[From Speed Up Firefox web browser | Ubuntu Geek]

What bothered me about this was that all the files scanned always end up named "scan12345.pdf". Because of the way I file, I like having my things as "year/company/year-monty-date.pdf" instead.

For a while there I was spending hours manually moving files to the correct folders while watching TV or doing another "using half-my-brain" activity. I thought there had to be a better way. I'm a bit handy with software, so I started a little ruby script to do the move for me.

The script uses pdftotext (you can get pdftotext using MacPorts ) and ruby to determine information inside the actual PDF (scan them as "searchable PDFs" so it OCRs the text). Once it makes that determination, it does the file move to the appropriate place. A series of Regular expressions inside the script determine what company it belongs to, and a strong date parser takes a look at anything that looks significant enough that has a date attached to it in order to determine an appropriate date IN the scanned document, as opposed to the date of the file (which it uses if it can't find any parsable dates inside the document).

So now I had a script that could do those moves properly, but still didn't want to have to remember to continuously running the script. Here is where the Mac's Folder Actions feature comes in.

Using Folder Actions, I wrote the following script to wait for the file to drop in the folder (the scanning program on the Windows side uses a VMWare shared folder to drop the PDF on a mac folder):

It's not perfect. It uses a delay mechanism to wait for the Windows side to finish writing the PDF (dumb scanning program creates the zero-byte file and waits to fill it until it has run OCR on the whole thing). But because the folder action works on all PDFs on the document, it can pick up the ones it couldn't run pdftotext on the last time, so it's good enough for now. It also has a real problem pulling the correct date on documents where a lot of patterns could be a date (need to work on a good algorithm - I haven't found one yet).

Anyway, here are the auto_file script and the applescript folder actions script that triggers it.

Hopefully it will be useful for you if you have a similar need or just for learning how you can streamline some of the stuff you do every day with your Mac. Cheers!

I have determined that the problem was the lack of an "always there" todo list and note taking device. I write too slowly and am not organized enough to use my moleskine (never mind that I also tend to forget it), and my lifestyle is too mobile to just use a laptop.

I tried iGTD and Omnifocus. Ominfocus was too heavy on the resources, though by now I should probably try it again (maybe when they come up with an iPhone rich client). I also used Circus Ponies' Notebook, which was pretty nice but didn't do spotlight with enough granularity. Then I tried Journler (for journaling, never used it for GTD) and now I'm trying out Evernote. So how does Evernote compare?

Well, it's not as nice as Journler. But it syncs online and does OCR, which is pretty unique. But only kind of works out for me because it's not ubiquitous.

The problem with syncing online is that sometimes you just don't have access to the online component. For example, I have a shopping list. I go to it online using the iPhone. When I enter the supermarket, signal drops and I no longer have access to the stuff I'm shopping for! (this *has* happened to me). As soon as there's an iPhone rich client that syncs locally, I'll be much happier.

Having said that, I also like features. And the last program I tried, Journler, spoiled me with them. To the point that I keep comparing Evernote to Journler, and find Evernote lacking.
Both Evernote and Journler support full Spotlight searching which is nice. But Journler (probably because they only have to worry about the client app) is a lot more polished. For example, in Journler you can add multiple pieces of content on the same entry, and these can be audio, video, and photos (through the iSight). When taking a picture with evernote (the only thing that's supported), it always creates a new entry. I can't seem to drag images from one note to another, though it does suport dragging and dropping multiple images.
A big selling point for Evernote is the text recognition that happens on the server. But it seems like it's only half a feature. For example, it recognizes the text and keeps it hidden, like an OCR'd PDF. It helpfully highlights the found words as it finds them during a search, but it doesn't let me drag and select text inside the image, like an OCR PDF would do in something like Preview. I know there's backing text being stored, and I would like to be able to select it to copy/paste, particularly for things like quickly updating my address book with information I photograph off a business card, or preparing a document based on some text that came in as a graphic.

Journler supports real smart folder searches, categories and tags, which should make it possible to set up a GTD-like system based on what you write. Evernote only has tags, and saved searches, which let you save a textual search and a search that is only limited to a particular set of tags, except that multiple tags work like an AND instead of an OR (and you can't change this), so it makes it less flexible that it could be in practice.