@Giuseppe - you had the error, so you should have seen what the result was based on whether it extracted isbns for other books in the batch or not. Of the top of my head I can't remember what the code would do, and having that PDF component crash is not something I see in my own usage to replicate (but then I try to avoid keeping PDFs where possible).
At a guess from memory this plugin breaks the work up into batches of some default size (50?). Each batch will get treated independently from each other, so an error in one batch should not affect others. As for what happens within a batch, my guess would be that with it being an unmanaged exception thrown by pdftohtml you will not get any isbn values from that batch back.