base (LazyByteString)
-import           Data.ByteString.Lazy ()
-import qualified Data.ByteString.Lazy.Char8 as B
-
--- |@'Parser' a@ is obviously a parser which parses and returns @a@.
-newtype Parser a = Parser {
-      runParser :: State ParserState (ParserResult a)
-    }
-
-
-data ParserState
-    = PST {
-        pstInput      :: LazyByteString
-      , 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
-                             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 -> LazyByteString -> (ParserResult a, LazyByteString)
-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, LazyByteString)
-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)
-
-
-eof :: Parser ()
-eof = Parser $!
-      do PST input _ <- get
-         if B.null input then
-             return $! Success ()
-           else
-             return IllegalInput
-
--- |@'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 = f `seq`
-            do c <- anyChar
-               if f c then
-                   return c
-                 else
-                   failP
-
-
-char :: Char -> Parser Char
-char c = c `seq` satisfy (== c)
-
-
-string :: String -> Parser String
-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
-    = 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
-                                         return ReachedEOF
-                                     else
-                                         do put saved
-                                            runParser g
-
-
-oneOf :: [Char] -> Parser Char
-oneOf = foldl (<|>) failP . map char
-
-
-notFollowedBy :: Parser a -> Parser ()
-notFollowedBy p = p `seq`
-                  (p >> failP) <|> return ()
-
-
-digit :: Parser Char
-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 :: Parser a -> Parser [a]
-many p = p `seq`
-         do x  <- p
-            xs <- many p
-            return (x:xs)
-         <|>
-         return []
-
-
-many1 :: Parser a -> Parser [a]
-many1 p = p `seq`
-          do x  <- p
-             xs <- many p
-             return (x:xs)
-
-
-count :: Int -> Parser a -> Parser [a]
-count 0 _ = return []
-count n p = n `seq` p `seq`
-            do x  <- p
-               xs <- count (n-1) p
-               return (x:xs)
-
--- def may be a _|_
-option :: a -> Parser a -> Parser a
-option def p = p `seq`
-               p <|> return def
-
-
-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 $! sep >> p
-                  return (x:xs)
-
-
-sp :: Parser Char
-sp = char ' '
-
-
-ht :: Parser Char
-ht = char '\t'
-
-
-crlf :: Parser String
-crlf = string "\x0d\x0a"
+import Control.Applicative
+import Control.Applicative.Unicode
+import Control.Monad.Unicode
+import Data.Attoparsec.Char8
+import Prelude.Unicode
+
+-- |@'atMost' n v@ is like @'P.many' v@ but accumulates @v@ at most
+-- @n@ times.
+atMost ∷ Alternative f ⇒ Int → f α → f [α]
+{-# INLINEABLE atMost #-}
+atMost 0 _ = pure []
+atMost n v = ( (:) <$> v ⊛ atMost (n-1) v )
+             <|>
+             pure []
+
+-- |@'finishOff' p@ is equivalent to @p '>>=' \\a -> 'endOfInput' '>>'
+-- 'return' a@.
+finishOff ∷ Parser α → Parser α
+{-# INLINE finishOff #-}
+finishOff = ((endOfInput *>) ∘ return =≪)
+
+-- |Similar to 'manyTill' but discards the result.
+skipManyTill ∷ Alternative f ⇒ f α → f β → f ()
+{-# INLINEABLE skipManyTill #-}
+skipManyTill p end = go
+    where
+      go = (end *> pure ()) <|> (p *> go)
+
+-- |Similar to 'skipWhile' but consumes at least one character.
+skipWhile1 ∷ (Char → Bool) → Parser ()
+{-# INLINE skipWhile1 #-}
+skipWhile1 p = takeWhile1 p *> pure ()
+
+-- |Similar to 'skipSpace' but consumes at least one whitespace.
+skipSpace1 ∷ Parser ()
+{-# INLINE skipSpace1 #-}
+skipSpace1 = skipMany1 space
+
+-- |@'isAlphaNum' c@ returns 'True' iff @'isDigit' c ||
+-- 'isAlpha_ascii' c@.
+isAlphaNum ∷ Char → Bool
+{-# INLINE isAlphaNum #-}
+isAlphaNum c = isDigit c ∨ isAlpha_ascii c