From: Robin Rosenberg <robin.rosenberg@dewire.com>
Date: Tue, 29 Nov 2011 23:16:40 +0000 (+0100)
Subject: Add percent-encoding of reserved characters in URIish
X-Git-Tag: v1.3.0.201202121842-rc4~66^2
X-Git-Url: https://source.dussan.org/?a=commitdiff_plain;h=refs%2Fchanges%2F50%2F4550%2F8;p=jgit.git

Add percent-encoding of reserved characters in URIish

We do this for the the names that have an explicit scheme and
do it both ways. The URIish is parsed before decoding. Only
a few special characters are encoded for the path part of the
URI, i.e. space, non-ASCII and control characters. The percent
encoding is assumed to be a stream encoding so we interpret it
as UTF-8.

Change-Id: I82d1910df9472e21d7212a2b984ff7d8fb2cbf0f
---

diff --git a/org.eclipse.jgit.test/tst/org/eclipse/jgit/transport/URIishTest.java b/org.eclipse.jgit.test/tst/org/eclipse/jgit/transport/URIishTest.java
index d797dd435e..b5a753bbba 100644
--- a/org.eclipse.jgit.test/tst/org/eclipse/jgit/transport/URIishTest.java
+++ b/org.eclipse.jgit.test/tst/org/eclipse/jgit/transport/URIishTest.java
@@ -88,8 +88,10 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertNull(u.getScheme());
 		assertFalse(u.isRemote());
+		assertEquals(str, u.getRawPath());
 		assertEquals(str, u.getPath());
 		assertEquals(str, u.toString());
+		assertEquals(str, u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -99,8 +101,10 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertNull(u.getScheme());
 		assertFalse(u.isRemote());
+		assertEquals(str, u.getRawPath());
 		assertEquals(str, u.getPath());
 		assertEquals(str, u.toString());
+		assertEquals(str, u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -110,8 +114,10 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertNull(u.getScheme());
 		assertFalse(u.isRemote());
+		assertEquals("D:\\m y", u.getRawPath());
 		assertEquals("D:\\m y", u.getPath());
 		assertEquals("D:\\m y", u.toString());
+		assertEquals("D:\\m y", u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -121,8 +127,10 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertNull(u.getScheme());
 		assertFalse(u.isRemote());
+		assertEquals(str, u.getRawPath());
 		assertEquals(str, u.getPath());
 		assertEquals(str, u.toString());
+		assertEquals(str, u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -132,8 +140,10 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertNull(u.getScheme());
 		assertFalse(u.isRemote());
+		assertEquals("\\\\some\\place", u.getRawPath());
 		assertEquals("\\\\some\\place", u.getPath());
 		assertEquals("\\\\some\\place", u.toString());
+		assertEquals("\\\\some\\place", u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -143,8 +153,49 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertEquals("file", u.getScheme());
 		assertFalse(u.isRemote());
+		assertEquals("/home/m y", u.getRawPath());
 		assertEquals("/home/m y", u.getPath());
-		assertEquals(str, u.toString());
+		assertEquals("file:///home/m y", u.toString());
+		assertEquals("file:///home/m%20y", u.toASCIIString());
+		assertEquals(u, new URIish(str));
+	}
+
+	@Test
+	public void testURIEncode_00() throws Exception {
+		final String str = "file:///home/m%00y";
+		URIish u = new URIish(str);
+		assertEquals("file", u.getScheme());
+		assertFalse(u.isRemote());
+		assertEquals("/home/m%00y", u.getRawPath());
+		assertEquals("/home/m\u0000y", u.getPath());
+		assertEquals("file:///home/m%00y", u.toString());
+		assertEquals("file:///home/m%00y", u.toASCIIString());
+		assertEquals(u, new URIish(str));
+	}
+
+	@Test
+	public void testURIEncode_0a() throws Exception {
+		final String str = "file:///home/m%0ay";
+		URIish u = new URIish(str);
+		assertEquals("file", u.getScheme());
+		assertFalse(u.isRemote());
+		assertEquals("/home/m%0ay", u.getRawPath());
+		assertEquals("/home/m\ny", u.getPath());
+		assertEquals("file:///home/m%0ay", u.toString());
+		assertEquals("file:///home/m%0ay", u.toASCIIString());
+		assertEquals(u, new URIish(str));
+	}
+
