From a4004904fed081ac6f46c84178073900205bd325 Mon Sep 17 00:00:00 2001
From: Romain DEP <rom1dep@gmail.com>
Date: Mon, 16 Jul 2018 20:50:23 +0200
Subject: [PATCH] Patterns: attempt at fixing complex xmpp URIs (typically
 xmpp:#mychan%irc.server@xmpp.server.tld?join) (#3115)

---
 src/main/java/eu/siacs/conversations/utils/Patterns.java | 9 +++++----
 src/main/java/eu/siacs/conversations/utils/XmppUri.java  | 5 ++++-
 2 files changed, 9 insertions(+), 5 deletions(-)

diff --git a/src/main/java/eu/siacs/conversations/utils/Patterns.java b/src/main/java/eu/siacs/conversations/utils/Patterns.java
index fae13aaea..dcbb24a0c 100644
--- a/src/main/java/eu/siacs/conversations/utils/Patterns.java
+++ b/src/main/java/eu/siacs/conversations/utils/Patterns.java
@@ -28,10 +28,11 @@ import java.util.regex.Pattern;
 public class Patterns {
 
     public static final Pattern XMPP_PATTERN = Pattern
-            .compile("xmpp\\:(?:(?:["
-                    + Patterns.GOOD_IRI_CHAR
-                    + "\\;\\/\\?\\@\\&\\=\\#\\~\\-\\.\\+\\!\\*\\'\\(\\)\\,\\_])"
-                    + "|(?:\\%[a-fA-F0-9]{2}))+");
+            .compile("xmpp:"
+                    + "(?:\\S+)" // any (one or more) non-whitespace prefix
+                    + "@(?:[." + Patterns.GOOD_IRI_CHAR + "])+" // domain part, added the dot
+                    + "(?:\\?[a-zA-Z]+)?" // optional action, can be any letter but no digit
+                    );
 
     /**
      *  Regular expression to match all IANA top-level domains.
diff --git a/src/main/java/eu/siacs/conversations/utils/XmppUri.java b/src/main/java/eu/siacs/conversations/utils/XmppUri.java
index 0deb66a80..e9ebb6534 100644
--- a/src/main/java/eu/siacs/conversations/utils/XmppUri.java
+++ b/src/main/java/eu/siacs/conversations/utils/XmppUri.java
@@ -91,7 +91,10 @@ public class XmppUri {
 			if (uri.getAuthority() != null) {
 				jid = uri.getAuthority();
 			} else {
-				String[] parts = uri.getSchemeSpecificPart().split("\\?");
+				// At this point, it might be safe to assume that any instance of '%' or '#' in the
+				// passed URI is the result of the user input and can safely be (should be?) encoded
+				Uri replaced = Uri.parse(lameUrlEncode(uri.toString()));
+				String[] parts = replaced.getSchemeSpecificPart().split("\\?");
 				if (parts.length > 0) {
 					jid = parts[0];
 				} else {