Is there any way to use zotero to extract metadata from a pdf as a third-party library? Or using an online service that receives a pdf file and returns its metadata?
The service is purposefully not run as a public service with an API (as e.g. APIs it in turn relies on have rate limits, sites may perform differently etc.)
I'm afraid not, no (sorry, the code is private because parts of it can't be properly exposed and the public version was thus useless. I thought it was self-contained)
Upgrade Storage
But the server code is public in its entirety and could be run on relatively easily as a service: https://github.com/zotero/recognizer-server