+	@Test
+	public void testURIEncode_unicode() throws Exception {
+		final String str = "file:///home/m%c3%a5y";
+		URIish u = new URIish(str);
+		assertEquals("file", u.getScheme());
+		assertFalse(u.isRemote());
+		assertEquals("/home/m%c3%a5y", u.getRawPath());
+		assertEquals("/home/m\u00e5y", u.getPath());
+		assertEquals("file:///home/m%c3%a5y", u.toString());
+		assertEquals("file:///home/m%c3%a5y", u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -154,8 +205,10 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertEquals("file", u.getScheme());
 		assertFalse(u.isRemote());
+		assertEquals("D:/m y", u.getRawPath());
 		assertEquals("D:/m y", u.getPath());
-		assertEquals(str, u.toString());
+		assertEquals("file:///D:/m y", u.toString());
+		assertEquals("file:///D:/m%20y", u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -166,8 +219,10 @@ public class URIishTest {
 		assertEquals("git", u.getScheme());
 		assertTrue(u.isRemote());
 		assertEquals("example.com", u.getHost());
+		assertEquals("/home/m y", u.getRawPath());
 		assertEquals("/home/m y", u.getPath());
-		assertEquals(str, u.toString());
+		assertEquals("git://example.com/home/m y", u.toString());
+		assertEquals("git://example.com/home/m%20y", u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -178,9 +233,11 @@ public class URIishTest {
 		assertEquals("git", u.getScheme());
 		assertTrue(u.isRemote());
 		assertEquals("example.com", u.getHost());
+		assertEquals("/home/m y", u.getRawPath());
 		assertEquals("/home/m y", u.getPath());
 		assertEquals(333, u.getPort());
-		assertEquals(str, u.toString());
+		assertEquals("git://example.com:333/home/m y", u.toString());
+		assertEquals("git://example.com:333/home/m%20y", u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -190,10 +247,12 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertEquals("git", u.getScheme());
 		assertTrue(u.isRemote());
+		assertEquals("D:/m y", u.getRawPath());
 		assertEquals("D:/m y", u.getPath());
 		assertEquals(338, u.getPort());
 		assertEquals("example.com", u.getHost());
-		assertEquals(str, u.toString());
+		assertEquals("git://example.com:338/D:/m y", u.toString());
+		assertEquals("git://example.com:338/D:/m%20y", u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -203,10 +262,27 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertEquals("git", u.getScheme());
 		assertTrue(u.isRemote());
+		assertEquals("D:/m y", u.getRawPath());
 		assertEquals("D:/m y", u.getPath());
 		assertEquals("example.com", u.getHost());
 		assertEquals(-1, u.getPort());
+		assertEquals("git://example.com/D:/m y", u.toString());
+		assertEquals("git://example.com/D:/m%20y", u.toASCIIString());
+		assertEquals(u, new URIish(str));
+	}
+
+	@Test
+	public void testScpStyleNoURIDecoding() throws Exception {
+		final String str = "example.com:some/p%20ath";
+		URIish u = new URIish(str);
+		assertNull(u.getScheme());
+		assertTrue(u.isRemote());
+		assertEquals("some/p%20ath", u.getRawPath());
+		assertEquals("some/p%20ath", u.getPath());
+		assertEquals("example.com", u.getHost());
+		assertEquals(-1, u.getPort());
 		assertEquals(str, u.toString());
+		assertEquals(str, u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -216,10 +292,12 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertNull(u.getScheme());
 		assertTrue(u.isRemote());
+		assertEquals("some/p ath", u.getRawPath());
 		assertEquals("some/p ath", u.getPath());
 		assertEquals("example.com", u.getHost());
 		assertEquals(-1, u.getPort());
 		assertEquals(str, u.toString());
+		assertEquals(str, u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -229,10 +307,12 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertNull(u.getScheme());
 		assertTrue(u.isRemote());
+		assertEquals("/some/p ath", u.getRawPath());
 		assertEquals("/some/p ath", u.getPath());
 		assertEquals("example.com", u.getHost());
 		assertEquals(-1, u.getPort());
 		assertEquals(str, u.toString());
