Discrepancy Between OuterXml and document.xml Contents

[Varorbc]

Describe the bug

I noticed that the value of OuterXml doesn't match the content in document.xml. For example, in document.xml there's a section like <pic:nvPr></pic:nvPr>, but in OuterXml it shows up as <pic:nvPr />.

Observed behavior

mismatch

Expected behavior

match

Desktop (please complete the following information):

• OS: Windows
• .NET Target: .NET Core
• DocumentFormat.OpenXml Version: 3.2.0

[mikeebowen]

@Varorbc, does this create any issues or problems with your application?

[Varorbc]

@mikeebowen

First off, I'm really puzzled why this issue only shows up for this particular item. Secondly, I think the values in both places should be consistent, even though they mean the same thing. My test reads the raw file content, and the inconsistency is causing the tests to fail.

[mikeebowen]

I'm really puzzled why this issue only shows up for this particular item.

This only happens with the document.xml, but no other parts?

I think the values in both places should be consistent, even though they mean the same thing

I agree, but I was curious what kinds of issues this would cause.

[Varorbc]

This only happens with the document.xml, but no other parts?

I mean pic:nvPr

I agree, but I was curious what kinds of issues this would cause.

My test reads the raw file content, and the inconsistency is causing the tests to fail.

[mikeebowen]

Hi @Varorbc,

I am closing this as by design. The OuterXml is not a copy of the xml string from the parts but generated by the SDK, so it is not a character for character copy of the xml from the part. To resolve the issue with your tests you can use the XML Diff Patch Tool Using the XML Diff and Patch Tool in Your Applications, which can evaluate XML and know, for example, that <pic:nvPr /> and <pic:nvPr></pic:nvPr> are functionally equivalent. There are also other tools on Nuget if you need a package.

[Varorbc]

although I can fix this difference, I can't comprehend this design. OuterXml is generated by the SDK, and document.xml is also generated by the SDK. The inconsistency between them is really hard to understand at all.

[mikeebowen]

The representation of the part in the SDK can differ from what is in the file when it has been updated/manipulated, but before saving, so the SDK has to generate the OuterXml property, so that it matches what the current state is, even if it has not been saved to the file yet.

[Varorbc]

I've noticed that they are inconsistent even though I just opened the file and didn't do anything.

[mikeebowen]

The OuterXml property is always generated, whether you've changed anything or not. The SDK reads in an xml part and creates a representation of the part in code then the OuterXml property is generated from that representation.

[Varorbc]

So, why on earth is there this difference? What's the reason behind it? And you know, it's not like all elements have this kind of difference. It appears that only a few specific elements have it.

[mikeebowen]

I am not familiar enough with the OuterXml generation process to answer why certain elements are generated as self-closing elements instead of separate open and close elements. The OuterXml property however will be the functional equivalent of the xml in the part file, but it won't necessarily be a character for character match.

[Varorbc]

So, shouldn't you guys look into this issue instead of closing it rashly?