// Sometimes encounter cells that have no r attribute, so infer it if possible.
// These are invalid spreadsheets, but attempt to get the data anyway.
private static void FixUpCellsThatHaveNoRAtt(XDocument shXDoc)
{
// if there are any rows that have all cells with no r attribute, then fix them up
var invalidRows = shXDoc
.Descendants(S.row)
.Where(r => !r.Elements(S.c).Any(c => c.Attribute("r") != null))
.ToList();
foreach (var row in invalidRows)
{
var rowNumberStr = (string)row.Attribute("r");
var colNumber = 0;
foreach (var cell in row.Elements(S.c))
{
var newCellRef = XlsxTables.IndexToColumnAddress(colNumber) + rowNumberStr;
cell.Add(new XAttribute("r", newCellRef));
}
}
// repeat iteratively until no further fixes can be made
while (true)
{
var invalidCells = shXDoc
.Descendants(S.c)
.Where(c => c.Attribute("r") == null)
.ToList();
bool didFixup = false;
foreach (var cell in invalidCells)
{
var followingCell = cell.ElementsAfterSelf(S.c).FirstOrDefault();
if (followingCell != null)
{
var followingR = (string)followingCell.Attribute("r");
if (followingR != null)
{
var spl = XlsxTables.SplitAddress(followingR);
var colIdxFollowing = XlsxTables.ColumnAddressToIndex(spl[0]);
var newRef = XlsxTables.IndexToColumnAddress(colIdxFollowing - 1) + spl[1];
cell.Add(new XAttribute("r", newRef));
didFixup = true;
}
else
{
didFixup = FixUpBasedOnPrecedingCell(didFixup, cell);
}
}
else
{
didFixup = FixUpBasedOnPrecedingCell(didFixup, cell);
}
}
if (!didFixup)
{
break;
}
}
}
private static bool FixUpBasedOnPrecedingCell(bool didFixup, XElement cell)
{
XElement precedingCell = GetPrevousElement(cell);
if (precedingCell != null)
{
var precedingR = (string)precedingCell.Attribute("r");
if (precedingR != null)
{
var spl = XlsxTables.SplitAddress(precedingR);
var colIdxFollowing = XlsxTables.ColumnAddressToIndex(spl[0]);
var newRef = XlsxTables.IndexToColumnAddress(colIdxFollowing + 1) + spl[1];
cell.Add(new XAttribute("r", newRef));
didFixup = true;
}
}
return(didFixup);
}
private static XElement GetCellProps_InTable(SpreadsheetDocument sDoc, XDocument styleXDoc, Table table, Cell tc)
{
var style = tc.Style;
if (style == null)
{
return(null);
}
var colIdStr = tc.ColumnAddress;
int colNbr = XlsxTables.ColumnAddressToIndex(colIdStr);
TableColumn column = table.TableColumns().FirstOrDefault(z => z.ColumnNumber == colNbr);
if (column == null)
{
throw new FileFormatException("Invalid spreadsheet");
}
var cellProps = new XElement("CellProps");
var d = column.DataDxfId;
if (d != null)
{
var dataDxf = styleXDoc
.Root
.Elements(S.dxfs)
.Elements(S.dxf)
.Skip((int)d)
.FirstOrDefault();
if (dataDxf == null)
{
throw new FileFormatException("Invalid spreadsheet");
}
var numFmt = dataDxf.Element(S.numFmt);
if (numFmt != null)
{
var numFmtId = (int?)numFmt.Attribute("numFmtId");
if (numFmtId != null)
{
cellProps.Add(new XAttribute("numFmtId", numFmtId));
}
var formatCode = (string)numFmt.Attribute("formatCode");
if (formatCode != null)
{
cellProps.Add(new XAttribute("formatCode", formatCode));
}
}
}
var xf = styleXDoc.Root
.Elements(S.cellXfs)
.Elements(S.xf)
.Skip((int)style)
.FirstOrDefault();
if (xf == null)
{
throw new FileFormatException("Invalid spreadsheet");
}
// if xf has different numFmtId, then replace the ones from the table definition
var numFmtId2 = (int?)xf.Attribute("numFmtId");
if (numFmtId2 != null)
{
AddNumFmtIdAndFormatCode(styleXDoc, cellProps, numFmtId2);
}
var masterXfId = (int?)xf.Attribute("xfId");
if (masterXfId != null)
{
var masterXf = styleXDoc
.Root
.Elements(S.cellStyleXfs)
.Elements(S.xf)
.Skip((int)masterXfId)
.FirstOrDefault();
if (masterXf != null)
{
AddFormattingToCellProps(styleXDoc, cellProps, masterXf);
}
}
AddFormattingToCellProps(styleXDoc, cellProps, xf);
AugmentAndCleanUpProps(cellProps);
if (!cellProps.HasElements && !cellProps.HasAttributes)
{
return(null);
}
return(cellProps);
}
public static XElement RetrieveRange(SpreadsheetDocument sDoc, string sheetName, int leftColumn, int topRow, int rightColumn, int bottomRow)
{
var wbXdoc = sDoc.WorkbookPart.GetXDocument();
var sheet = wbXdoc
.Root
.Elements(S.sheets)
.Elements(S.sheet)
.FirstOrDefault(s => (string)s.Attribute("name") == sheetName);
if (sheet == null)
{
throw new ArgumentException("Invalid sheet name passed to RetrieveRange", "sheetName");
}
var rId = (string)sheet.Attribute(R.id);
if (rId == null)
{
throw new FileFormatException("Invalid spreadsheet");
}
var sheetPart = sDoc.WorkbookPart.GetPartById(rId);
if (sheetPart == null)
{
throw new FileFormatException("Invalid spreadsheet");
}
var shXDoc = sheetPart.GetXDocument();
if (sDoc.WorkbookPart.WorkbookStylesPart == null)
{
throw new FileFormatException("Invalid spreadsheet. No WorkbookStylesPart.");
}
var styleXDoc = sDoc.WorkbookPart.WorkbookStylesPart.GetXDocument();
// if there is no shared string table, sharedStringTable will be null
// it will only be used if there is a cell type == "s", in which case, referencing this
// part would indicate an invalid spreadsheet.