+		assertEquals(str, u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -242,11 +322,13 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertNull(u.getScheme());
 		assertTrue(u.isRemote());
+		assertEquals("some/p ath", u.getRawPath());
 		assertEquals("some/p ath", u.getPath());
 		assertEquals("user", u.getUser());
 		assertEquals("example.com", u.getHost());
 		assertEquals(-1, u.getPort());
 		assertEquals(str, u.toString());
+		assertEquals(str, u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -256,10 +338,12 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertEquals("git+ssh", u.getScheme());
 		assertTrue(u.isRemote());
+		assertEquals("/some/p ath", u.getRawPath());
 		assertEquals("/some/p ath", u.getPath());
 		assertEquals("example.com", u.getHost());
 		assertEquals(-1, u.getPort());
-		assertEquals(str, u.toString());
+		assertEquals("git+ssh://example.com/some/p ath", u.toString());
+		assertEquals("git+ssh://example.com/some/p%20ath", u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -269,10 +353,12 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertEquals("ssh+git", u.getScheme());
 		assertTrue(u.isRemote());
+		assertEquals("/some/p ath", u.getRawPath());
 		assertEquals("/some/p ath", u.getPath());
 		assertEquals("example.com", u.getHost());
 		assertEquals(-1, u.getPort());
-		assertEquals(str, u.toString());
+		assertEquals("ssh+git://example.com/some/p ath", u.toString());
+		assertEquals("ssh+git://example.com/some/p%20ath", u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -282,10 +368,12 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertEquals("ssh", u.getScheme());
 		assertTrue(u.isRemote());
+		assertEquals("/some/p ath", u.getRawPath());
 		assertEquals("/some/p ath", u.getPath());
 		assertEquals("example.com", u.getHost());
 		assertEquals(-1, u.getPort());
-		assertEquals(str, u.toString());
+		assertEquals("ssh://example.com/some/p ath", u.toString());
+		assertEquals("ssh://example.com/some/p%20ath", u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -295,12 +383,15 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertEquals("ssh", u.getScheme());
 		assertTrue(u.isRemote());
+		assertEquals("/some/p ath", u.getRawPath());
 		assertEquals("/some/p ath", u.getPath());
 		assertEquals("example.com", u.getHost());
 		assertEquals("user", u.getUser());
 		assertNull(u.getPass());
 		assertEquals(33, u.getPort());
-		assertEquals(str, u.toString());
+		assertEquals("ssh://user@example.com:33/some/p ath", u.toString());
+		assertEquals("ssh://user@example.com:33/some/p%20ath",
+				u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -310,13 +401,18 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertEquals("ssh", u.getScheme());
 		assertTrue(u.isRemote());
+		assertEquals("/some/p ath", u.getRawPath());
 		assertEquals("/some/p ath", u.getPath());
 		assertEquals("example.com", u.getHost());
 		assertEquals("user", u.getUser());
 		assertEquals("pass", u.getPass());
 		assertEquals(33, u.getPort());
-		assertEquals(str, u.toPrivateString());
+		assertEquals("ssh://user:pass@example.com:33/some/p ath",
+				u.toPrivateString());
+		assertEquals("ssh://user:pass@example.com:33/some/p%20ath",
+				u.toPrivateASCIIString());
 		assertEquals(u.setPass(null).toPrivateString(), u.toString());
+		assertEquals(u.setPass(null).toPrivateASCIIString(), u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -326,13 +422,62 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertEquals("ssh", u.getScheme());
 		assertTrue(u.isRemote());
+		assertEquals("/some/p ath", u.getRawPath());
 		assertEquals("/some/p ath", u.getPath());
 		assertEquals("example.com", u.getHost());
 		assertEquals("DOMAIN\\user", u.getUser());
 		assertEquals("pass", u.getPass());
 		assertEquals(33, u.getPort());
-		assertEquals(str, u.toPrivateString());
+		assertEquals("ssh://DOMAIN\\user:pass@example.com:33/some/p ath",
+				u.toPrivateString());
+		assertEquals("ssh://DOMAIN\\user:pass@example.com:33/some/p%20ath",
