Copyright 2009 Brian Taylor
el.wubo@gmail.com

The files that make up the scan server are under the GNU General Public License (version 2 or greater)

tornado is included in this repo for convenience. It was written by the Facebook team and is redistributed under the apache 2.0 license.

INSTALLATION
============

The server is pretty messy and expects to be able to store its database and image files relative to its current working directory. In practice, for me, this has been just fine and seems pretty workable long-term. The script will create a database file in it's current working directory called "scan_data.db" and will expect a directory called `static' (or link by that name) to exist so it can store all its images. The script must be able to write to that directory as whatever user it happens to be running as.

Make sure you have imagemagick installed (need convert)

Drop the code somewhere.

Go somewhere (maybe where the code is) and make sure you have a static directory that can be written to.

Run python $somepath/scan_server.py in the directory that contains static

Go to localhost:8000 and bask in the glory

MORE PERMANENT INSTALLATION
===========================

Copy init.d/scan_server into /etc/init.d

Modify CWD to point at the directory where you put static

Modify USER to be the user you want the server to run as (must be able to read and create files and directories in static)

Symlink scan_server.py and scan_data.py into the directory where you put static

Run your platforms startup service utility to make sure that scan_server is executed at startup. 

For ubuntu: rc.update scan_server defaults

CONFIGURATION
=============

The variables that are most configurable are loosely oriented towards the top of the scan_server.py file. The defaults assume that SANE is installed and can automatically find your scanner. All scanned images, thumbnails, and medium scale images are stored in the static subdirectory of the current working directory of this program. The database is stored in scan_data.db.

THINGS THAT CAN IMPROVE
=======================

* The main document listing page is produced by iterating every key in the database. Obviously this won't scale.
* Image resizing blocks the server. It would be simple to make it happen in the background.
* Need to be able to change the name and tags for a document
* Need to be able to delete pages or move them to other documents
* Nicer styling would be cool
* Some image post processing
  - background smoothing (get rid of some of the paper effects)
  - detect BW/greyscale documents and convert
  - auto rotate assuming the document is a rectangle
* Is PNG the best choice for this? My images are HUGE!