From 646bf9c4d838f2cbb59e3eb0ae7bf0636ff2f7e4 Mon Sep 17 00:00:00 2001
From: bjorn <bjorn@bringert.net>
Date: Sun, 24 Aug 2008 17:39:24 +0000
Subject: [PATCH] URL decode text input in fastcgi server, including %uXXXX
 sequences.

---
 src/server/MainFastCGI.hs |  3 ++-
 src/server/URLEncoding.hs | 18 ++++++++++++++++++
 2 files changed, 20 insertions(+), 1 deletion(-)
 create mode 100644 src/server/URLEncoding.hs

diff --git a/src/server/MainFastCGI.hs b/src/server/MainFastCGI.hs
index 230e09b00..b7af9fce4 100644
--- a/src/server/MainFastCGI.hs
+++ b/src/server/MainFastCGI.hs
@@ -3,6 +3,7 @@
 import PGF (PGF)
 import qualified PGF
 import FastCGIUtils
+import URLEncoding
 
 import Network.CGI
 import Text.JSON
@@ -42,7 +43,7 @@ cgiMain pgf =
        outputJSON json
   where
     getText :: CGI String
-    getText = liftM (fromMaybe "") $ getInput "input"
+    getText = liftM (maybe "" urlDecodeUnicode) $ getInput "input"
 
     getTree :: CGI PGF.Tree
     getTree = do mt <- getInput "tree"
diff --git a/src/server/URLEncoding.hs b/src/server/URLEncoding.hs
new file mode 100644
index 000000000..ad5fb0dd9
--- /dev/null
+++ b/src/server/URLEncoding.hs
@@ -0,0 +1,18 @@
+module URLEncoding where
+
+import Data.Bits (shiftL, (.|.))
+import Data.Char (chr,digitToInt,isHexDigit)
+
+
+urlDecodeUnicode :: String -> String
+urlDecodeUnicode [] = ""
+urlDecodeUnicode ('%':'u':x1:x2:x3:x4:s) 
+    | all isHexDigit [x1,x2,x3,x4] =
+    chr (    digitToInt x1 `shiftL` 12 
+         .|. digitToInt x2 `shiftL` 8
+         .|. digitToInt x3 `shiftL` 4
+         .|. digitToInt x4) : urlDecodeUnicode s
+urlDecodeUnicode ('%':x1:x2:s) | isHexDigit x1 && isHexDigit x2 =
+    chr (    digitToInt x1 `shiftL` 4
+         .|. digitToInt x2) : urlDecodeUnicode s
+urlDecodeUnicode (c:s) = c : urlDecodeUnicode s