tdf#155496: Export list header to (X)HTML using 'display: block' style

This re-implements commit 013a4f1f5c9ea5fb511568c53a7e76d1b365a65d
(sw XHTML export: fix handling of list labels, 2021-05-13), and
instead of not putting headers (ODF 'text:list-header' elements)
into lists, this adds 'style="display: block"' attribute to the
respective list items. This makes sure that the items use proper
list indentation, and produces correct markup.

Change-Id: I900e4aebbe562830dc2ce5400e3e33b38c2f2faa
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/152280
Tested-by: Jenkins
Reviewed-by: Miklos Vajna <vmiklos@collabora.com>
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/152298

3 files changed, 17 insertions, 92 deletions

diff --git a/sw/source/filter/html/htmlatr.cxx b/sw/source/filter/html/htmlatr.cxx
index 87b38fc0165f..79c00be34fad 100644
--- a/sw/source/filter/html/htmlatr.cxx
+++ b/sw/source/filter/html/htmlatr.cxx
@@ -683,12 +683,11 @@ static void OutHTML_SwFormat( Writer& rWrt, const SwFormat& rFormat,
     if( nNewDefListLvl != rHWrt.m_nDefListLvl )
         rHWrt.OutAndSetDefList( nNewDefListLvl );
 
-    bool bAtLeastOneNumbered = false;
     // if necessary, start a bulleted or numbered list
     if( rInfo.bInNumberBulletList )
     {
         OSL_ENSURE( !rHWrt.m_nDefListLvl, "DL cannot be inside OL!" );
-        OutHTML_NumberBulletListStart( rHWrt, aNumInfo, bAtLeastOneNumbered );
+        OutHTML_NumberBulletListStart( rHWrt, aNumInfo );
 
         if( bNumbered )
         {
@@ -758,18 +757,26 @@ static void OutHTML_SwFormat( Writer& rWrt, const SwFormat& rFormat,
 
     // if necessary, start a new list item
     bool bNumberedForListItem = bNumbered;
-    if (!bNumberedForListItem && rHWrt.mbXHTML && bAtLeastOneNumbered)
+    if (!bNumberedForListItem)
     {
-        // OutHTML_NumberBulletListEnd() will end a list item if at least one text node is numbered
-        // in the list, so open the list item with the same condition here.
-        bNumberedForListItem = true;
+        // Open a list also for the leading unnumbered nodes (= list headers in ODF terminology);
+        // to do that, detect if this unnumbered node is the first in this list
+        const auto& rPrevListInfo = rHWrt.GetNumInfo();
+        if (rPrevListInfo.GetNumRule() != aNumInfo.GetNumRule() || aNumInfo.IsRestart(rPrevListInfo)
+            || rPrevListInfo.GetDepth() < aNumInfo.GetDepth())
+            bNumberedForListItem = true;
     }
     if( rInfo.bInNumberBulletList && bNumberedForListItem )
     {
         HtmlWriter html(rWrt.Strm(), rHWrt.maNamespace);
         html.prettyPrint(rHWrt.m_bPrettyPrint);
         html.start(OOO_STRING_SVTOOLS_HTML_li);
-        if( USHRT_MAX != nNumStart )
+        if (!bNumbered)
+        {
+            // Handles list headers (<text:list-header> ODF element)
+            html.attribute(OOO_STRING_SVTOOLS_HTML_O_style, "display: block");
+        }
+        else if (USHRT_MAX != nNumStart)
             html.attribute(OOO_STRING_SVTOOLS_HTML_O_value, OString::number(nNumStart));
         // Finish the opening element, but don't close it.
         html.characters("");
diff --git a/sw/source/filter/html/htmlnumwriter.cxx b/sw/source/filter/html/htmlnumwriter.cxx
index 365864810551..a3bc30891f76 100644
--- a/sw/source/filter/html/htmlnumwriter.cxx
+++ b/sw/source/filter/html/htmlnumwriter.cxx
@@ -84,8 +84,7 @@ void SwHTMLWriter::SetNextNumInfo( std::unique_ptr<SwHTMLNumRuleInfo> pNxt )
 }
 
