tdf#127484 - Use HTML caption tag to name externally linked tables

Change-Id: I0abca8e1966e93337c16609211c146c0ed97c62f
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/114743
Tested-by: Jenkins
Reviewed-by: Eike Rathke <erack@redhat.com>

3 files changed, 41 insertions, 8 deletions

diff --git a/sc/source/filter/html/htmlimp.cxx b/sc/source/filter/html/htmlimp.cxx
index 4a9a3a93b099..f1004ba0fe73 100644
--- a/sc/source/filter/html/htmlimp.cxx
+++ b/sc/source/filter/html/htmlimp.cxx
@@ -181,14 +181,17 @@ void ScHTMLImport::WriteToDocument(
             assert(!"can't move");
         }
         // insert table number as name
-        InsertRangeName( *mpDoc, ScfTools::GetNameFromHTMLIndex( nTableId ), aNewRange );
+        OUStringBuffer aName(ScfTools::GetNameFromHTMLIndex(nTableId));
         // insert table id as name
         if (!pTable->GetTableName().isEmpty())
-        {
-            OUString aName( ScfTools::GetNameFromHTMLName( pTable->GetTableName() ) );
-            if (!mpDoc->GetRangeName()->findByUpperName(ScGlobal::getCharClassPtr()->uppercase(aName)))
-                InsertRangeName( *mpDoc, aName, aNewRange );
-        }
+            aName.append(" - " + pTable->GetTableName());
+        // insert table caption as name
+        if (!pTable->GetTableCaption().isEmpty())
+            aName.append(" - " + pTable->GetTableCaption());
+        if (!mpDoc->GetRangeName()->findByUpperName(
+                ScGlobal::getCharClassPtr()->uppercase(aName.toString())))
+            InsertRangeName(*mpDoc, aName.toString(), aNewRange);
+
     }
 }
 
diff --git a/sc/source/filter/html/htmlpars.cxx b/sc/source/filter/html/htmlpars.cxx
index 394691b70cbe..e320269352ed 100644
--- a/sc/source/filter/html/htmlpars.cxx
+++ b/sc/source/filter/html/htmlpars.cxx
@@ -1804,7 +1804,8 @@ ScHTMLTable::ScHTMLTable( ScHTMLTable& rParentTable, const HtmlImportInfo& rInfo
     mbPreFormText( bPreFormText ),
     mbRowOn( false ),
     mbDataOn( false ),
-    mbPushEmptyLine( false )
+    mbPushEmptyLine( false ),
+    mbCaptionOn ( false )
 {
     if( mbPreFormText )
     {
@@ -1851,7 +1852,8 @@ ScHTMLTable::ScHTMLTable(
     mbPreFormText( false ),
     mbRowOn( false ),
     mbDataOn( false ),
-    mbPushEmptyLine( false )
+    mbPushEmptyLine( false ),
+    mbCaptionOn ( false )
 {
     // open the first "cell" of the document
     ImplRowOn();
@@ -1901,6 +1903,9 @@ void ScHTMLTable::PutText( const HtmlImportInfo& rInfo )
             mxCurrEntry->AdjustStart( rInfo );
         else
             mxCurrEntry->AdjustEnd( rInfo );
+        if (mbCaptionOn)
+            maCaptionBuffer.append(rInfo.aText);
+
     }
 }
 
@@ -1950,6 +1955,20 @@ ScHTMLTable* ScHTMLTable::TableOff( const HtmlImportInfo& rInfo )
     return mbPreFormText ? this : CloseTable( rInfo );
 }
 
+void ScHTMLTable::CaptionOn()
+{
+    mbCaptionOn = true;
+    maCaptionBuffer.setLength(0);
+}
+
+void ScHTMLTable::CaptionOff()
+{
+    if (!mbCaptionOn)
+        return;
+    maCaption = maCaptionBuffer.makeStringAndClear().trim();
+    mbCaptionOn = false;
+}
+
 ScHTMLTable* ScHTMLTable::PreOn( const HtmlImportInfo& rInfo )
 {
     PushEntry( rInfo );
@@ -2823,6 +2842,8 @@ void ScHTMLQueryParser::ProcessToken( const HtmlImportInfo& rInfo )
 // --- table handling ---
         case HtmlTokenId::TABLE_ON:         TableOn( rInfo );               break;  // <table>
         case HtmlTokenId::TABLE_OFF:        TableOff( rInfo );              break;  // </table>
+        case HtmlTokenId::CAPTION_ON:       mpCurrTable->CaptionOn();       break;  // <caption>
+        case HtmlTokenId::CAPTION_OFF:      mpCurrTable->CaptionOff();      break;  // </caption>
         case HtmlTokenId::TABLEROW_ON:      mpCurrTable->RowOn( rInfo );    break;  // <tr>
         case HtmlTokenId::TABLEROW_OFF:     mpCurrTable->RowOff( rInfo );   break;  // </tr>
         case HtmlTokenId::TABLEHEADER_ON:                                           // <th>
diff --git a/sc/source/filter/inc/htmlpars.hxx b/sc/source/filter/inc/htmlpars.hxx
index 1bf8c9b91008..a683a0c984d7 100644
--- a/sc/source/filter/inc/htmlpars.hxx
+++ b/sc/source/filter/inc/htmlpars.hxx
@@ -338,6 +338,8 @@ public:
 
     /** Returns the name of the table, specified in the TABLE tag. */
     const OUString& GetTableName() const { return maTableName; }
+    /** Returns the caption of the table, specified in the <caption> tag. */
+    const OUString& GetTableCaption() const { return maCaption; }
     /** Returns the unique identifier of the table. */
     ScHTMLTableId GetTableId() const { return maTableId.mnTableId; }
     /** Returns the cell spanning of the specified cell. */
@@ -368,6 +370,10 @@ public:
     /** Closes *this* table (</table> tag).
         @return  Pointer to the parent table. */
     ScHTMLTable*        TableOff( const HtmlImportInfo& rInfo );
+    /** Processes the caption of the table (<caption> tag). */
+    void                CaptionOn();
+    /** Processes the caption of the table (</caption> tag). */
+    void                CaptionOff();
     /** Starts a *new* table based on preformatted text (<pre> tag).
         @return  Pointer to the new table. */
     ScHTMLTable*        PreOn( const HtmlImportInfo& rInfo );
@@ -517,6 +523,8 @@ private:
     ScHTMLTable*        mpParentTable;      /// Pointer to parent table.
     ScHTMLTableMapPtr   mxNestedTables;     /// Table of nested HTML tables.
     OUString       maTableName;        /// Table name from <table id> option.
+    OUString            maCaption;          /// Caption name of the table from <caption> </caption>
+    OUStringBuffer      maCaptionBuffer;    /// Caption buffer of the table from <caption> </caption>
     ScHTMLTableAutoId   maTableId;          /// Unique identifier of this table.
     SfxItemSet          maTableItemSet;     /// Items for the entire table.
     SfxItemSetPtr       mxRowItemSet;       /// Items for the current table row.
@@ -539,6 +547,7 @@ private:
     bool                mbRowOn:1;          /// true = Inside of <tr> </tr>.
     bool                mbDataOn:1;         /// true = Inside of <td> </td> or <th> </th>.
     bool                mbPushEmptyLine:1;  /// true = Insert empty line before current entry.
+    bool                mbCaptionOn:1;      /// true = Inside of <caption> </caption>
 };
 
 /** The "global table" representing the entire HTML document. */