Replaced whitespaces by "%20" in urls on redirects and .connect(String

url)
jhy · Nov 11, 2013 · c765b81 · c765b81
1 parent c1bdb9d
commit c765b81
Show file tree

Hide file tree

Showing 2 changed files with 22 additions and 2 deletions.
diff --git a/src/main/java/org/jsoup/helper/HttpConnection.java b/src/main/java/org/jsoup/helper/HttpConnection.java
@@ -34,6 +34,12 @@ public static Connection connect(URL url) {
         return con;
     }
 
+	private static String encodeUrl(String url) {
+		if(url == null)
+			return null;
+    	return url.replaceAll(" ", "%20");
+	}
+
     private Connection.Request req;
     private Connection.Response res;
 
@@ -50,7 +56,7 @@ public Connection url(URL url) {
     public Connection url(String url) {
         Validate.notEmpty(url, "Must supply a valid URL");
         try {
-            req.url(new URL(url));
+            req.url(new URL(encodeUrl(url)));
         } catch (MalformedURLException e) {
             throw new IllegalArgumentException("Malformed URL: " + url, e);
         }
@@ -447,7 +453,7 @@ else if (!req.ignoreHttpErrors())
                 if (needsRedirect && req.followRedirects()) {
                     req.method(Method.GET); // always redirect with a get. any data param from original req are dropped.
                     req.data().clear();
-                    req.url(new URL(req.url(), res.header("Location")));
+                    req.url(new URL(req.url(), encodeUrl(res.header("Location"))));
                     for (Map.Entry<String, String> cookie : res.cookies.entrySet()) { // add response cookies to request (for e.g. login posts)
                         req.cookie(cookie.getKey(), cookie.getValue());
                     }

diff --git a/src/test/java/org/jsoup/integration/UrlConnectTest.java b/src/test/java/org/jsoup/integration/UrlConnectTest.java
@@ -29,6 +29,13 @@ public void fetchURl() throws IOException {
         assertTrue(doc.title().contains("Google"));
     }
 
+    @Test
+    public void fetchURIWithWihtespace() throws IOException {
+        Connection con = Jsoup.connect("http://try.jsoup.org/#with whitespaces");
+        Document doc = con.get();
+        assertTrue(doc.title().contains("jsoup"));
+    }
+
     @Test
     public void fetchBaidu() throws IOException {
         Connection.Response res = Jsoup.connect("http://www.baidu.com/").timeout(10*1000).execute();
@@ -140,6 +147,13 @@ public void followsRelativeRedirect() throws IOException {
         assertTrue(doc.title().contains("HTML Tidy Online"));
     }
 
+    @Test
+    public void followsRedirectsWithWithespaces() throws IOException {
+        Connection con = Jsoup.connect("http://tinyurl.com/kgofxl8"); // to http://www.google.com/?q=white spaces
+        Document doc = con.get();
+        assertTrue(doc.title().contains("Google"));
+    }
+
     @Test
     public void throwsExceptionOnError() {
         String url = "http://direct.infohound.net/tools/404";