SharedStringTablePart sharedStringTable = sDoc.WorkbookPart.SharedStringTablePart;
FixUpCellsThatHaveNoRAtt(shXDoc);
// assemble the transform
var sheetData = shXDoc
.Root
.Elements(S.sheetData)
.Elements(S.row)
.Select(row =>
{
// filter
string ra = (string)row.Attribute("r");
if (ra == null)
{
return(null);
}
int rowNbr;
if (!int.TryParse(ra, out rowNbr))
{
return(null);
}
if (rowNbr < topRow)
{
return(null);
}
if (rowNbr > bottomRow)
{
return(null);
}
var cells = row
.Elements(S.c)
.Select(cell =>
{
var cellAddress = (string)cell.Attribute("r");
if (cellAddress == null)
{
throw new FileFormatException("Invalid spreadsheet - cell does not have r attribute.");
}
var splitCellAddress = XlsxTables.SplitAddress(cellAddress);
var columnAddress = splitCellAddress[0];
var columnIndex = XlsxTables.ColumnAddressToIndex(columnAddress);
// filter
if (columnIndex < leftColumn || columnIndex > rightColumn)
{
return(null);
}
var cellType = (string)cell.Attribute("t");
string sharedString = null;
if (cellType == "s")
{
int sharedStringIndex;
string sharedStringBeforeParsing = (string)cell.Element(S.v);
if (sharedStringBeforeParsing == null)
{
sharedStringBeforeParsing = (string)cell.Elements(S._is).Elements(S.t).FirstOrDefault();
}
if (sharedStringBeforeParsing == null)
{
throw new FileFormatException("Invalid document");
}
if (!int.TryParse(sharedStringBeforeParsing, out sharedStringIndex))
{
throw new FileFormatException("Invalid document");
}
XElement sharedStringElement = null;
if (sharedStringTable == null)
{
throw new FileFormatException("Invalid spreadsheet. Shared string, but no Shared String Part.");
}
sharedStringElement =
sharedStringTable
.GetXDocument()
.Root
.Elements(S.si)
.Skip(sharedStringIndex)
.FirstOrDefault();
if (sharedStringElement == null)
{
throw new FileFormatException("Invalid spreadsheet. Shared string reference not valid.");
}
sharedString =
sharedStringElement
.Descendants(S.t)
.StringConcatenate(e => (string)e);
}
if (sharedString != null)
{
XElement cellProps = GetCellProps_NotInTable(sDoc, styleXDoc, cell);
string value = sharedString;
string displayValue;
string color = null;
if (cellProps != null)
{
displayValue = SmlCellFormatter.FormatCell(
(string)cellProps.Attribute("formatCode"),
value,
out color);
}
else
{
displayValue = value;
}
XElement newCell1 = new XElement("Cell",
new XAttribute("Ref", (string)cell.Attribute("r")),
new XAttribute("ColumnId", columnAddress),
new XAttribute("ColumnNumber", columnIndex),
cell.Attribute("f") != null ? new XAttribute("Formula", (string)cell.Attribute("f")) : null,
cell.Attribute("s") != null ? new XAttribute("Style", (string)cell.Attribute("s")) : null,
cell.Attribute("t") != null ? new XAttribute("Type", (string)cell.Attribute("t")) : null,
cellProps,
new XElement("Value", value),
new XElement("DisplayValue", displayValue),
color != null ? new XElement("DisplayColor", color) : null
);
return(newCell1);
}
else
{
var type = (string)cell.Attribute("t");
XElement value = new XElement("Value", cell.Value);
if (type != null && type == "inlineStr")
{
type = "s";
}
XAttribute typeAttr = null;
if (type != null)
{
typeAttr = new XAttribute("Type", type);
}
XElement cellProps = GetCellProps_NotInTable(sDoc, styleXDoc, cell);
string displayValue;
string color = null;
if (cellProps != null)
{
displayValue = SmlCellFormatter.FormatCell(
(string)cellProps.Attribute("formatCode"),
cell.Value,
out color);
}
else
{
displayValue = displayValue = SmlCellFormatter.FormatCell(
(string)"General",
cell.Value,
out color);
}
XElement newCell2 = new XElement("Cell",
new XAttribute("Ref", (string)cell.Attribute("r")),
new XAttribute("ColumnId", columnAddress),
new XAttribute("ColumnNumber", columnIndex),
typeAttr,
cell.Attribute("f") != null ? new XAttribute("Formula", (string)cell.Attribute("f")) : null,
cell.Attribute("s") != null ? new XAttribute("Style", (string)cell.Attribute("s")) : null,
cellProps,
value,
new XElement("DisplayValue", displayValue),
color != null ? new XElement("DisplayColor", color) : null);
return(newCell2);
}
});
XElement dataRow = new XElement("Row",
row.Attribute("r") != null ? new XAttribute("RowNumber", (int)row.Attribute("r")) : null,
cells);
return(dataRow);
});
var dataProps = GetDataProps(shXDoc);
XElement data = new XElement("Data",
dataProps,
sheetData);
return(data);
}