Merge branch 'master' into attoparsec

[Lucu.git] / Network / HTTP / Lucu / Utils.hs

diff --git a/Network/HTTP/Lucu/Utils.hs b/Network/HTTP/Lucu/Utils.hs
index d92516ee15875a6791fc912b8d11ebe1eb765ffb..d2541691ced99dd41ac579d146224fa7657a8f7a 100644 (file)
--- a/Network/HTTP/Lucu/Utils.hs
+++ b/Network/HTTP/Lucu/Utils.hs
@@ -1,92 +1,84 @@
+{-# LANGUAGE
+    BangPatterns
+  , OverloadedStrings
+  , UnicodeSyntax
+  #-}
 -- |Utility functions used internally in the Lucu httpd. These
 -- functions may be useful too for something else.
 module Network.HTTP.Lucu.Utils
     ( splitBy
     , joinWith
-    , trim
-    , noCaseEq
-    , noCaseEq'
-    , isWhiteSpace
     , quoteStr
     , parseWWWFormURLEncoded
     )
     where
-
-import Data.Char
-import Data.List
+import Control.Monad
+import Data.Ascii (Ascii, AsciiBuilder)
+import qualified Data.Ascii as A
+import qualified Data.ByteString.Char8 as BS
+import Data.List hiding (last)
+import Data.Monoid.Unicode
 import Network.URI
+import Prelude hiding (last)
+import Prelude.Unicode
 
 -- |> splitBy (== ':') "ab:c:def"
 --  > ==> ["ab", "c", "def"]
-splitBy :: (a -> Bool) -> [a] -> [[a]]
-splitBy isSeparator src
-    = isSeparator `seq`
-      case break isSeparator src
-      of (last , []      ) -> last  : []
-         (first, sep:rest) -> first : splitBy isSeparator rest
+splitBy ∷ (a → Bool) → [a] → [[a]]
+splitBy isSep src
+    = case break isSep src
+      of (last , []       ) → [last]
+         (first, _sep:rest) → first : splitBy isSep rest
 
 -- |> joinWith ":" ["ab", "c", "def"]
 --  > ==> "ab:c:def"
-joinWith :: [a] -> [[a]] -> [a]
-joinWith separator xs
-    = separator `seq` xs `seq`
-      foldr (++) [] $! intersperse separator xs
-
--- |> trim (== '_') "__ab_c__def___"
---  > ==> "ab_c__def"
-trim :: (a -> Bool) -> [a] -> [a]
-trim p = p `seq` trimTail . trimHead
+joinWith ∷ Ascii → [AsciiBuilder] → AsciiBuilder
+{-# INLINEABLE joinWith #-}
+joinWith sep = flip go (∅)
     where
-      trimHead = dropWhile p
-      trimTail = reverse . trimHead . reverse
-
--- |@'noCaseEq' a b@ is equivalent to @(map toLower a) == (map toLower
--- b)@. See 'noCaseEq''.
-noCaseEq :: String -> String -> Bool
-noCaseEq a b
-    = (map toLower a) == (map toLower b)
-{-# INLINE noCaseEq #-}
-
--- |@'noCaseEq'' a b@ is a variant of 'noCaseEq' which first checks
--- the length of two strings to avoid possibly unnecessary comparison.
-noCaseEq' :: String -> String -> Bool
-noCaseEq' a b
-    | length a /= length b = False
-    | otherwise            = noCaseEq a b
-{-# INLINE noCaseEq' #-}
-
--- |@'isWhiteSpace' c@ is True iff c is one of SP, HT, CR and LF.
-isWhiteSpace :: Char -> Bool
-isWhiteSpace ' '  = True
-isWhiteSpace '\t' = True
-isWhiteSpace '\r' = True
-isWhiteSpace '\n' = True
-isWhiteSpace _    = False
-{-# INLINE isWhiteSpace #-}
+      go ∷ [AsciiBuilder] → AsciiBuilder → AsciiBuilder
+      {-# INLINE go #-}
+      go []     ab = ab
+      go (x:[]) ab = ab ⊕ x
+      go (x:xs) ab = go xs (ab ⊕ A.toAsciiBuilder sep ⊕ x)
 
 -- |> quoteStr "abc"
 --  > ==> "\"abc\""
 --
 --  > quoteStr "ab\"c"
 --  > ==> "\"ab\\\"c\""
-quoteStr :: String -> String
-quoteStr str = str `seq`
-               foldr (++) "" (["\""] ++ map quote str ++ ["\""])
+quoteStr ∷ Ascii → AsciiBuilder
+quoteStr str = A.toAsciiBuilder "\"" ⊕
+               go (A.toByteString str) (∅) ⊕
+               A.toAsciiBuilder "\""
     where
-      quote :: Char -> String
-      quote '"' = "\\\""
-      quote c   = [c]
+      go ∷ BS.ByteString → AsciiBuilder → AsciiBuilder
+      go bs ab
+          = case BS.break (≡ '"') bs of
+              (x, y)
+                  | BS.null y → ab ⊕ b2ab x
+                  | otherwise → go (BS.tail y) (ab ⊕ b2ab x
+                                                   ⊕ A.toAsciiBuilder "\\\"")
 
+      b2ab ∷ BS.ByteString → AsciiBuilder
+      b2ab = A.toAsciiBuilder ∘ A.unsafeFromByteString
 
 -- |> parseWWWFormURLEncoded "aaa=bbb&ccc=ddd"
 --  > ==> [("aaa", "bbb"), ("ccc", "ddd")]
-parseWWWFormURLEncoded :: String -> [(String, String)]
+parseWWWFormURLEncoded ∷ String → [(String, String)]
 parseWWWFormURLEncoded src
-    | src == "" = []
-    | otherwise = do pairStr <- splitBy (\ c -> c == ';' || c == '&') src
-                     let (key, value) = break (== '=') pairStr
-                     return ( unEscapeString key
-                            , unEscapeString $ case value of
-                                                 ('=':val) -> val
-                                                 ""        -> ""
+    | null src  = []
+    | otherwise = do pairStr ← splitBy (\ c → c ≡ ';' ∨ c ≡ '&') src
+                     let (key, value) = break (≡ '=') pairStr
+                     return ( unescape key
+                            , unescape $ case value of
+                                           ('=':val) → val
+                                           val       → val
                             )
+    where
+      unescape ∷ String → String
+      unescape = unEscapeString ∘ map plusToSpace
+
+      plusToSpace ∷ Char → Char
+      plusToSpace '+' = ' '
+      plusToSpace c   = c