+				u.toPrivateASCIIString());
+		assertEquals(u.setPass(null).toPrivateString(), u.toString());
+		assertEquals(u.setPass(null).toPrivateASCIIString(), u.toASCIIString());
+		assertEquals(u, new URIish(str));
+	}
+
+	@Test
+	public void testSshProtoWithEscapedADUserPassAndPort() throws Exception {
+		final String str = "ssh://DOMAIN%5c\u00fcser:pass@example.com:33/some/p ath";
+		URIish u = new URIish(str);
+		assertEquals("ssh", u.getScheme());
+		assertTrue(u.isRemote());
+		assertEquals("/some/p ath", u.getRawPath());
+		assertEquals("/some/p ath", u.getPath());
+		assertEquals("example.com", u.getHost());
+		assertEquals("DOMAIN\\\u00fcser", u.getUser());
+		assertEquals("pass", u.getPass());
+		assertEquals(33, u.getPort());
+		assertEquals("ssh://DOMAIN\\\u00fcser:pass@example.com:33/some/p ath",
+				u.toPrivateString());
+		assertEquals(
+				"ssh://DOMAIN\\%c3%bcser:pass@example.com:33/some/p%20ath",
+				u.toPrivateASCIIString());
+		assertEquals(u.setPass(null).toPrivateString(), u.toString());
+		assertEquals(u.setPass(null).toPrivateASCIIString(), u.toASCIIString());
+		assertEquals(u, new URIish(str));
+	}
+
+	@Test
+	public void testURIEncodeDecode() throws Exception {
+		final String str = "ssh://%3ax%25:%40%41x@example.com:33/some%c3%a5/p%20a th";
+		URIish u = new URIish(str);
+		assertEquals("ssh", u.getScheme());
+		assertTrue(u.isRemote());
+		assertEquals("/some%c3%a5/p%20a th", u.getRawPath());
+		assertEquals("/some\u00e5/p a th", u.getPath());
+		assertEquals("example.com", u.getHost());
+		assertEquals(":x%", u.getUser());
+		assertEquals("@Ax", u.getPass());
+		assertEquals(33, u.getPort());
+		assertEquals("ssh://%3ax%25:%40Ax@example.com:33/some%c3%a5/p%20a th",
+				u.toPrivateString());
+		assertEquals(
+				"ssh://%3ax%25:%40Ax@example.com:33/some%c3%a5/p%20a%20th",
+				u.toPrivateASCIIString());
 		assertEquals(u.setPass(null).toPrivateString(), u.toString());
+		assertEquals(u.setPass(null).toPrivateASCIIString(), u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -342,13 +487,17 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertEquals("git", u.getScheme());
 		assertTrue(u.isRemote());
+		assertEquals("~some/p ath", u.getRawPath());
 		assertEquals("~some/p ath", u.getPath());
 		assertEquals("example.com", u.getHost());
 		assertNull(u.getUser());
 		assertNull(u.getPass());
 		assertEquals(-1, u.getPort());
-		assertEquals(str, u.toPrivateString());
+		assertEquals("git://example.com/~some/p ath", u.toPrivateString());
+		assertEquals("git://example.com/~some/p%20ath",
+				u.toPrivateASCIIString());
 		assertEquals(u.setPass(null).toPrivateString(), u.toString());
+		assertEquals(u.setPass(null).toPrivateASCIIString(), u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -359,13 +508,16 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertEquals("git", u.getScheme());
 		assertTrue(u.isRemote());
+		assertEquals("~some/p ath", u.getRawPath());
 		assertEquals("~some/p ath", u.getPath());
 		assertEquals("example.com", u.getHost());
 		assertNull(u.getUser());
 		assertNull(u.getPass());
 		assertEquals(-1, u.getPort());
 		assertEquals(str, u.toPrivateString());
+		assertEquals(str, u.toPrivateASCIIString());
 		assertEquals(u.setPass(null).toPrivateString(), u.toString());
+		assertEquals(u.setPass(null).toPrivateASCIIString(), u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -375,13 +527,16 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertNull(u.getScheme());
 		assertFalse(u.isRemote());
+		assertEquals("/~some/p ath", u.getRawPath());
 		assertEquals("/~some/p ath", u.getPath());
 		assertNull(u.getHost());
 		assertNull(u.getUser());
 		assertNull(u.getPass());
 		assertEquals(-1, u.getPort());
 		assertEquals(str, u.toPrivateString());
+		assertEquals(str, u.toPrivateASCIIString());
 		assertEquals(u.setPass(null).toPrivateString(), u.toString());
+		assertEquals(u.setPass(null).toPrivateASCIIString(), u.toASCIIString());
 		assertEquals(u, new URIish(str));
 	}
 
