tdf#159483 sc HTML paste: handle data-sheets-formula attribute

When a formula cell gets copied from google docs, the value of the <td>
element only contains the formula result. This means once value cells
and formula cells gets copied together, the pasted formula cell won't
updated anymore when the input values change.

Turns out there is a data-sheets-formula attribute on <td> that contains
the formula, it seems it uses the R1C1 format.

Fix the problem by extending ScHTMLLayoutParser::TableDataOn() to parse
this attribute and set a formula on the target cell (rather than a
value) if the formula is available.

This required also extending ScEEImport a bit, since the HTML paste
builds on top of the RTF one in Calc.

Change-Id: I720df96ce74a5e865b7329d06f3b719551f31b96
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/163234
Reviewed-by: Miklos Vajna <vmiklos@collabora.com>
Tested-by: Jenkins

include/svtools/htmlkywd.hxx

@@ -448,6 +448,7 @@
 #define OOO_STRING_SVTOOLS_HTML_O_DSval "data-sheets-value"
 #define OOO_STRING_SVTOOLS_HTML_O_SDnum "sdnum"
 #define OOO_STRING_SVTOOLS_HTML_O_DSnum "data-sheets-numberformat"
+#define OOO_STRING_SVTOOLS_HTML_O_DSformula "data-sheets-formula"
 #define OOO_STRING_SVTOOLS_HTML_O_sdlibrary "sdlibrary"
 #define OOO_STRING_SVTOOLS_HTML_O_sdmodule "sdmodule"
 #define OOO_STRING_SVTOOLS_HTML_O_sdevent "sdevent-"

include/svtools/htmltokn.h

@@ -347,6 +347,7 @@ STRING_START        = BOOL_END,
     DSVAL,
     SDNUM, // StarDiv NumberFormat
     DSNUM,
+    DSFORMULA,
     SDLIBRARY,
     SDMODULE,
 STRING_END,

sc/qa/filter/html/data/formula.html

@@ -0,0 +1,7 @@
+<table>
+  <tr>
+    <td data-sheets-value="{&quot;1&quot;:3,&quot;3&quot;:1}">1</td>
+    <td data-sheets-value="{&quot;1&quot;:3,&quot;3&quot;:2}">2</td>
+    <td data-sheets-value="{&quot;1&quot;:3,&quot;3&quot;:3}" data-sheets-formula="=SUM(R[0]C[-2]:R[0]C[-1])">3</td>
+  </tr>
+</table>

sc/qa/filter/html/html.cxx

@@ -143,6 +143,31 @@ CPPUNIT_TEST_FIXTURE(Test, testPasteTdAsFormattedNumber)
     CPPUNIT_ASSERT_EQUAL(static_cast<double>(1000),
                          pDoc->GetValue(/*col=*/0, /*row=*/0, /*tab=*/0));
 }
+
+CPPUNIT_TEST_FIXTURE(Test, testPasteTdAsFormula)
+{
+    // Given an empty document:
+    createScDoc();
+
+    // When pasting HTML with cells containing a formula:
+    ScDocument* pDoc = getScDoc();
+    ScAddress aCellPos(/*nColP=*/0, /*nRowP=*/0, /*nTabP=*/0);
+    ScImportExport aImporter(*pDoc, aCellPos);
+    SvFileStream aFile(createFileURL(u"formula.html"), StreamMode::READ);
+    SvMemoryStream aMemory;
+    aMemory.WriteStream(aFile);
+    aMemory.Seek(0);
+    CPPUNIT_ASSERT(aImporter.ImportStream(aMemory, OUString(), SotClipboardFormatId::HTML));
+
+    // Then make sure C1 is a sum and it evaluates to 3:
+    // Without the accompanying fix in place, this test would have failed with:
+    // - Expected: =SUM(A1:B1)
+    // - Actual  :
+    // i.e. only the formula result was imported, not the formula.
+    CPPUNIT_ASSERT_EQUAL(OUString("=SUM(A1:B1)"),
+                         pDoc->GetFormula(/*col=*/2, /*row=*/0, /*tab=*/0));
+    CPPUNIT_ASSERT_EQUAL(static_cast<double>(3), pDoc->GetValue(/*col=*/2, /*row=*/0, /*tab=*/0));
+}
 }
 
