A “no text” PDF simply indicates that the PDF doesn’t contain searchable text. That’s not an error, but information.
For example, an image of a paper receipt saved as PDF by a scanner is merely a picture of the receipt. In order to make that PDF searchable, it can be subjected to OCR (optical character recognition), so as to produce a searchable PDF that contains a text layer as well as an image layer.
You can add a Kind column to a DEVONthink database view window (View > Columns > Kind). Searchable PDFs are identified as Kind = PDF+Text. Non-searchable PDFs are identified as Kind = PDF.
You can create a smart group that lists non-searchable PDFs using these criteria:
Kind = PDF/PS
Word Count = 0
Not all non-searchable PDFs need to be OCRed. For example, I’ve got a friend who occasionally sends image files (photos or graphics) as PDF files. They don’t include pictures of text, so there’s no point in OCRing them.