@@ -415,6 +570,33 @@ public class URIishTest {
 		}
 	}
 
+	@Test
+	public void testGetSet() throws Exception {
+		final String str = "ssh://DOMAIN\\user:pass@example.com:33/some/p ath%20";
+		URIish u = new URIish(str);
+		u = u.setHost(u.getHost());
+		u = u.setPass(u.getPass());
+		u = u.setPort(u.getPort());
+		assertEquals("ssh", u.getScheme());
+		assertTrue(u.isRemote());
+		u = u.setRawPath(u.getRawPath());
+		assertEquals("/some/p ath%20", u.getRawPath());
+		u = u.setPath(u.getPath());
+		assertEquals("/some/p ath ", u.getRawPath());
+		assertEquals("/some/p ath ", u.getPath());
+		assertEquals("example.com", u.getHost());
+		assertEquals("DOMAIN\\user", u.getUser());
+		assertEquals("pass", u.getPass());
+		assertEquals(33, u.getPort());
+		assertEquals("ssh://DOMAIN\\user:pass@example.com:33/some/p ath ",
+				u.toPrivateString());
+		assertEquals("ssh://DOMAIN\\user:pass@example.com:33/some/p%20ath%20",
+				u.toPrivateASCIIString());
+		assertEquals(u.setPass(null).toPrivateString(), u.toString());
+		assertEquals(u.setPass(null).toPrivateASCIIString(), u.toASCIIString());
+		assertEquals(u, new URIish(str));
+	}
+
 	@Test
 	public void testGetValidWithEmptySlashDotGitHumanishName()
 			throws IllegalArgumentException, URISyntaxException {
@@ -537,6 +719,7 @@ public class URIishTest {
 		URIish u = new URIish(str);
 		assertEquals("http", u.getScheme());
 		assertTrue(u.isRemote());
+		assertEquals("/some/path", u.getRawPath());
 		assertEquals("/some/path", u.getPath());
 		assertEquals("host.xy", u.getHost());
 		assertEquals(80, u.getPort());
@@ -556,6 +739,27 @@ public class URIishTest {
 		assertEquals(u, new URIish(str));
 	}
 
