Alright, let's break down why only some of your images are working (and none after the 2/3 mark) in this Word document processing project you inherited. I’ve looked through your code and the problem description, and there are a few critical issues causing those frustrating red Xs.

1. The Core Matching Logic Is Broken #

Your current CopyImages method uses targetBlip.Embed.Value == sourceBlip.Embed.Value to match source and target images—but this makes no sense!

• The Embed value on a Blip is a relationship ID specific to its parent document. The source blips' IDs belong to testData (your source document), while the target blips' IDs belong to testReport (your template). These IDs are completely unrelated across documents. The few images that work are just lucky coincidences where the IDs happened to overlap by chance. Once you pass the point where IDs stop matching, all subsequent images fail to update, hence the 2/3 cutoff.

Fix for CopyImages #

Instead of matching on unrelated IDs, assume your source and target images are in sequential order (since you're extracting content from the source to insert into the template). Use an index-based loop to pair them one-to-one:

internal static void CopyImages(OpenXmlElement oldTable, OpenXmlElement newTable, WordprocessingDocument testData, WordprocessingDocument testReport)
{
    List<Blip> sourceBlips = DocumentHelper.GetAllBlips(oldTable);
    List<Blip> targetBlips = DocumentHelper.GetAllBlips(newTable);

    // Pair source and target images by their position (adjust if you use a different matching strategy)
    for (int i = 0; i < Math.Min(sourceBlips.Count, targetBlips.Count); i++)
    {
        var sourceBlip = sourceBlips[i];
        var targetBlip = targetBlips[i];

        if (testData.MainDocumentPart.GetPartById(sourceBlip.Embed.Value) is ImagePart imagePart)
        {
            // Create a new ImagePart and copy the source stream safely
            ImagePart newImagePart = testReport.MainDocumentPart.AddImagePart(imagePart.ContentType);
            using (var sourceStream = imagePart.GetStream(FileMode.Open, FileAccess.Read))
            {
                newImagePart.FeedData(sourceStream);
            }
            // Update the target blip to point to the new image part in the report document
            targetBlip.Embed.Value = testReport.MainDocumentPart.GetIdOfPart(newImagePart);
        }
    }
}

2. CopyEmbeddedVisioImages Has Two Critical Flaws #

Your updated CopyEmbeddedVisioImages method has two big issues that prevent images from working:

• You’re looping through every source object for every target object, which means later source objects overwrite earlier ones on the same target.
• You create a new ImagePart but never assign its relationship ID to the target’s ImageData—so the template never knows where to find the new image.

Fix for CopyEmbeddedVisioImages #

Again, use index-based pairing, and make sure you update the target’s relationship IDs after creating new parts:

internal static void CopyEmbeddedVisioImages(OpenXmlElement oldTable, OpenXmlElement newTable, WordprocessingDocument testData, WordprocessingDocument testReport)
{
    List<EmbeddedObject> sourceObjects = oldTable.Descendants<EmbeddedObject>().ToList();
    List<EmbeddedObject> targetObjects = newTable.Descendants<EmbeddedObject>().ToList();

    for (int i = 0; i < Math.Min(sourceObjects.Count, targetObjects.Count); i++)
    {
        var sourceObj = sourceObjects[i];
        var targetObj = targetObjects[i];

        // Handle ImageData
        var sourceImageData = sourceObj.Descendants<ImageData>().FirstOrDefault();
        var targetImageData = targetObj.Descendants<ImageData>().FirstOrDefault();
        if (sourceImageData != null && targetImageData != null 
            && testData.MainDocumentPart.GetPartById(sourceImageData.RelationshipId) is ImagePart oldImagePart)
        {
            ImagePart newImagePart = testReport.MainDocumentPart.AddImagePart(oldImagePart.ContentType);
            using (var sourceStream = oldImagePart.GetStream(FileMode.Open, FileAccess.Read))
            {
                newImagePart.FeedData(sourceStream);
            }
            // Assign the new relationship ID to the target's ImageData (this was missing!)
            targetImageData.RelationshipId = testReport.MainDocumentPart.GetIdOfPart(newImagePart);
        }

        // Handle OleObject
        var sourceOleObject = sourceObj.Descendants<OleObject>().FirstOrDefault();
        var targetOleObject = targetObj.Descendants<OleObject>().FirstOrDefault();
        if (sourceOleObject != null && targetOleObject != null
            && testData.MainDocumentPart.GetPartById(sourceOleObject.Id) is OpenXmlPart openXmlPart)
        {
            EmbeddedObjectPart newEmbeddedObj = (EmbeddedObjectPart)testReport.MainDocumentPart.AddPart(openXmlPart);
            targetOleObject.Id = testReport.MainDocumentPart.GetIdOfPart(newEmbeddedObj);
        }
    }
}

3. Document Open Mode Might Be Read-Only #

You mentioned using using(WordprocessingDocument foo = WordprocessingDocument.Open(bar, false){}—the false here means you’re opening the document in read-only mode. Any changes you make won’t be saved to the file!

Make sure you open your template document in writable mode:

using (var testReport = WordprocessingDocument.Open("your-template-path.docx", true)) // true = writable
{
    // Run your image copy methods here
    testReport.Save(); // Explicit save is safe, though using block will auto-save on disposal
}

Final Notes #

• Always wrap streams in using blocks to ensure they’re properly closed and flushed—this prevents partial image data from being written.
• If your template uses named placeholders for images (instead of sequential order), adjust the matching logic to target those placeholders (e.g., by checking image alt text or shape IDs) instead of using indexes.

内容的提问来源于stack exchange，提问作者emsimpson92