 CPPUNIT_PLUGIN_IMPLEMENT();

sc/source/filter/html/htmlpars.cxx

@@ -155,6 +155,14 @@ void ParseDataSheetsNumberformat(const OUString& rDataSheetsValue, std::optional
         }
     }
 }
+
+/// data-sheets-formula from google sheets, grammar is R1C1 reference style.
+void ParseDataSheetsFormula(const OUString& rDataSheetsFormula, std::optional<OUString>& rVal,
+                            std::optional<formula::FormulaGrammar::Grammar>& rGrammar)
+{
+    rVal = rDataSheetsFormula;
+    rGrammar = formula::FormulaGrammar::GRAM_ENGLISH_XL_R1C1;
+}
 }
 
 ScHTMLStyles::ScHTMLStyles() : maEmpty() {}
@@ -1074,6 +1082,12 @@ void ScHTMLLayoutParser::TableDataOn( HtmlImportInfo* pInfo )
                 ParseDataSheetsNumberformat(rOption.GetString(), mxActEntry->pNumStr);
             }
             break;
+            case HtmlOptionId::DSFORMULA:
+            {
+                ParseDataSheetsFormula(rOption.GetString(), mxActEntry->moFormulaStr,
+                                       mxActEntry->moFormulaGrammar);
+            }
+            break;
             default: break;
         }
     }

sc/source/filter/inc/eeparser.hxx

@@ -56,6 +56,8 @@ struct ScEEParseEntry
     ESelection          aSel;           // Selection in EditEngine
     std::optional<OUString>
                         pValStr;        // HTML possibly SDVAL string
+    std::optional<OUString> moFormulaStr;
+    std::optional<formula::FormulaGrammar::Grammar> moFormulaGrammar;
     std::optional<OUString>
                         pNumStr;        // HTML possibly SDNUM string
     std::optional<OUString>

sc/source/filter/rtf/eeimpars.cxx

@@ -340,6 +340,11 @@ void ScEEImport::WriteToDocument( bool bSizeColsRows, double nOutputFactor, SvNu
 
                 if (!aValStr.isEmpty())
                     mpDoc->SetValue( nCol, nRow, nTab, fVal );
+                else if (pE->moFormulaStr && pE->moFormulaGrammar)
+                {
+                    mpDoc->SetFormula(ScAddress(nCol, nRow, nTab), *pE->moFormulaStr,
+                                      *pE->moFormulaGrammar);
+                }
                 else if ( !pE->aSel.HasRange() )
                 {
                     // maybe ALT text of IMG or similar

svtools/source/svhtml/htmlkywd.cxx

@@ -527,6 +527,7 @@ static HTML_OptionEntry aHTMLOptionTab[] = {
     {std::u16string_view(u"" OOO_STRING_SVTOOLS_HTML_O_DSval),     HtmlOptionId::DSVAL},
     {std::u16string_view(u"" OOO_STRING_SVTOOLS_HTML_O_SDnum),     HtmlOptionId::SDNUM}, // StarDiv NumberFormat
     {std::u16string_view(u"" OOO_STRING_SVTOOLS_HTML_O_DSnum),     HtmlOptionId::DSNUM},
+    {std::u16string_view(u"" OOO_STRING_SVTOOLS_HTML_O_DSformula), HtmlOptionId::DSFORMULA},
     {std::u16string_view(u"" OOO_STRING_SVTOOLS_HTML_O_sdlibrary), HtmlOptionId::SDLIBRARY},
     {std::u16string_view(u"" OOO_STRING_SVTOOLS_HTML_O_sdmodule),  HtmlOptionId::SDMODULE},