From 7f14ea4431cb4182bccbd9998a60db2f9d9ae07a Mon Sep 17 00:00:00 2001
From: Michael Stahl <michael.stahl@allotropia.de>
Date: Wed, 3 Feb 2021 20:05:15 +0100
Subject: (related: tdf#121842) sw: encode URLs generated in ToX

The "|" is used as a separator there and it looks like it has never been
allowed in a URI, even in a fragment.

Change-Id: I1e4b9e12f5409f93c2291494fd4350431f68fe2b
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/110388
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@allotropia.de>
---
 sw/source/core/tox/ToxLinkProcessor.cxx | 10 +++++++++-
 sw/source/core/tox/txmsrt.cxx           |  6 +-----
 2 files changed, 10 insertions(+), 6 deletions(-)

(limited to 'sw/source/core')

diff --git a/sw/source/core/tox/ToxLinkProcessor.cxx b/sw/source/core/tox/ToxLinkProcessor.cxx
index c86cde923fef..825e37c16f65 100644
--- a/sw/source/core/tox/ToxLinkProcessor.cxx
+++ b/sw/source/core/tox/ToxLinkProcessor.cxx
@@ -13,6 +13,7 @@
 #include <SwStyleNameMapper.hxx>
 #include <ndtxt.hxx>
 #include <sal/log.hxx>
+#include <rtl/uri.hxx>
 
 namespace sw {
 
@@ -37,8 +38,15 @@ ToxLinkProcessor::CloseLink(sal_Int32 endPosition, const OUString& url)
         return;
     }
 
+    // url contains '|' which must be encoded; also in some cases contains
+    // arbitrary strings that need to be encoded
+    assert(url[0] == '#'); // all links are internal
+    OUString const uri("#" + rtl::Uri::encode(url.copy(1),
+        rtl_UriCharClassUricNoSlash,
+        rtl_UriEncodeIgnoreEscapes, RTL_TEXTENCODING_UTF8));
+
     std::unique_ptr<ClosedLink> pClosedLink(
-            new ClosedLink(url, m_pStartedLink->mStartPosition, endPosition));
+            new ClosedLink(uri, m_pStartedLink->mStartPosition, endPosition));
 
     const OUString& characterStyle = m_pStartedLink->mCharacterStyle;
     sal_uInt16 poolId = ObtainPoolId(characterStyle);
diff --git a/sw/source/core/tox/txmsrt.cxx b/sw/source/core/tox/txmsrt.cxx
index f626f09afc40..3bdd2871c23c 100644
--- a/sw/source/core/tox/txmsrt.cxx
+++ b/sw/source/core/tox/txmsrt.cxx
@@ -19,7 +19,6 @@
 
 #include <unotools/charclass.hxx>
 #include <osl/diagnose.h>
-#include <rtl/uri.hxx>
 #include <txtfld.hxx>
 #include <doc.hxx>
 #include <IDocumentLayoutAccess.hxx>
@@ -204,10 +203,7 @@ std::pair<OUString, bool> SwTOXSortTabBase::GetURL(SwRootFrame const*const pLayo
         + OUStringChar(toxMarkSeparator) + typeName
         + OUStringChar(cMarkSeparator) + "toxmark" );
 
-    OUString const uri(rtl::Uri::encode(decodedUrl, rtl_UriCharClassUricNoSlash,
-        rtl_UriEncodeIgnoreEscapes, RTL_TEXTENCODING_UTF8));
-
-    return std::make_pair(uri, true);
+    return std::make_pair(decodedUrl, true);
 }
 
 bool SwTOXSortTabBase::IsFullPara() const
-- 
cgit