+	/**
+	 * Exemplify what happens with the special case of encoding '/' as %2F. Web
+	 * services in general parse path components before decoding the characters.
+	 *
+	 * @throws URISyntaxException
+	 */
+	@Test
+	public void testPathSeparator() throws URISyntaxException {
+		String str = "http://user:secret@host.xy:80/some%2Fpath";
+		URIish u = new URIish(str);
+		assertEquals("http", u.getScheme());
+		assertTrue(u.isRemote());
+		assertEquals("/some%2Fpath", u.getRawPath());
+		assertEquals("/some/path", u.getPath());
+		assertEquals("host.xy", u.getHost());
+		assertEquals(80, u.getPort());
+		assertEquals("user", u.getUser());
+		assertEquals("secret", u.getPass());
+		assertEquals(u, new URIish(str));
+	}
+
 	@Test
 	public void testFileProtocol() throws IllegalArgumentException,
 			URISyntaxException, IOException {
@@ -565,6 +769,7 @@ public class URIishTest {
 		assertFalse(u.isRemote());
 		assertNull(u.getHost());
 		assertNull(u.getPass());
+		assertEquals("/a/b.txt", u.getRawPath());
 		assertEquals("/a/b.txt", u.getPath());
 		assertEquals(-1, u.getPort());
 		assertNull(u.getUser());
@@ -586,6 +791,7 @@ public class URIishTest {
 		assertFalse(u.isRemote());
 		assertNull(u.getHost());
 		assertNull(u.getPass());
+		assertEquals("/a/b.txt", u.getRawPath());
 		assertEquals("/a/b.txt", u.getPath());
 		assertEquals(-1, u.getPort());
 		assertNull(u.getUser());
diff --git a/org.eclipse.jgit/src/org/eclipse/jgit/transport/URIish.java b/org.eclipse.jgit/src/org/eclipse/jgit/transport/URIish.java
index 8254c1f20e..de18ece355 100644
--- a/org.eclipse.jgit/src/org/eclipse/jgit/transport/URIish.java
+++ b/org.eclipse.jgit/src/org/eclipse/jgit/transport/URIish.java
@@ -46,22 +46,26 @@
 
 package org.eclipse.jgit.transport;
 
+import java.io.ByteArrayOutputStream;
 import java.io.File;
 import java.io.Serializable;
+import java.io.UnsupportedEncodingException;
 import java.net.URISyntaxException;
 import java.net.URL;
+import java.util.BitSet;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
 
 import org.eclipse.jgit.JGitText;
 import org.eclipse.jgit.lib.Constants;
+import org.eclipse.jgit.util.RawParseUtils;
 import org.eclipse.jgit.util.StringUtils;
 
 /**
  * This URI like construct used for referencing Git archives over the net, as
- * well as locally stored archives. The most important difference compared to
- * RFC 2396 URI's is that no URI encoding/decoding ever takes place. A space or
- * any special character is written as-is.
+ * well as locally stored archives. It is similar to RFC 2396 URI's, but also
+ * support SCP and the malformed file://<path> syntax (as opposed to the correct
+ * file:<path> syntax.
  */
 public class URIish implements Serializable {
 	/**
@@ -179,6 +183,8 @@ public class URIish implements Serializable {
 
 	private String path;
 
+	private String rawPath;
+
 	private String user;
 
