Adding html-files with inline images (base64-encoded) pollutes search index

[microsoft]

Adding html-files with inline images (base64-encoded) pollutes search index

[dayland]

We will have a separate explicit processing and pipeline for images coming in our Delta release. We will be using Azure Cognitive Services to extract metadata about an image and including them into the search index directly.

We don't have any official support for images in the current Gamma release and have not tested embedded images yet.

[AThomsen]

I'm not so much interested in analyzing the images as to just having them shown in the document preview. Will the Delta release have a solution for this or should we start preparing a solution for storing images elsewhere for previewing?

Also, when do you expect Delta to be released? :-)

[dayland]

@AThomsen , We are not planning to add any special processing to image files in Delta. We are however switching non-PDF text-based document processing to Unstructured.io. #256. They may have a better way to return inline images.

Delta is loosely targeted for end of October timeframe.