Summarise and index new documents in a Box folder

Point this at the Box folder your documents land in. Every hour it lists the folder, works out which PDFs it has not read before, and for each new one writes a summary document into your Google Drive and adds one row to your index spreadsheet.

There is no such thing as a reliable 'a file appeared' signal here, so this does the honest version: it lists the folder on a schedule and uses your own index sheet as the record of what it has already read. Column C of that sheet holds the Box file id, and a file whose id is already in column C is left alone. That means the sheet layout is fixed: A is the date read, B the file name, C the Box file id, D one line saying what the document is, and E a link to the summary document. Put those five headers on the tab before the first run. Nothing is ever overwritten - rows are only ever added.

Each document is read by field extraction, which returns the exact words it finds and an empty value for anything it cannot find - it never invents. The written note is then composed from those extracted details alone, so it cannot assert something the document does not say, and both the note and the raw details go into the summary document so you can check one against the other.

Reading a document is a smart action, so the first runs pause for you to review what was pulled out before they carry on. The summary document is written before the index row, so a run that fails half way repeats a document rather than losing it. Only PDFs are read; anything else in the folder is counted and left alone, and one file that cannot be read does not stop the rest. Ten documents per run by default - raise 'maxDocumentsPerRun' if your folder fills faster than that.