 	private String pass;
@@ -201,19 +207,21 @@ public class URIish implements Serializable {
 		Matcher matcher = SINGLE_SLASH_FILE_URI.matcher(s);
 		if (matcher.matches()) {
 			scheme = matcher.group(1);
-			path = cleanLeadingSlashes(matcher.group(2), scheme);
+			rawPath = cleanLeadingSlashes(matcher.group(2), scheme);
+			path = unescape(rawPath);
 			return;
 		}
 		matcher = FULL_URI.matcher(s);
 		if (matcher.matches()) {
 			scheme = matcher.group(1);
-			user = matcher.group(2);
-			pass = matcher.group(3);
-			host = matcher.group(4);
+			user = unescape(matcher.group(2));
+			pass = unescape(matcher.group(3));
+			host = unescape(matcher.group(4));
 			if (matcher.group(5) != null)
 				port = Integer.parseInt(matcher.group(5));
-			path = cleanLeadingSlashes(
+			rawPath = cleanLeadingSlashes(
 					n2e(matcher.group(6)) + n2e(matcher.group(7)), scheme);
+			path = unescape(rawPath);
 			return;
 		}
 		matcher = RELATIVE_SCP_URI.matcher(s);
@@ -221,7 +229,8 @@ public class URIish implements Serializable {
 			user = matcher.group(1);
 			pass = matcher.group(2);
 			host = matcher.group(3);
-			path = matcher.group(4);
+			rawPath = matcher.group(4);
+			path = rawPath;
 			return;
 		}
 		matcher = ABSOLUTE_SCP_URI.matcher(s);
@@ -229,17 +238,95 @@ public class URIish implements Serializable {
 			user = matcher.group(1);
 			pass = matcher.group(2);
 			host = matcher.group(3);
-			path = matcher.group(4);
+			rawPath = matcher.group(4);
+			path = rawPath;
 			return;
 		}
 		matcher = LOCAL_FILE.matcher(s);
 		if (matcher.matches()) {
-			path = matcher.group(1);
+			rawPath = matcher.group(1);
+			path = rawPath;
 			return;
 		}
 		throw new URISyntaxException(s, JGitText.get().cannotParseGitURIish);
 	}
 
+	private static String unescape(String s) throws URISyntaxException {
+		if (s == null)
+			return null;
+		if (s.indexOf('%') < 0)
+			return s;
+
+		byte[] bytes;
+		try {
+			bytes = s.getBytes(Constants.CHARACTER_ENCODING);
+		} catch (UnsupportedEncodingException e) {
+			throw new RuntimeException(e); // can't happen
+		}
+
+		byte[] os = new byte[bytes.length];
+		int j = 0;
+		for (int i = 0; i < bytes.length; ++i) {
+			byte c = bytes[i];
+			if (c == '%') {
+				if (i + 2 >= bytes.length)
+					throw new URISyntaxException(s, JGitText.get().cannotParseGitURIish);
+				int val = (RawParseUtils.parseHexInt4(bytes[i + 1]) << 4)
+						| RawParseUtils.parseHexInt4(bytes[i + 2]);
+				os[j++] = (byte) val;
+				i += 2;
+			} else
+				os[j++] = c;
+		}
+		return RawParseUtils.decode(os, 0, j);
+	}
+
+	private static final BitSet reservedChars = new BitSet(127);
+
+	static {
+		for (byte b : Constants.encodeASCII("!*'();:@&=+$,/?#[]"))
+			reservedChars.set(b);
+	}
+
+	/**
+	 * Escape unprintable characters optionally URI-reserved characters
+	 *
+	 * @param s
+	 *            The Java String to encode (may contain any character)
+	 * @param escapeReservedChars
+	 *            true to escape URI reserved characters
+	 * @param encodeNonAscii
+	 *            encode any non-ASCII characters
+	 * @return a URI-encoded string
+	 */
+	private static String escape(String s, boolean escapeReservedChars,
+			boolean encodeNonAscii) {
+		if (s == null)
+			return null;
+		ByteArrayOutputStream os = new ByteArrayOutputStream(s.length());
+		byte[] bytes;
+		try {
+			bytes = s.getBytes(Constants.CHARACTER_ENCODING);
+		} catch (UnsupportedEncodingException e) {
+			throw new RuntimeException(e); // cannot happen
+		}
+		for (int i = 0; i < bytes.length; ++i) {
+			int b = bytes[i] & 0xFF;
+			if (b <= 32 || (encodeNonAscii && b > 127) || b == '%'
+					|| (escapeReservedChars && reservedChars.get(b))) {
+				os.write('%');
+				byte[] tmp = Constants.encodeASCII(String.format("%02x",
+						Integer.valueOf(b)));
+				os.write(tmp[0]);
+				os.write(tmp[1]);
+			} else {
+				os.write(b);
+			}
+		}
+		byte[] buf = os.toByteArray();
+		return RawParseUtils.decode(buf, 0, buf.length);
+	}
+
 	private String n2e(String s) {
 		if (s == null)
 			return "";
@@ -272,6 +359,11 @@ public class URIish implements Serializable {
 	public URIish(final URL u) {
 		scheme = u.getProtocol();
 		path = u.getPath();
+		try {
+			rawPath = u.toURI().getRawPath();
+		} catch (URISyntaxException e) {
+			throw new RuntimeException(e); // Impossible
+		}
 
