Resource paths should not be assumed to be encoded in UTF-8. HTTP/1.1 says nothing...

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

diff --git a/Network/HTTP/Lucu/Utils.hs b/Network/HTTP/Lucu/Utils.hs
index d2541691ced99dd41ac579d146224fa7657a8f7a..4db7c0555e1e05513052ef0573e4dfab2874fc24 100644 (file)
--- a/Network/HTTP/Lucu/Utils.hs
+++ b/Network/HTTP/Lucu/Utils.hs
@@ -1,6 +1,5 @@
 {-# LANGUAGE
-    BangPatterns
-  , OverloadedStrings
+    OverloadedStrings
   , UnicodeSyntax
   #-}
 -- |Utility functions used internally in the Lucu httpd. These
@@ -10,11 +9,16 @@ module Network.HTTP.Lucu.Utils
     , joinWith
     , quoteStr
     , parseWWWFormURLEncoded
+    , splitPathInfo
+    , show3
     )
     where
+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
@@ -25,6 +29,7 @@ import Prelude.Unicode
 -- |> splitBy (== ':') "ab:c:def"
 --  > ==> ["ab", "c", "def"]
 splitBy ∷ (a → Bool) → [a] → [[a]]
+{-# INLINEABLE splitBy #-}
 splitBy isSep src
     = case break isSep src
       of (last , []       ) → [last]
@@ -65,10 +70,12 @@ quoteStr str = A.toAsciiBuilder "\"" ⊕
 
 -- |> parseWWWFormURLEncoded "aaa=bbb&ccc=ddd"
 --  > ==> [("aaa", "bbb"), ("ccc", "ddd")]
-parseWWWFormURLEncoded ∷ String → [(String, String)]
+parseWWWFormURLEncoded ∷ Ascii → [(ByteString, ByteString)]
 parseWWWFormURLEncoded src
-    | null src  = []
-    | otherwise = do pairStr ← splitBy (\ c → c ≡ ';' ∨ c ≡ '&') src
+    -- 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
@@ -76,9 +83,29 @@ parseWWWFormURLEncoded src
                                            val       → val
                             )
     where
-      unescape ∷ String → String
-      unescape = unEscapeString ∘ map plusToSpace
+      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 → [ByteString]
+splitPathInfo uri
+    = let reqPathStr = uriPath uri
+          reqPath    = [unEscapeString x | x ← splitBy (≡ '/') reqPathStr, (¬) (null x)]
+      in
+        map 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)