Can remote image Urls be passed into Markdown export when using the ImageRefMode.REFERENCED

[RichardGrundy]

Hi,

Have been building a pipeline for processing docs into Markdown and want to extract images into a blob storage container, setting the ImageRef to a https:// url rather than data:// or file://

looking at the code it ignores https:// urls for the images in the following section of document.py:

  elif image_mode == ImageRefMode.REFERENCED:
            if not isinstance(self.image, ImageRef) or (
                isinstance(self.image.uri, AnyUrl) and self.image.uri.scheme == "data"
            ):
                return default_response

            if (
                isinstance(self.image.uri, AnyUrl) and self.image.uri.scheme == "file"
            ) or isinstance(self.image.uri, Path):
                text = f"\n![Image]({str(self.image.uri)})\n"
                return text

            else:
                return default_response

        else:
            return default_response

Or am I using this wrong? Should I be using a different method?

I am happy to make a PR with the change if this is needed.

R

[dolfim-ibm]

It is definitely one of the use cases we want to support, but I think it was missed in the recent changes. We will fix it quickly.

Our intent was to avoid fetching the remote urls (or better, some nice design should be put in place to allow for it), but the url is already a remote one, it should simply be in the generated output.

cc @cau-git @PeterStaar-IBM

[PeterStaar-IBM]

Yes, I would just like to have an option that the user can use in order to be aware he is pulling data from remote urls. I am not yet sure what the correct API design is here. Happy to get suggestions.

[simjak]

maybe api should look like this?

 # Save markdown with externally referenced pictures using base_url
    md_filename = output_dir / f"{doc_filename}-with-image-refs.md"
    conversion_result.document.save_as_markdown(
        md_filename,
        image_mode=ImageRefMode.REFERENCED,
        base_url=base_url  # Set the base URL for image references
    )

    # Save HTML with externally referenced pictures using base_url
    html_filename = output_dir / f"{doc_filename}-with-image-refs.html"
    conversion_result.document.save_as_html(
        html_filename,
        image_mode=ImageRefMode.REFERENCED,
        base_url=base_url  # Set the base URL for image references
    )

@cau-git @PeterStaar-IBM