 		final String ui = u.getUserInfo();
 		if (ui != null) {
@@ -291,6 +383,7 @@ public class URIish implements Serializable {
 
 	private URIish(final URIish u) {
 		this.scheme = u.scheme;
+		this.rawPath = u.rawPath;
 		this.path = u.path;
 		this.user = u.user;
 		this.pass = u.pass;
@@ -352,6 +445,13 @@ public class URIish implements Serializable {
 		return path;
 	}
 
+	/**
+	 * @return path name component
+	 */
+	public String getRawPath() {
+		return rawPath;
+	}
+
 	/**
 	 * Return a new URI matching this one, but with a different path.
 	 *
@@ -362,6 +462,22 @@ public class URIish implements Serializable {
 	public URIish setPath(final String n) {
 		final URIish r = new URIish(this);
 		r.path = n;
+		r.rawPath = n;
+		return r;
+	}
+
+	/**
+	 * Return a new URI matching this one, but with a different (raw) path.
+	 *
+	 * @param n
+	 *            the new value for path.
+	 * @return a new URI with the updated value.
+	 * @throws URISyntaxException
+	 */
+	public URIish setRawPath(final String n) throws URISyntaxException {
+		final URIish r = new URIish(this);
+		r.path = unescape(n);
+		r.rawPath = n;
 		return r;
 	}
 
@@ -475,14 +591,15 @@ public class URIish implements Serializable {
 	 * @return the URI, including its password field, if any.
 	 */
 	public String toPrivateString() {
-		return format(true);
+		return format(true, false, false);
 	}
 
 	public String toString() {
-		return format(false);
+		return format(false, false, false);
 	}
 
-	private String format(final boolean includePassword) {
+	private String format(final boolean includePassword, boolean escape,
+			boolean escapeNonAscii) {
 		final StringBuilder r = new StringBuilder();
 		if (getScheme() != null) {
 			r.append(getScheme());
@@ -490,17 +607,17 @@ public class URIish implements Serializable {
 		}
 
 		if (getUser() != null) {
-			r.append(getUser());
+			r.append(escape(getUser(), true, escapeNonAscii));
 			if (includePassword && getPass() != null) {
 				r.append(':');
-				r.append(getPass());
+				r.append(escape(getPass(), true, escapeNonAscii));
 			}
 		}
 
 		if (getHost() != null) {
 			if (getUser() != null)
 				r.append('@');
-			r.append(getHost());
+			r.append(escape(getHost(), false, escapeNonAscii));
 			if (getScheme() != null && getPort() > 0) {
 				r.append(':');
 				r.append(getPort());
@@ -513,12 +630,33 @@ public class URIish implements Serializable {
 					r.append('/');
 			} else if (getHost() != null)
 				r.append(':');
-			r.append(getPath());
+			if (getScheme() != null)
+				if (escapeNonAscii)
+					r.append(escape(getPath(), false, escapeNonAscii));
+				else
+					r.append(getRawPath());
+			else
+				r.append(getPath());
 		}
 
 		return r.toString();
 	}
 
+	/**
+	 * @return the URI as an ASCII string. Password is not included.
+	 */
+	public String toASCIIString() {
+		return format(false, true, true);
+	}
+
+	/**
+	 * @return the URI including password, formatted with only ASCII characters
+	 *         such that it will be valid for use over the network.
+	 */
+	public String toPrivateASCIIString() {
+		return format(true, true, true);
+	}
+
 	/**
 	 * Get the "humanish" part of the path. Some examples of a 'humanish' part
 	 * for a full path: