From 98bb0b98fcf280ae19db6a457ddc6dd1be904af7 Mon Sep 17 00:00:00 2001
From: James Reeves <jreeves@weavejester.com>
Date: Fri, 23 Mar 2012 22:42:25 +0000
Subject: [PATCH] Technically correct version of URL/percent-encoding

Java's URLEncoder and URLDecoder classes don't follow the URL-encoding
format precisely, instead encoding strings in the slightly different
www-form-urlencoded format. This commit ensures that the url-encode and
url-decode functions work correctly, leaving it up to the form-encode and
form-decode functions to handle data encoded in the www-form-urlencoded
format.
---
 ring-core/src/ring/util/codec.clj       | 43 +++++++++++++++++++++++++++------
 ring-core/test/ring/util/test/codec.clj | 24 ++++++++++++------
 2 files changed, 53 insertions(+), 14 deletions(-)

diff --git a/ring-core/src/ring/util/codec.clj b/ring-core/src/ring/util/codec.clj
index 9dbff02..bee2006 100644
--- a/ring-core/src/ring/util/codec.clj
+++ b/ring-core/src/ring/util/codec.clj
@@ -1,24 +1,53 @@
 (ns ring.util.codec
   "Encoding and decoding utilities."
   (:use ring.util.data)
-  (:require [clojure.string :as string])
+  (:require [clojure.string :as str])
   (:import java.io.File
            (java.net URLEncoder URLDecoder)
            org.apache.commons.codec.binary.Base64))
 
+(defn- double-escape [^String x]
+  (.replace x "\\" "\\\\"))
+
+(defn percent-encode
+  "Percent-encode every character in the given string using either the specified
+  encoding, or UTF-8 by default."
+  [unencoded & [encoding]]
+  (->> (.getBytes unencoded (or encoding "UTF-8"))
+       (map (partial format "%%%02X"))
+       (str/join)))
+
+(defn- parse-bytes [encoded-bytes]
+  (->> (re-seq #"%.." encoded-bytes)
+       (map #(subs % 1))
+       (map #(.byteValue (Integer/parseInt % 16)))
+       (byte-array)))
+
+(defn percent-decode
+  "Decode every percent-encoded character in the given string using the
+  specified encoding, or UTF-8 by default."
+  [encoded & [encoding]]
+  (str/replace encoded
+               #"(?:%..)+"
+               (fn [chars]
+                 (-> (parse-bytes chars)
+                     (String. (or encoding "UTF-8"))
+                     (double-escape)))))
+
 (defn url-encode
   "Returns the url-encoded version of the given string, using either a specified
   encoding or UTF-8 by default."
   [unencoded & [encoding]]
-  (URLEncoder/encode unencoded (or encoding "UTF-8")))
+  (str/replace
+    unencoded
+    #"[^A-Za-z0-9_~.+-]+"
+    #(double-escape (percent-encode % encoding))))
 
 (defn url-decode
   "Returns the url-decoded version of the given string, using either a specified
   encoding or UTF-8 by default. If the encoding is invalid, nil is returned."
   [encoded & [encoding]]
-  (try
-    (URLDecoder/decode encoded (or encoding "UTF-8"))
-    (catch Exception e nil)))
+  (percent-decode encoded encoding))
 
 (defn base64-encode
   "Encode an array of bytes into a base64 encoded string."
@@ -43,7 +72,7 @@
                   (url-decode (or val "") encoding))
           param-map))
       {}
-      (string/split param-string #"&"))))
+      (str/split param-string #"&"))))
 
 (defn form-encode
   "Encode parameters from a map into a string."
@@ -54,7 +83,7 @@
                   (vals param-map)
                   encoding))
   ([params values encoding]
-     (string/join #"&"
+     (str/join #"&"
                   (map (fn [param value]
                          (if (vector? value)
                            (form-encode (repeat (count value) param)
diff --git a/ring-core/test/ring/util/test/codec.clj b/ring-core/test/ring/util/test/codec.clj
index f99b051..58cdf1d 100644
--- a/ring-core/test/ring/util/test/codec.clj
+++ b/ring-core/test/ring/util/test/codec.clj
@@ -3,16 +3,26 @@
         ring.util.codec)
   (:import java.util.Arrays))
 
+(deftest test-percent-encode
+  (is (= (percent-encode " ") "%20"))
+  (is (= (percent-encode "+") "%2B"))
+  (is (= (percent-encode "foo") "%66%6F%6F")))
+
+(deftest test-percent-decode
+  (is (= (percent-decode "%20") " "))
+  (is (= (percent-decode "foo%20bar") "foo bar"))
+  (is (= (percent-decode "foo%FE%FF%00%2Fbar" "UTF-16") "foo/bar")))
+
 (deftest test-url-encode
-  (is (= "foo%2Fbar" (url-encode "foo/bar")))
-  (is (= "foo%FE%FF%00%2Fbar") (url-encode "foo/bar" "UTF-16")))
+  (is (= (url-encode "foo/bar") "foo%2Fbar"))
+  (is (= (url-encode "foo/bar" "UTF-16") "foo%FE%FF%00%2Fbar"))
+  (is (= (url-encode "foo+bar") "foo+bar"))
+  (is (= (url-encode "foo bar") "foo%20bar")))
 
 (deftest test-url-decode
-  (testing "standard behavior"
-    (is (= "foo/bar" (url-decode "foo%2Fbar")))
-    (is (= "foo/bar" (url-decode "foo%FE%FF%00%2Fbar" "UTF-16"))))
-  (testing "returns nil when underlying Java methods throw an exception"
-    (is (nil? (url-decode "%")))))
+  (is (= (url-decode "foo%2Fbar") "foo/bar" ))
+  (is (= (url-decode "foo%FE%FF%00%2Fbar" "UTF-16") "foo/bar"))
+  (is (= (url-decode "%") "%")))
 
 (deftest test-base64-encoding
   (let [str-bytes (.getBytes "foo?/+" "UTF-8")]