sw HTML import: allow custom XHTML namespace alias

This helps in case the HTML filter is given an XHTML that has an
explicit namespace alias for <http://www.w3.org/1999/xhtml>.

Change-Id: I437fa85ba19ce907c9c4b10c8d10aaf2217dc0ea
Reviewed-on: https://gerrit.libreoffice.org/50380Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>

include/svtools/parhtml.hxx

@@ -168,6 +168,9 @@ private:
 
     OUString aEndToken;
 
+    /// XML namespace, in case of XHTML.
+    OUString maNamespace;
+
 protected:
     OUString sSaveToken;             // the read tag as string
 
@@ -182,6 +185,8 @@ protected:
 
     void FinishHeader() { bIsInHeader = false; }
 
+    void SetNamespace(const OUString& rNamespace);
+
 public:
     HTMLParser( SvStream& rIn, bool bReadNewDoc = true );
 

svtools/source/svhtml/parhtml.cxx

@@ -237,6 +237,12 @@ HTMLParser::~HTMLParser()
 {
 }
 
+void HTMLParser::SetNamespace(const OUString& rNamespace)
+{
+    // Convert namespace alias to a prefix.
+    maNamespace = rNamespace + ":";
+}
+
 namespace
 {
     class RefGuard
@@ -1071,6 +1077,10 @@ HtmlTokenId HTMLParser::GetNextToken_()
                     // Search token in table:
                     sSaveToken = aToken;
                     aToken = aToken.toAsciiLowerCase();
+
+                    if (!maNamespace.isEmpty() && aToken.startsWith(maNamespace))
+                        aToken = aToken.copy(maNamespace.getLength());
+
                     if( HtmlTokenId::NONE == (nRet = GetHTMLToken( aToken )) )
                         // Unknown control
                         nRet = HtmlTokenId::UNKNOWNCONTROL_ON;

sw/qa/extras/htmlimport/data/reqif-br.xhtml

+aaa<reqif-xhtml:br/>bbb

sw/qa/extras/htmlimport/htmlimport.cxx

@@ -26,6 +26,18 @@ class HtmlImportTest : public SwModelTestBase
 {
     public:
         HtmlImportTest() : SwModelTestBase("sw/qa/extras/htmlimport/data/", "HTML (StarWriter)") {}
+    private:
+        std::unique_ptr<Resetter> preTest(const char* /*filename*/) override
+        {
+            if (getTestName().indexOf("ReqIf") != -1)
+            {
+                setImportFilterOptions("xhtmlns=reqif-xhtml");
+                // Bypass type detection, this is an XHTML fragment only.
+                setImportFilterName("HTML (StarWriter)");
+            }
+
+            return nullptr;
+        }
 };
 
 #define DECLARE_HTMLIMPORT_TEST(TestName, filename) DECLARE_SW_IMPORT_TEST(TestName, filename, nullptr, HtmlImportTest)
@@ -279,6 +291,12 @@ DECLARE_HTMLIMPORT_TEST(testOutlineLevel, "outline-level.html")
                          getProperty<sal_Int32>(getParagraph(1), "OutlineLevel"));
 }
 
+DECLARE_HTMLIMPORT_TEST(testReqIfBr, "reqif-br.xhtml")
+{
+    // <reqif-xhtml:br/> was not recognized as a line break from a ReqIf file.
+    CPPUNIT_ASSERT(getParagraph(1)->getString().startsWith("aaa\nbbb"));
+}
+
 CPPUNIT_PLUGIN_IMPLEMENT();
 
 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */

sw/qa/extras/inc/swmodeltestbase.hxx

@@ -136,6 +136,8 @@ class SwModelTestBase : public test::BootstrapFixture, public unotest::MacrosTes
 {
 private:
     OUString maFilterOptions;
+    OUString maImportFilterOptions;
+    OUString maImportFilterName;
 
 protected:
     uno::Reference< lang::XComponent > mxComponent;
@@ -184,6 +186,16 @@ public:
         maFilterOptions = rFilterOptions;
     }
 
+    void setImportFilterOptions(const OUString &rFilterOptions)
+    {
+        maImportFilterOptions = rFilterOptions;
+    }
+
+    void setImportFilterName(const OUString &rFilterName)
+    {
+        maImportFilterName = rFilterName;
+    }
+
     SwModelTestBase(const OUString& pTestDocumentPath = OUString(), const char* pFilter = "")
         : mpXmlBuffer(nullptr)
         , mpTestDocumentPath(pTestDocumentPath)
@@ -662,6 +674,22 @@ protected:
             setTestInteractionHandler(pPassword, aFilterOptions);
         }
 
+        if (!maImportFilterOptions.isEmpty())
+        {
+            beans::PropertyValue aValue;
+            aValue.Name = "FilterOptions";
+            aValue.Value <<= maImportFilterOptions;
+            aFilterOptions.push_back(aValue);
+        }
+
+        if (!maImportFilterName.isEmpty())
+        {
+            beans::PropertyValue aValue;
+            aValue.Name = "FilterName";
+            aValue.Value <<= maImportFilterName;
+            aFilterOptions.push_back(aValue);
+        }
+
         // Output name early, so in the case of a hang, the name of the hanging input file is visible.
         if (pName)
             std::cout << pName << ":\n";

sw/source/filter/html/swhtml.cxx

@@ -417,6 +417,8 @@ SwHTMLParser::SwHTMLParser( SwDoc* pD, SwPaM& rCursor, SvStream& rIn,
             }
         }
     }
+
+    SetupFilterOptions();
 }
 
 SwHTMLParser::~SwHTMLParser()
@@ -5540,6 +5542,25 @@ void SwHTMLParser::AddMetaUserDefined( OUString const & i_rMetaName )
     }
 }
 
+void SwHTMLParser::SetupFilterOptions()
+{
+    if (!GetMedium())
+        return;
+
+    const SfxItemSet* pItemSet = GetMedium()->GetItemSet();
+    if (!pItemSet)
+        return;
+
+    auto pItem = pItemSet->GetItem<SfxStringItem>(SID_FILE_FILTEROPTIONS);
+    if (!pItem)
+        return;
+
+    OUString aFilterOptions = pItem->GetValue();
+    const OUString aXhtmlNsKey("xhtmlns=");
+    if (aFilterOptions.startsWith(aXhtmlNsKey))
+        SetNamespace(aFilterOptions.copy(aXhtmlNsKey.getLength()));
+}
+
 namespace
 {
     class FontCacheGuard

sw/source/filter/html/swhtml.hxx

@@ -889,6 +889,9 @@ private:
 
     bool PendingObjectsInPaM(SwPaM& rPam) const;
 
+    /// Parse FilterOptions passed to the importer.
+    void SetupFilterOptions();
+
     class TableDepthGuard
     {
     private: