Each XML part of the generated documents has a BOM prefix

[elkinmilen]

Description

Cannot generate document without BOM sequence prefix on every part of the document.
Some viewers throw an error and do not open the document because of this. When first opened and saved with MS Excel the BOM sequences for each part gets removed.

Information

• .NET Target: any
• DocumentFormat.OpenXml Version: (ie 2.7.2)

Repro

Generate basic Excel document. Change the extension of the generated document to ZIP. Open the ZIP archive. Open an XML file inside into a HEX editor of choice.

Observed

The file contents start with EF BB BF (or BOM) and you cannot avoid that.

Expected

No BOM sequence or a setting in the API to omit it.

[twsouthwick]

Thanks for letting us know! In the future, for a repro, providing code would make it much easier to diagnose the problem. I'll take a look and see if this is sufficient.

[twsouthwick]

I've reproed it:

[Fact]
public void PartsShouldBeEncodedWithUTF8WithoutBOM()
{
    using (var ms = new MemoryStream())
    {
        using (var package = SpreadsheetDocument.Create(ms, DocumentFormat.OpenXml.SpreadsheetDocumentType.Workbook))
        {
            var wb = package.AddWorkbookPart();

            wb.Workbook = new Workbook();
        }

        using (var package = Package.Open(ms))
        {
            var part = package.GetPart(new Uri("/xl/workbook.xml", UriKind.Relative));

            using (var stream = part.GetStream())
            using (var reader = new StreamReader(stream))
            {
                reader.Read();

                var encoding = Assert.IsType<UTF8Encoding>(reader.CurrentEncoding);
                Assert.Empty(encoding.GetPreamble());
            }
        }
    }
}

@tomjebo Is there anything in the spec that specifies that parts must be written in a specific encoding? Changing to output UTF8 w/o the BOM is easy enough, but I want to make sure this is ok in terms of the spec.

We'll probably want to find all places where the an XmlWriter is created and ensure we update all those. We may want to look at consolidating the XmlWriter instances if possible.