emsane

Emsane

Emsane is an Emancs frontend for SANE. Emsane runs scanadf with various jobs, to make it easy to scan books, magazines etc.

SANE is an acronym for "scanner acces now easy", scanner support available for free operating systems, see http://www.sane-project.org/

Author: Joakim Verona, (C) FSF 2009, 2010, GPL joakim@verona.se

Here are some of the interesting Emsane features:

• Emsane is a convenient Emacs interface to scanning
• Emsane handles one or several scanners
• Emsane can use different scanner settings for different sections of a book
• the power of Tramp lets you run scanners remotely over ssh
• you can dired your scans
• you can postprocess scans:
  • format conversions, to djvu for instance
  • unpaper
• there is a multi-scanner mode, scan different sections of a book with different scanners to speed up scanning

Brief usage:

• Define scanners and jobs in a file, "emsane-config.el"

for instance. See sample.

• load "emsane" and "emsane-config"
• start scanning with "emsane-scan-start
• check out the keybindings, but mostly you use "return" and "n" while scanning.

Configuring Emsane

Emsane is configured by writing Elisp expressions. This is a great advantage when you have many complicated presets. Primarily the configuration is made by creating instances of EIEIO classes. This is an advantage when you have many similar but not identical configuration items.

Emsane eieio classes

emsane-scanner

a set of options describing a scanner.

there are three supported scanner types:

• scanadf
• gphoto2
• fswebcam

emsane-section

a list of scanner and postprocessing settings. a section corresponds to a part of a printed matter. For a book the page numbering and colour content can change between different parts for instance.

A section definition can contain:

• paper size; expressed as an alias(ISO sizes or common book sizes), or width x height
• front or duplex scan
• color/lineart
• resolution
• file name pattern
• postprocessing options

emsane-job

a list of sections. you start a scan by selecting a job, then switching between the different sections.

optional extras

emsane.el is complemented by other Emacs packages such as dired, and some of my own packages such as dired-sequence.el, and a patch to the Emacs core to optionaly use imagemagick for image display, so djvu files can be shown in Emacs. (This was accepted upstream in Emacs 24)

buffers

• there is a control buffer per scanner.
• there is a post-operation buffer for each job.

different types of devices

Emsane was originally only supported scanadf, which is a Sane frontend for scanners with automatic document feeders. But scanadf also supports all the scanners Sane supports. In principle you could then use digital cameras with sanes gphoto2 backend. However, this is only beta in sane, so therefore Emsane also supports using gphoto2 directly, and also fswebcam for good measure.

History of Emsane

In 2010 I got a call from my wife while at work. I couldnt believe my ears, but apparently a fire had started at the neighbours house. I rushed home in a taxi, telling myself that it would work out fine.

When I arrieved, the fire department where already there and it seemed like it would turn out fine. Except the swedish climate makes for houses that have a very tight seal to the outside, so you cant easily get water inside to extinguish a fire. Also, since our type of house sits together in a long row, the fire department must decide on the most efficient way of saving all houses. So they made a hole in our roof and poured, I were told, 20 milion liters of water inside. The neighbours house was completely destroyed by fire. Ours was drenched in water and smoke, and the roof was destroyed by the fire.

Anyway, being a bibliophile, I had hamstered a large library at home. All my books were now to some degree damaged, and in the not so long term all my library would rot away.

I decided to convert my library into digital form. The most efficient method I could figure out was getting a number of ADF scanners, removing the bindings from the books and feed them through the scanners, and throw away the originals.

Remarcably enough there didnt seem to be a suitable scanner frontend that worked in accordance with my needs. Therefore I wrote Emsane.

Emsane is a bit quirky and maybe not really suitable for one-shot scans, but I've scanned tens of thousands of pages with it, so its actually industrial strength efficient when you get the hang of it.

not all of emsane is equally well used of course. patches welcome.

requirements

on Fedora: yum instal sane-frontends

WIP

DONE gphoto scanner

I want to be able to use gphoto to aquire scans(as well as existing support). Book pages are manually turned, and each scan also manually triggered.

gphoto2 --capture-image-and-download

scanadf has a gphoto backend, but it was tricky to control.

It turned out easier to add separate scanner classes.

TODO emsane ELPA package

TODO support more complex file name patterns

Each scan file name should correspond to the pages it contains. Normally theres one page per scan, but sometimes there are more(like when you scan spreads)

Scanadf is given a file name pattern and controls the file name. Therefore its easier to change the file name in the post operation stage.

TODO replace tx-vars with closures

the postop code has expressions like this: (emsane-postop-setenv tx 'SCANFILE filename) which are basically used as a poor mans closures. Now that Emacs have closures we can use them.

TODO image format conversions doesnt handle gphoto well

scanadf only produces pbm or pnm, and you can tell the scanner which image depth to produce.

gphoto2 is more dependant on what the camera produces, colour jpg, normally. cjb2 cant handle jpg input, so an aditional conversion stage is needed:

convert 0100-0001.scn -threshold 128 pbm:0100-0001.pbm && cjb2 -verbose -lossy 0100-0001.pbm 0100-0001.djvu

cjb2 doesnt seem to accept pipe input, and its not atm clear that this will ever produce good djvu output.