sw XHTML export: fix handling of list labels

This is building on top of commit
119b6876c92e4cdae44583c4b1b1419d3533e3ee (sw XHTML export: properly
write <li>...</li> around multiple paragraphs, 2020-05-21), but the
use-case here is a numbering with list labels only.

The first problem was that the list label had its <li> suppressed, but
not its </li>.

The other problem is that <ul> can only have <li> child elements, so at
least fix the case where the list only has list labels, in which case
even <ul> and </ul> should be omitted.

(cherry picked from commit 013a4f1f5c9ea5fb511568c53a7e76d1b365a65d)

Change-Id: Id38978a40d8618f483e3e9d499f75838d5b2adb0

2 files changed, 98 insertions, 1 deletions

diff --git a/sw/qa/extras/htmlexport/htmlexport.cxx b/sw/qa/extras/htmlexport/htmlexport.cxx
index 3746c3e7e786..b19ab0d655f2 100644
--- a/sw/qa/extras/htmlexport/htmlexport.cxx
+++ b/sw/qa/extras/htmlexport/htmlexport.cxx
@@ -1389,6 +1389,37 @@ CPPUNIT_TEST_FIXTURE(SwHtmlDomExportTest, testReqifObjdataPresentationDataSize)
     CPPUNIT_ASSERT_EQUAL(static_cast<sal_uInt32>(565994), aOle1Reader.m_nPresentationDataSize);
 }
 
+CPPUNIT_TEST_FIXTURE(SwHtmlDomExportTest, testListHeading)
+{
+    // Given a document with a list heading:
+    loadURL("private:factory/swriter", nullptr);
+    SwXTextDocument* pTextDoc = dynamic_cast<SwXTextDocument*>(mxComponent.get());
+    SwWrtShell* pWrtShell = pTextDoc->GetDocShell()->GetWrtShell();
+    pWrtShell->Insert("list header");
+    SwDoc* pDoc = pWrtShell->GetDoc();
+    sal_uInt16 nPos = pDoc->MakeNumRule(pDoc->GetUniqueNumRuleName());
+    SwNumRule* pNumRule = pDoc->GetNumRuleTable()[nPos];
+    SwNode& rNode = pWrtShell->GetCursor()->GetPoint()->nNode.GetNode();
+    SwTextNode& rTextNode = *rNode.GetTextNode();
+    rTextNode.SetAttr(SwNumRuleItem(pNumRule->GetName()));
+    rTextNode.SetCountedInList(false);
+
+    // When exporting to ReqIF:
+    ExportToReqif();
+
+    // Then make sure the output is valid xhtml:
+    SvMemoryStream aStream;
+    HtmlExportTest::wrapFragment(maTempFile, aStream);
+    xmlDocUniquePtr pXmlDoc = parseXmlStream(&aStream);
+    CPPUNIT_ASSERT(pDoc);
+
+    // Without the accompanying fix in place, this test would have failed:
+    // - expected: <div><p>...</p></div>
+    // - actual  : <div><ol><p>...</p></li></ol></div>
+    // because a </li> but no <li> is not well-formed and <ol> with a non-li children is invalid.
+    assertXPathContent(pXmlDoc, "/reqif-xhtml:html/reqif-xhtml:div/reqif-xhtml:p", "list header");
+}
+
 CPPUNIT_PLUGIN_IMPLEMENT();
 
 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */
diff --git a/sw/source/filter/html/htmlnumwriter.cxx b/sw/source/filter/html/htmlnumwriter.cxx
index ea3c949c3801..84ed431699d8 100644
--- a/sw/source/filter/html/htmlnumwriter.cxx
+++ b/sw/source/filter/html/htmlnumwriter.cxx
@@ -94,6 +94,39 @@ Writer& OutHTML_NumberBulletListStart( SwHTMLWriter& rWrt,
         return rWrt;
     }
 
+    if (rWrt.mbXHTML && !rInfo.IsNumbered())
+    {
+        // If the list only consists of non-numbered text nodes, then don't start the list.
+        bool bAtLeastOneNumbered = false;
+        sal_uLong nPos = rWrt.m_pCurrentPam->GetPoint()->nNode.GetIndex() + 1;
+        while (true)
+        {
+            const SwNode* pNode = rWrt.m_pDoc->GetNodes()[nPos];
+            if (!pNode->IsTextNode())
+            {
+                break;
+            }
+
+            const SwTextNode* pTextNode = pNode->GetTextNode();
+            if (!pTextNode->GetNumRule())
+            {
+                break;
+            }
+
+            if (pTextNode->IsNumbered())
+            {
+                bAtLeastOneNumbered = true;
+                break;
+            }
+            ++nPos;
+        }
+
+        if (!bAtLeastOneNumbered)
+        {
+            return rWrt;
+        }
+    }
+
     bool bStartValue = false;
     if( !bSameRule && rInfo.GetDepth() )
     {
@@ -283,7 +316,7 @@ Writer& OutHTML_NumberBulletListEnd( SwHTMLWriter& rWrt,
 
     if (rWrt.mbXHTML)
     {
-        if (bListEnd || (!bListEnd && rNextInfo.IsNumbered()))
+        if ((bListEnd && rInfo.IsNumbered()) || (!bListEnd && rNextInfo.IsNumbered()))
         {
             HTMLOutFuncs::Out_AsciiTag(rWrt.Strm(),
                                        rWrt.GetNamespace() + OOO_STRING_SVTOOLS_HTML_li, false);
@@ -295,6 +328,39 @@ Writer& OutHTML_NumberBulletListEnd( SwHTMLWriter& rWrt,
         return rWrt;
     }
 
+    if (rWrt.mbXHTML && !rInfo.IsNumbered())
+    {
+        // If the list only consisted of non-numbered text nodes, then don't end the list.
+        bool bAtLeastOneNumbered = false;
+        sal_uLong nPos = rWrt.m_pCurrentPam->GetPoint()->nNode.GetIndex() - 1;
+        while (true)
+        {
+            const SwNode* pNode = rWrt.m_pDoc->GetNodes()[nPos];
+            if (!pNode->IsTextNode())
+            {
+                break;
+            }
+
+            const SwTextNode* pTextNode = pNode->GetTextNode();
+            if (!pTextNode->GetNumRule())
+            {
+                break;
+            }
+
+            if (pTextNode->IsNumbered())
+            {
+                bAtLeastOneNumbered = true;
+                break;
+            }
+            --nPos;
+        }
+
+        if (!bAtLeastOneNumbered)
+        {
+            return rWrt;
+        }
+    }
+
     OSL_ENSURE( rWrt.m_nLastParaToken == HtmlTokenId::NONE,
                 "<PRE> was not closed before </OL>." );
     sal_uInt16 nNextDepth =