New ImagePart is created as .bin instead of real extension

[Asbjoedt]

Describe the bug
I have asked the question on Stackoverflow:
https://stackoverflow.com/questions/75100804/embedding-image-as-new-imagepart-creates-a-bin-file-using-open-xml-sdk

I don't expect to get any answer there, and I am also increasingly curious whether, this is an implementation error but an issue in the SDK.

Issue is, I create new embedded image in Excel and it is saved to /xl/media/image.[WhatEverExtensionItIs]
If I create a new ImagePart with the SDK and feed it data like this:

ImagePart new_ImagePart = worksheetPart.AddImagePart(ImagePartType.Tiff);
stream.Position = 0;
new_ImagePart.FeedData(stream);

Then the file will be saved as /xl/media/image.bin.

This behaviour it should not do. It should write the extension as .tiff, since this is the ImagePartType, that I specify. Or secondly it should read the stream and identify it as .tiff through i.e. a magic number.

If you unzip the package and change the image extension to real extension, then the image will open perfectly and I have used a TIFF validator to confirm the image is well-formed.

What do you think of this? Am I doing anything wrong in my implementation, or can we change the SDK to write real extension like Excel does?

Screenshots
If applicable, add screenshots to help explain your problem.

To Reproduce
Steps to reproduce the behavior:

1. Use the supplied code to stream any TIFF image (or change ImagePartType to any other image type) to a new ImagePart
2. Unzip the spreadsheet and go to /xl/media/ to find the file as a .bin
3. Change extension to .tiff and the image will open perfectly

Observed behavior
ImagePart creates a new image with .bin extension

Expected behavior
ImagePart should create a new image with extension based on ImagePartType.

Desktop (please complete the following information):

• OS: Windows 11
• Office version: 2212
• .NET Target: .Net latest version
• DocumentFormat.OpenXml Version: 2.19

[Asbjoedt]

Will deleting the following lines be enough to prevent this behaviour? I am imitating this #1292 fix, that @mikeebowen made. He replaced .xml with .bin. Instead, I delete the same lines.

From: ImagePart.json line 6

"Extension": ".bin",

From: PartConstraintData.json line 1518

"TargetFileExtension": ".bin",

From: ImagePart.cs line 38

string ITargetFeature.Extension => ".bin";

[faerbersteve]

I'm having the same issue within the Word document when embedding png files.
In compare with the older version 2.0.5022 of DocumentFormat.OpenXml the image is added there as .png.

@Asbjoedt Do you have a working fix for the issue?

[Asbjoedt]

@faerbersteve I think deleting the above lines are maybe only a part of the way.

I found the following methods in the SDK. I don't have a clear overview of how to proceed from here though.

       internal static string GetContentType(ImagePartType imageType)
            => imageType switch
            {
                ImagePartType.Bmp => "image/bmp",
                ImagePartType.Gif => "image/gif",
                ImagePartType.Png => "image/png",
                ImagePartType.Tiff => "image/tiff",

                // ImagePartType.Xbm => "image/xbm",
                ImagePartType.Icon => "image/x-icon",
                ImagePartType.Pcx => "image/x-pcx",

                // ImagePartType.Pcz => "image/x-pcz",
                // ImagePartType.Pict => "image/pict",
                ImagePartType.Jpeg => "image/jpeg",
                ImagePartType.Emf => "image/x-emf",
                ImagePartType.Wmf => "image/x-wmf",
                ImagePartType.Svg => "image/svg+xml",
                _ => throw new ArgumentOutOfRangeException(nameof(imageType)),
            };

        internal static string GetTargetExtension(ImagePartType imageType)
            => imageType switch
            {
                ImagePartType.Bmp => ".bmp",
                ImagePartType.Gif => ".gif",
                ImagePartType.Png => ".png",
                ImagePartType.Tiff => ".tiff",

                // ImagePartType.Xbm => ".xbm",
                ImagePartType.Icon => ".ico",
                ImagePartType.Pcx => ".pcx",

                // ImagePartType.Pcz => ".pcz",
                // ImagePartType.Pict => ".pict",
                ImagePartType.Jpeg => ".jpg",
                ImagePartType.Emf => ".emf",
                ImagePartType.Wmf => ".wmf",
                ImagePartType.Svg => ".svg",
                _ => ".image",
            };

        /// <summary>
        /// Adds a ImagePart to the DrawingsPart
        /// </summary>
        /// <param name="partType">The part type of the ImagePart</param>
        /// <param name="id">The relationship id</param>
        /// <return>The newly added part</return>
        public ImagePart AddImagePart(ImagePartType partType, string id)
        {
            var contentType = ImagePartTypeInfo.GetContentType(partType);
            var partExtension = ImagePartTypeInfo.GetTargetExtension(partType);
            Features.GetRequired<IPartExtensionFeature>().Register(contentType, partExtension);
            return AddImagePart(contentType, id);
        }

[Asbjoedt]

Hi @twsouthwick, @tomjebo

I have written a fix for this, you can see here: https://github.com/dotnet/Open-XML-SDK/compare/main...Asbjoedt:Open-XML-SDK:ImagePartFix?expand=1

I am not entirely sure the fix works, because I have problems building Open XML SDK to test the new code (approx. 200 namespace references are missing). Maybe you can help me out somehow.

The fix does two things:

• Remove any .bin references for ImagePart
• Clean up ImagePart code by removing several lines of outcommented code related to image file formats, that were not implemented when the initial code were made.
• The code also recommits the code in the recently approved PR for Add two extensions to GetImagePartType #1304 (this is a mistake from me, I will be better at keeping commits properly branched in the future, but the recommit should not be an issue)

[twsouthwick]

This file is where the .bin is coming from:

Open-XML-SDK/data/parts/ImagePart.json

Line 6 in d46f92b

"Extension": ".bin",

[Asbjoedt]

Great! thx.
I will verify this when 2.20 is released.

[Asbjoedt]

Confirmed fix.

[Wenfengcheng]

@mikeebowen Not fixed for ImagePart generated by this way in version 2.20:

public static string AppendImagePart(this TypedOpenXmlPart typedOpenXmlPart, byte[] stream, string contentType)
{
    var rId = typedOpenXmlPart.GetNextRelationshipId();

    var imagePart = typedOpenXmlPart.AddNewPart<ImagePart>(contentType, rId);

    using var mStream = new MemoryStream();

    mStream.Write(stream, 0, stream.Length);

    mStream.Position = 0;

    imagePart.FeedData(mStream);

    return rId;
}

[mikeebowen]

@Wenfengcheng , What is the issue you're seeing? Does it use the wrong extension? Does it throw an error? Something else?

[Wenfengcheng]

@mikeebowen yeah, the extension is wrong which is .bin file, not the origin image extension.

[psyvision]

+1 can confirm the same issue of a .bin in /media/image.bin using v2.20 when saving a word document:

var codes = ImageCodecInfo.GetImageDecoders();
var mimeType = codes.First(c => c.FormatID == image.RawFormat.Guid).MimeType;

var imagePart = mainPart.AddImagePart(mimeType); // image/jpeg

using var stream = new MemoryStream();
image.Save(stream, image.RawFormat);
stream.Position = 0;
imagePart.FeedData(stream);

[twsouthwick]

@mikeebowen can you take a look to see what's different here?

[twsouthwick]

We'll track this in the new issue as it looks subtly different