I have attached a sample database which definitely works including a numbers spreadsheet of what the result should look like. There is nothing special about the construction of the database.
The script simply looks through the selected files and extracts information from each document. If the file has a url which ends in the form ?page= then it assumes that the file is an annotation and includes a reference to the original document and the page number in the export to numbers. The script does nothing with files that do not have a url ending in ?page= except export the file name to numbers.
I personally keep annotations in a separate folder but even if you don’t simply select all the documents that might be relevant. The spreadsheet will show you which documents are annotations and the rest can be deleted from the spreadsheet.
I hope this helps.
Frederiko
export to numbers test.dtBase2.zip (128 KB)