From f94e34058c45ea9ba38511e0bad09e8125419600 Mon Sep 17 00:00:00 2001
From: reger <reger18@arcor.de>
Date: Wed, 11 Mar 2015 01:05:14 +0100
Subject: [PATCH] fix url (path) %-decoding
 http://mantis.tokeek.de/view.php?id=519 - add test case for this

---
 .../cora/document/id/MultiProtocolURL.java    | 13 ++++++++--
 .../document/id/MultiProtocolURLTest.java     | 24 +++++++++++++++++++
 2 files changed, 35 insertions(+), 2 deletions(-)

diff --git a/source/net/yacy/cora/document/id/MultiProtocolURL.java b/source/net/yacy/cora/document/id/MultiProtocolURL.java
index bb0e69cb1..fdc7847ef 100644
--- a/source/net/yacy/cora/document/id/MultiProtocolURL.java
+++ b/source/net/yacy/cora/document/id/MultiProtocolURL.java
@@ -124,6 +124,15 @@ public class MultiProtocolURL implements Serializable, Comparable<MultiProtocolU
         this.port = url.port;
     }
 
+    /**
+     * Create MultiProtocolURL
+     *
+     * decoding exception: if url string contains http url with char '%' the url string must be url encoded (percent-escaped) before
+     * as internal encoding is skipped if url string contains '%'.
+     *
+     * @param url '%' char url encoded before
+     * @throws MalformedURLException
+     */
     public MultiProtocolURL(String url) throws MalformedURLException {
         if (url == null) throw new MalformedURLException("url string is null");
 
@@ -133,8 +142,7 @@ public class MultiProtocolURL implements Serializable, Comparable<MultiProtocolU
         // identify protocol
         assert (url != null);
         url = url.trim();
-        url = UTF8.decodeURL(url); // normalization here
-        //url = patternSpace.matcher(url).replaceAll(" ");
+        
         if (url.startsWith("//")) {
             // patch for urls starting with "//" which can be found in the wild
             url = "http:" + url;
@@ -195,6 +203,7 @@ public class MultiProtocolURL implements Serializable, Comparable<MultiProtocolU
             identSearchpart();
             escape();
         } else {
+            url = UTF8.decodeURL(url); // normalization here
             // this is not a http or ftp url
             if (this.protocol.equals("mailto")) {
                 // parse email url
diff --git a/test/net/yacy/cora/document/id/MultiProtocolURLTest.java b/test/net/yacy/cora/document/id/MultiProtocolURLTest.java
index 394fc1a25..e3c7f5a2d 100644
--- a/test/net/yacy/cora/document/id/MultiProtocolURLTest.java
+++ b/test/net/yacy/cora/document/id/MultiProtocolURLTest.java
@@ -141,6 +141,30 @@ public class MultiProtocolURLTest {
 
         }
     }
+
+    /**
+     * Test of toNormalform method, of class MultiProtocolURL.
+     */
+    @Test
+    public void testToNormalform() throws Exception {
+        // some test url/uri with problems in the past
+        String[][] testStrings = new String[][]{
+            // teststring , expectedresult
+            new String[]{"http://www.heise.de/newsticker/thema/%23saukontrovers", "http://www.heise.de/newsticker/thema/%23saukontrovers"}, // http://mantis.tokeek.de/view.php?id=519
+            new String[]{"http://www.heise.de/newsticker/thema/#saukontrovers", "http://www.heise.de/newsticker/thema/"}
+        };
+
+        for (String[] testString : testStrings) {
+            // desired conversion result
+            System.out.print("orig uri: " + testString[0]);
+            String shouldBe = testString[1];
+            // conversion result
+            String resolvedHost = new MultiProtocolURL(testString[0]).toNormalform(true);
+            // test if equal
+            assertEquals(shouldBe, resolvedHost);
+            System.out.println(" -> " + resolvedHost);
+        }
+    }
 }