OCR Fails for ZotMoov PDFs in Google Drive (Error: "PDF file is empty")

[noa-no-h]

Thank you all for the great work on this plugin! I'm trying to OCR pdfs stored in Google Drive (thanks to ZotMoov). They are downloaded to my mac and available offline. When ocr runs, I get an empty .ocr file and an empty note. The Zotero debug log reports an "InvalidPDFException: The PDF file is empty, i.e. its size is zero bytes" error. When i do manage attachment -> convert linked file to stored file, the ocr works perfectly.

• MacOS M4 15.4.1
• Zotero 7.0.15
• Zotero OCR 0.9.2
• ZotMoov 1.2.21
• Google Drive desktop 108.0.1.0
• pdftoppm path: /opt/homebrew/bin/pdftoppm
• tesseract path: /opt/homebrew/bin/tesseract

Debug log:

(4)(+0000000): INSERT INTO itemAttachments (itemID, parentItemID, linkMode, contentType, charsetID, path, syncState, storageModTime, storageHash, lastProcessedModificationTime) VALUES (?,?,?,?,NULL,?,?,NULL,NULL,NULL) [3166, 3112, 2, 'application/pdf', '/Users//Library/CloudStorage/GoogleDrive-/My Drive/Articles/Mead%2C+The+I+_+the+Me.ocr.pdf', 0]
...
(1)(+0000000): Error: Worker 'getFullText' failed: {"error":"{"message":"The PDF file is empty, i.e. its size is zero bytes.","name":"InvalidPDFException"}"}
getFullText/<@chrome://zotero/content/xpcom/pdfWorker/manager.js:650:21

Let me know if anything else would be useful to send. Thanks!

[aborel]

I'm not sure we can actually support files on Google Drive at all. At the moment I would only advise against it, but I'll try to investigate.

Can you try linked files (using Zotmov) in a local directory, just to check if that works?

[noa-no-h]

Thanks so much! And I was hoping that since the files were downloaded to my mac, they wouldn't be pulled from Google Drive.

Can you try linked files (using Zotmov) in a local directory, just to check if that works?

That works.

[aborel]

Thanks, I guess we can focus on the Google Drive aspect and leave Zotmov out of the way for now.

[noa-no-h]

thank you!

[aborel]

My first test with a linked file stored on Google Drive has worked normally. I've tried a locally stored output, a linked file output in a local directory and a linked file output on Google Drive: all tests were successful.

How large is your original PDF? Was it added from the same computer where you are running Zotero-OCR? Was it added through the MacOS Finder or through a browser?

ETA: my bad, you wrote explicitly about Zotmov, forget about the entry method. The size might still be a factor, so that question still stands.

[noa-no-h]

the most recent was 42kb

[aborel]

I have tested Zotero-OCR combined with Zotmoov on my Mac, I haven't observed any error.
Is there actually a 0-byte PDF file in your Google Drive folder?

[noa-no-h]

Good thinking, but this is a problem for every PDF file and it goes away as soon as i move those files onto my computer. For now, I just decided to store articles on my mac and then have that folder be synced to Drive instead of the other way around. It's working like that. Although I still don't know what went wrong before.

[aborel]

It's good if you found a workaround. I think we'll keep the issue open in case someone else encounters the same problem, or maybe we find a clue to the root cause.

[noa-no-h]

Great! thanks for all your help!

[aborel]

@noa-no-h if you can reproduce the problem with version 0.9.3, perhaps it will catch an error message that provides new information?