That’s what I was assuming I’d have to do. I was hoping for something that would require one less step as well as less exposure of data on the file system.
I don’t need a two-way sync, because I have no intention to edit the material from within DEVONthink. Each of these apps has data elements attached on a UUID and distinct URL scheme, which is how I would assume I could create the unique filename per application.
That said, I’d love to be able to link back to the original item in the database, so if there is a way to do that, I’d like to know. I assume that an unsophisticated version of this would be to stuff a link to the original item somewhere in the exported content.
I’d also welcome suggestions on file naming and formatting to make for effective indexing as well as how to take advantage of item metadata.
For Bear and Day One and Agenda, the content structures are basic note structures (title and text content), so I’m assuming that would be relatively straightforward to index once it’s been exported. That said, extracting information from each of those isn’t terribly scriptable at the moment without using AppleScript’s System Events, so I figure there might be some kludges necessary to get the data available from those.
OmniFocus is much more scriptable, yet the object schema is more complex. Tasks, Projects, Folders, and Tags are their own types of objects, and while they each have a notes field, they have varying degrees of additional and useful metadata to include as well.
Another thing I’ve considered is skipping the indexing and adding these items as Bookmarks, assuming DEVONthink indexes bookmarked content from regular URLs. That said and as far as I can tell, there’s no way to read the content from URLs that aren’t http-/https-based. A workaround could be to create a custom web server on my computer to mimic/expose read-only x-callback-urls via a custom web server which would handle the web request for the resource then call the x-callback-urls locally and send back the content in the http response. That seems like its own kind of Rube Goldberg machine / security nightmare, so I’d like to avoid that for now.