--- |Utility functions used internally in the Lucu httpd. These
--- functions may be useful too for something else.
+{-# LANGUAGE
+ FlexibleContexts
+ , OverloadedStrings
+ , UnicodeSyntax
+ #-}
+-- |Utility functions used internally in this package.
module Network.HTTP.Lucu.Utils
- ( splitBy
- , joinWith
- , trim
- , noCaseEq
- , isWhiteSpace
+ ( Host
+ , PathSegment
+ , Path
+
+ , splitBy
, quoteStr
, parseWWWFormURLEncoded
+ , uriHost
+ , uriPathSegments
+ , trim
+
+ , (⊲)
+ , (⊳)
+ , (⋈)
+ , mapM
+
+ , getLastModified
)
where
-
-import Control.Monad.Trans
+import Control.Applicative hiding (empty)
+import Control.Monad hiding (mapM)
+import Data.Ascii (Ascii, AsciiBuilder)
+import qualified Data.Ascii as A
+import Data.ByteString (ByteString)
+import qualified Data.ByteString.Char8 as BS
+import Data.CaseInsensitive (CI)
+import qualified Data.CaseInsensitive as CI
import Data.Char
-import Data.List
-import Foreign
-import Foreign.C
+import Data.Collections
+import Data.Collections.BaseInstances ()
+import Data.Convertible.Base
+import Data.Convertible.Instances.Ascii ()
+import Data.Convertible.Instances.Text ()
+import Data.Convertible.Instances.Time ()
+import Data.Maybe
+import Data.Monoid.Unicode
+import Data.Text (Text)
+import Data.Time
import Network.URI
+import Prelude hiding (last, mapM, null, reverse)
+import Prelude.Unicode
+import System.Directory
--- |> splitBy (== ':') "ab:c:def"
--- > ==> ["ab", "c", "def"]
-splitBy :: (a -> Bool) -> [a] -> [[a]]
-splitBy isSeparator src
- = case break isSeparator src
- of (last , [] ) -> last : []
- (first, sep:rest) -> first : splitBy isSeparator rest
-
--- |> joinWith ':' ["ab", "c", "def"]
--- > ==> "ab:c:def"
-joinWith :: [a] -> [[a]] -> [a]
-joinWith separator xs
- = foldr (++) [] $ intersperse separator xs
-
--- |> trim (== '_') "__ab_c__def___"
--- > ==> "ab_c__def"
-trim :: (a -> Bool) -> [a] -> [a]
-trim p = trimTail . trimHead
- where
- trimHead = dropWhile p
- trimTail = reverse . trimHead . reverse
+-- |'Host' represents an IP address or a host name in an URI
+-- authority.
+type Host = CI Text
+
+-- |'PathSegment' represents an URI path segment, split by slashes and
+-- percent-decoded.
+type PathSegment = ByteString
--- |@'noCaseEq' a b@ is equivalent to @(map toLower a) == (map toLower
--- b)@
-noCaseEq :: String -> String -> Bool
-noCaseEq a b
- = (map toLower a) == (map toLower b)
+-- |'Path' is a list of URI path segments.
+type Path = [PathSegment]
--- |@'isWhiteSpace' c@ is True iff c is one of SP, HT, CR and LF.