 Writer& OutHTML_NumberBulletListStart( SwHTMLWriter& rWrt,
-                                 const SwHTMLNumRuleInfo& rInfo,
-                                 bool& rAtLeastOneNumbered )
+                                 const SwHTMLNumRuleInfo& rInfo )
 {
     SwHTMLNumRuleInfo& rPrevInfo = rWrt.GetNumInfo();
     bool bSameRule = rPrevInfo.GetNumRule() == rInfo.GetNumRule();
@@ -95,43 +94,6 @@ 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;
-        SwNodeOffset nPos = rWrt.m_pCurrentPam->GetPoint()->GetNodeIndex() + 1;
-        SwNumRule* pNumRule = nullptr;
-        while (true)
-        {
-            const SwNode* pNode = rWrt.m_pDoc->GetNodes()[nPos];
-            if (!pNode->IsTextNode())
-            {
-                break;
-            }
-
-            const SwTextNode* pTextNode = pNode->GetTextNode();
-            if (!pTextNode->GetNumRule() || (pNumRule && pTextNode->GetNumRule() != pNumRule))
-            {
-                // Node is not in the same numbering as the previous one.
-                break;
-            }
-
-            pNumRule = pTextNode->GetNumRule();
-            if (pTextNode->IsNumbered())
-            {
-                bAtLeastOneNumbered = true;
-                break;
-            }
-            ++nPos;
-        }
-
-        rAtLeastOneNumbered = bAtLeastOneNumbered;
-        if (!bAtLeastOneNumbered)
-        {
-            return rWrt;
-        }
-    }
-
     bool bStartValue = false;
     if( !bSameRule && rInfo.GetDepth() )
     {
@@ -325,44 +287,10 @@ Writer& OutHTML_NumberBulletListEnd( SwHTMLWriter& rWrt,
     bool bListEnd = !bSameRule || rNextInfo.GetDepth() < rInfo.GetDepth() || rNextInfo.IsRestart(rInfo);
     bool bNextIsSubitem = !bListEnd && rNextInfo.GetDepth() > rInfo.GetDepth();
 
-    std::optional<bool> oAtLeastOneNumbered;
-    if (!rInfo.IsNumbered())
-    {
-        oAtLeastOneNumbered = false;
-        SwNodeOffset nPos = rWrt.m_pCurrentPam->GetPoint()->GetNodeIndex() - 1;
-        SwNumRule* pNumRule = nullptr;
-        while (true)
-        {
-            const SwNode* pNode = rWrt.m_pDoc->GetNodes()[nPos];
-            if (!pNode->IsTextNode())
-            {
-                break;
-            }
-
-            const SwTextNode* pTextNode = pNode->GetTextNode();
-            if (!pTextNode->GetNumRule() || (pNumRule && pTextNode->GetNumRule() != pNumRule))
-            {
-                // Node is not in the same numbering as the next one.
-                break;
-            }
-
-            pNumRule = pTextNode->GetNumRule();
-            if (pTextNode->IsNumbered())
-            {
-                oAtLeastOneNumbered = true;
-                break;
-            }
-            --nPos;
-        }
-    }
-
     if (rWrt.mbXHTML)
     {
-        // The list is numbered if the previous text node is numbered or any other previous text
-        // node is numbered.
-        bool bPrevIsNumbered = rInfo.IsNumbered() || *oAtLeastOneNumbered;
         // XHTML </li> for the list item content, if there is an open <li>.
-        if ((bListEnd && bPrevIsNumbered) || (!bListEnd && !bNextIsSubitem && rNextInfo.IsNumbered()))
+        if (bListEnd || (!bNextIsSubitem && rNextInfo.IsNumbered()))
         {
             HTMLOutFuncs::Out_AsciiTag(
                 rWrt.Strm(), Concat2View(rWrt.GetNamespace() + OOO_STRING_SVTOOLS_HTML_li),
@@ -375,15 +303,6 @@ 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.
-        if (!*oAtLeastOneNumbered)
-        {
-            return rWrt;
-        }
-    }
-
     OSL_ENSURE( rWrt.m_nLastParaToken == HtmlTokenId::NONE,
                 "<PRE> was not closed before </OL>." );
     sal_uInt16 nNextDepth =
diff --git a/sw/source/filter/html/wrthtml.hxx b/sw/source/filter/html/wrthtml.hxx
index a1986daf5dba..aa87bf245592 100644
--- a/sw/source/filter/html/wrthtml.hxx
+++ b/sw/source/filter/html/wrthtml.hxx
@@ -724,8 +724,7 @@ Writer& OutCSS1_NumberBulletListStyleOpt( Writer& rWrt, const SwNumRule& rNumRul
                                     sal_uInt8 nLevel );
 
 Writer& OutHTML_NumberBulletListStart( SwHTMLWriter& rWrt,
-                                 const SwHTMLNumRuleInfo& rInfo,
-                                 bool& rAtLeastOneNumbered );
+                                 const SwHTMLNumRuleInfo& rInfo );
 Writer& OutHTML_NumberBulletListEnd( SwHTMLWriter& rWrt,
                                const SwHTMLNumRuleInfo& rNextInfo );