"On Import" rule only runs on first document of batch

[bdamage]

I’m importing multiple single-page documents from ScanSnap, and have it configured like this.

and in DT4, I have this rule:

When I load six documents into the hopper and press the Scan button, I get six PDFs. All have names starting with recipe-card. But DT4 only runs this rule on the first of the batch — I have to select the other documents and manually run this rule against them.

I’m not sure if there’s something wrong with my setup, or if this is a bug. It’s been this way for a few months, but it’s still occurring in DT 4.2.2.

[cgrunenberg]

Please post a screenshot of the Rename Waitrose recipe cards smart rule, thanks.

[bdamage]

Thanks — here it is.

The AppleScript is

on scriptOutput(theRecord, theInput)
	do shell script "afplay /Users/user/Library/Sounds/Laugh.aiff"
	tell application id "DNtp"
		return name of theRecord
	end tell
end scriptOutput

(necessary because Play Sound still only allows me to choose system default sounds, rather than anything installed in $HOME/Library/Sounds)

[cgrunenberg]

I tested a similar combination of smart rules using the same script and also renaming via AI. Processing multiple selected documents worked flawlessly (DEVONthink 4.2.2, macOS 26.4). Which AI provider and model do you use?

[BLUEFROG]

• In your ScanSnap profile, to where are you saving the documents?
• Do you have a smart rule running OCR on the scans when DEVONthink receives them?

[bdamage]

I’m using a local Ollama install with the gemma3:12b model.

[bdamage]

I’m sending directly to DEVONthink. It probably would have helped if I’d noticed there was a scrollbar when I took that screenshot, as I could have included it

I’m not triggering OCR from any rules, but Settings > OCR > Convert Incoming Scans is set to “To searchable PDF”.

[BLUEFROG]

That’s the Send to. What is the path you’re saving to?

[bdamage]

Ah, sorry – it’s /private/tmp. I think I set it there because ScanSnap left the originals around after sending them to DT4, meaning I ended up with two copies — at least this way, they get cleaned up after a reboot

[BLUEFROG]

Try setting it to ~/Pictures and make sure DEVONthink’s OCR settings have Original document: Move to Trash enabled. That should remove the extra document.

[bdamage]

Thanks — that’s fixed the deletion issue, at least.

I thought it might be because Ollama spins the model down after it’s been idle for a set amount of time, and spinning it up again takes several seconds.

When the model needs to start, only the first document scanned seems to get processed correctly — but if I scan when the model is definitely running, it still doesn’t run the import action on all of them.

It seems to only process every other one (when I scanned 8 documents, I got documents 1, 3, 5, and 7 moved and renamed, while 2, 4, 6, and 8 staying in the Inbox with their original names of recipe-card00x and no recipe-card tag applied).

If I turn off OCR in DT4, then all documents get tagged, moved, and retitled correctly (and, bizarrely, all but the last are still PDF+Text, even though ScanSnap does not have “Convert to Searchable PDF” enabled!)

I do want to be able to leave DT4’s OCR enabled, though — I don’t really want to have to run it manually…