+{-# LANGUAGE
+ 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
+ , splitPathInfo
+ , show3
)
where
-
-import Data.Char
-import Data.List
+import Blaze.ByteString.Builder.ByteString as B
+import Blaze.Text.Int as BT
+import Control.Monad
+import Data.Ascii (Ascii, AsciiBuilder)
+import qualified Data.Ascii as A
+import Data.ByteString (ByteString)
+import qualified Data.ByteString.Char8 as BS
+import Data.List hiding (last)
+import Data.Monoid.Unicode
+import Data.Text (Text)
+import Data.Text.Encoding as T
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]]
+{-# INLINEABLE splitBy #-}
+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 ∷ Ascii → [(ByteString, ByteString)]
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
- "" -> ""
+ -- THINKME: We could gain some performance by using attoparsec
+ -- here.
+ | src ≡ "" = []
+ | otherwise = do pairStr ← splitBy (\ c → c ≡ ';' ∨ c ≡ '&') (A.toString src)
+ let (key, value) = break (≡ '=') pairStr
+ return ( unescape key
+ , unescape $ case value of
+ ('=':val) → val
+ val → val
)
+ where
+ unescape ∷ String → ByteString
+ unescape = BS.pack ∘ unEscapeString ∘ map plusToSpace
+
+ plusToSpace ∷ Char → Char
+ plusToSpace '+' = ' '
+ plusToSpace c = c
+
+-- |> splitPathInfo "http://example.com/foo/bar"
+-- > ==> ["foo", "bar"]
+splitPathInfo ∷ URI → [Text]
+splitPathInfo uri
+ = let reqPathStr = uriPath uri
+ reqPath = [unEscapeString x | x ← splitBy (≡ '/') reqPathStr, (¬) (null x)]
+ in
+ map (T.decodeUtf8 ∘ BS.pack) reqPath
+
+-- |> show3 5
+-- > ==> "005"
+show3 ∷ Integral n ⇒ n → AsciiBuilder
+{-# INLINEABLE show3 #-}
+show3 = A.unsafeFromBuilder ∘ go
+ where
+ go i | i ≥ 0 ∧ i < 10 = B.fromByteString "00" ⊕ BT.digit i
+ | i ≥ 0 ∧ i < 100 = B.fromByteString "0" ⊕ BT.integral i
+ | i ≥ 0 ∧ i < 1000 = BT.integral i
+ | otherwise = error ("show3: the integer i must satisfy 0 <= i < 1000: " ⧺ show i)