-isWhiteSpace :: Char -> Bool
-isWhiteSpace = flip elem " \t\r\n"
+-- |>>> splitBy (== ':') "ab:c:def"
+-- ["ab", "c", "def"]
+splitBy ∷ (a → Bool) → [a] → [[a]]
+{-# INLINEABLE splitBy #-}
+splitBy isSep src
+ = case break isSep src of
+ (last , [] ) → [last]
+ (first, _sep:rest) → first : splitBy isSep rest
--- |> quoteStr "abc"
--- > ==> "\"abc\""
+-- |>>> quoteStr "abc"
+-- "\"abc\""
--
--- > quoteStr "ab\"c"
--- > ==> "\"ab\\\"c\""
-quoteStr :: String -> String
-quoteStr str = foldr (++) "" (["\""] ++ map quote str ++ ["\""])
+-- >>> quoteStr "ab\"c"
+-- "\"ab\\\"c\""
+quoteStr ∷ Ascii → AsciiBuilder
+quoteStr str = cs ("\"" ∷ Ascii) ⊕
+ go (cs str) (∅) ⊕
+ cs ("\"" ∷ Ascii)
where
- quote :: Char -> String
- quote '"' = "\\\""
- quote c = [c]
+ go ∷ 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 ⊕ cs ("\\\"" ∷ Ascii))
+ b2ab ∷ ByteString → AsciiBuilder
+ b2ab = cs ∘ A.unsafeFromByteString
--- |> parseWWWFormURLEncoded "aaa=bbb&ccc=ddd"
--- > ==> [("aaa", "bbb"), ("ccc", "ddd")]
-parseWWWFormURLEncoded :: String -> [(String, String)]
+-- |>>> parseWWWFormURLEncoded "aaa=bbb&ccc=ddd"
+-- [("aaa", "bbb"), ("ccc", "ddd")]
+parseWWWFormURLEncoded ∷ Ascii → [(ByteString, ByteString)]
parseWWWFormURLEncoded src
- | src == "" = []
- | otherwise = do pairStr <- splitBy (\ c -> c == ';' || c == '&') src
- let pair = break (== '=') pairStr
- return ( unEscapeString $ fst pair
- , unEscapeString $ snd pair
+ -- THINKME: We could gain some performance by using attoparsec
+ -- here.
+ | src ≡ "" = []
+ | otherwise = do pairStr ← splitBy (\ c → c ≡ ';' ∨ c ≡ '&') (cs 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 ∘ (plusToSpace <$>)
+
+ plusToSpace ∷ Char → Char
+ plusToSpace '+' = ' '
+ plusToSpace c = c
+
+-- |>>> uriHost "http://example.com/foo/bar"
+-- "example.com"
+uriHost ∷ URI → Host
+{-# INLINE uriHost #-}
+uriHost = CI.mk ∘ cs ∘ uriRegName ∘ fromJust ∘ uriAuthority
+
+-- |>>> uriPathSegments "http://example.com/foo/bar"
+-- ["foo", "bar"]
+uriPathSegments ∷ URI → Path
+uriPathSegments uri
+ = let reqPathStr = uriPath uri
+ reqPath = [ unEscapeString x
+ | x ← splitBy (≡ '/') reqPathStr, (¬) (null x) ]
+ in
+ BS.pack <$> reqPath
+
+-- |>>> trim " ab c d "
+-- "ab c d"
+trim ∷ String → String
+trim = reverse ∘ f ∘ reverse ∘ f
+ where
+ f = dropWhile isSpace
+
+infixr 5 ⊲
+-- | (⊲) = ('<|')
+--
+-- U+22B2, NORMAL SUBGROUP OF
+(⊲) ∷ Sequence α a ⇒ a → α → α
+(⊲) = (<|)
+
+infixl 5 ⊳
+-- | (⊳) = ('|>')
+--
+-- U+22B3, CONTAINS AS NORMAL SUBGROUP
+(⊳) ∷ Sequence α a ⇒ α → a → α
+(⊳) = (|>)
+
+infixr 5 ⋈
+-- | (⋈) = ('><')
+--
+-- U+22C8, BOWTIE
+(⋈) ∷ Sequence α a ⇒ α → α → α
+(⋈) = (><)
+
+-- |Generalised @mapM@ from any 'Foldable' to 'Unfoldable'. Why isn't
+-- this in the @collections-api@?
+mapM ∷ (Foldable α a, Unfoldable β b, Functor m, Monad m)
+ ⇒ (a → m b) → α → m β
+{-# INLINE mapM #-}
+mapM = flip foldrM empty ∘ (flip ((<$>) ∘ flip insert) ∘)
+
+-- |Get the modification time of a given file.
+getLastModified ∷ FilePath → IO UTCTime
+getLastModified = (cs <$>) ∘ getModificationTime