sw HTML import: strip query & fragment for <object data="..."> and file URLs

There are two cases: when the <object> is a child of an outer <object>
(we create an embedded object) and the non-embedded <object> case (we
create an image). Call the new SwHTMLParser::StripQueryFromPath() in
both cases.

INetURLObject considers this query/fragment part as "name" (end of
"path"), so just do it manually.

Change-Id: Icb7991082b6e595db5729f3c4c84073c80834f27
Reviewed-on: https://gerrit.libreoffice.org/56989
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins

6 files changed, 29 insertions, 4 deletions

diff --git a/sw/qa/extras/htmlexport/data/reqif-ole-img.xhtml b/sw/qa/extras/htmlexport/data/reqif-ole-img.xhtml
index b552783068ae..6217412ae597 100644
--- a/sw/qa/extras/htmlexport/data/reqif-ole-img.xhtml
+++ b/sw/qa/extras/htmlexport/data/reqif-ole-img.xhtml
@@ -1,6 +1,6 @@
 <reqif-xhtml:div><reqif-xhtml:br/>
     <reqif-xhtml:object data="reqif-ole-data.ole" type="text/rtf">
-        <reqif-xhtml:object data="reqif-ole-img.png" type="image/png">OLE Object</reqif-xhtml:object>
+        <reqif-xhtml:object data="reqif-ole-img.png?test=true" type="image/png">OLE Object</reqif-xhtml:object>
     </reqif-xhtml:object>
 </reqif-xhtml:div>
 
diff --git a/sw/qa/extras/htmlexport/data/reqif-png-img.xhtml b/sw/qa/extras/htmlexport/data/reqif-png-img.xhtml
index f31795e35224..637a7c2ac4b6 100644
--- a/sw/qa/extras/htmlexport/data/reqif-png-img.xhtml
+++ b/sw/qa/extras/htmlexport/data/reqif-png-img.xhtml
@@ -1,4 +1,4 @@
 <reqif-xhtml:div><reqif-xhtml:br/>
-    <reqif-xhtml:object data="reqif-ole-img.png" type="image/png">OLE Object</reqif-xhtml:object>
+    <reqif-xhtml:object data="reqif-ole-img.png?test=true" type="image/png">OLE Object</reqif-xhtml:object>
 </reqif-xhtml:div>
 
diff --git a/sw/qa/extras/htmlexport/htmlexport.cxx b/sw/qa/extras/htmlexport/htmlexport.cxx
index 8cb60396691d..5d82b82c9524 100644
--- a/sw/qa/extras/htmlexport/htmlexport.cxx
+++ b/sw/qa/extras/htmlexport/htmlexport.cxx
@@ -391,6 +391,7 @@ DECLARE_HTMLEXPORT_ROUNDTRIP_TEST(testReqIfOleImg, "reqif-ole-img.xhtml")
     // This failed, OLE object had no replacement image.
     // And then it also failed when the export lost the replacement image.
     uno::Reference<graphic::XGraphic> xGraphic = xObject->getReplacementGraphic();
+    // This failed when query and fragment of file:// URLs were not ignored.
     CPPUNIT_ASSERT(xGraphic.is());
 
     uno::Reference<drawing::XShape> xShape(xObject, uno::UNO_QUERY);
@@ -456,6 +457,8 @@ DECLARE_HTMLEXPORT_ROUNDTRIP_TEST(testReqIfPngImg, "reqif-png-img.xhtml")
 
     // Make sure that both RTF and PNG versions are written.
     CPPUNIT_ASSERT(aStream.indexOf("text/rtf") != -1);
+    // This failed when images with a query in their file:// URL failed to
+    // import.
     CPPUNIT_ASSERT(aStream.indexOf("image/png") != -1);
 }
 
diff --git a/sw/source/filter/html/htmlgrin.cxx b/sw/source/filter/html/htmlgrin.cxx
index 0a9050f3475a..8b3ddaf0f601 100644
--- a/sw/source/filter/html/htmlgrin.cxx
+++ b/sw/source/filter/html/htmlgrin.cxx
@@ -349,7 +349,8 @@ void SwHTMLParser::InsertImage()
             case HtmlOptionId::DATA:
                 aGraphicData = rOption.GetString();
                 if (!InternalImgToPrivateURL(aGraphicData))
-                    aGraphicData = INetURLObject::GetAbsURL(m_sBaseURL, aGraphicData);
+                    aGraphicData = INetURLObject::GetAbsURL(
+                        m_sBaseURL, SwHTMLParser::StripQueryFromPath(m_sBaseURL, aGraphicData));
                 break;
             case HtmlOptionId::ALIGN:
                 eVertOri =
diff --git a/sw/source/filter/html/htmlplug.cxx b/sw/source/filter/html/htmlplug.cxx
index 30277b8bf3a9..22738a13b934 100644
--- a/sw/source/filter/html/htmlplug.cxx
+++ b/sw/source/filter/html/htmlplug.cxx
@@ -72,6 +72,7 @@
 #include <unotools/ucbstreamhelper.hxx>
 #include <comphelper/propertysequence.hxx>
 #include <filter/msfilter/msoleexp.hxx>
+#include <comphelper/fileurl.hxx>
 
 using namespace com::sun::star;
 
@@ -299,6 +300,19 @@ void SwHTMLParser::SetSpace( const Size& rPixSpace,
     }
 }
 
+OUString SwHTMLParser::StripQueryFromPath(const OUString& rBase, const OUString& rPath)
+{
+    if (!comphelper::isFileUrl(rBase))
+        return rPath;
+
+    sal_Int32 nIndex = rPath.indexOf('?');
+
+    if (nIndex != -1)
+        return rPath.copy(0, nIndex);
+
+    return rPath;
+}
+
 bool SwHTMLParser::InsertEmbed()
 {
     OUString aURL, aType, aName, aAlt, aId, aStyle, aClass;
@@ -420,9 +434,13 @@ bool SwHTMLParser::InsertEmbed()
                            INetURLObject(m_sBaseURL), aURL,
                            URIHelper::GetMaybeFileHdl()) );
     bool bHasData = !aData.isEmpty();
+
     try
     {
-        aURLObj.SetURL(rtl::Uri::convertRelToAbs(m_sBaseURL, aData));
+        // Strip query and everything after that for file:// URLs, browsers do
+        // the same.
+        aURLObj.SetURL(rtl::Uri::convertRelToAbs(
+            m_sBaseURL, SwHTMLParser::StripQueryFromPath(m_sBaseURL, aData)));
     }
     catch (const rtl::MalformedUriException& /*rException*/)
     {
diff --git a/sw/source/filter/html/swhtml.hxx b/sw/source/filter/html/swhtml.hxx
index d649f4a123e2..ae6a91ad3250 100644
--- a/sw/source/filter/html/swhtml.hxx
+++ b/sw/source/filter/html/swhtml.hxx
@@ -963,6 +963,9 @@ public:
     }
 
     void DeregisterHTMLTable(HTMLTable* pOld);
+
+    /// Strips query and fragment from a URL path if base URL is a file:// one.
+    static OUString StripQueryFromPath(const OUString& rBase, const OUString& rPath);
 };
 
 struct SwPendingStackData