Here it is. And some posts earlier. But this script works with Bookends, I didn’t use Zotero and don’t know if it is scriptable.
I wouldn’t rely on the OCR data, or data you can get automatically from PDF or PDF properties. There is often just garbage, and there is no sure regex to find what you need inside the PDF, unless it is some “hard format”. The best way I found - is to find the ref and click the script.