Re: [LUG] filtering recovered jpeg files

[Gordon Henderson <gordon+lug@xxxxxxxxxx>]

On Mon, 20 Aug 2018, Pentiddy wrote:

> Hello all...
> A few years back I made the novice mistake of losing all my photos due to a 
> disk crash...
> I did however manage to run a recovery program on the disk and get images off 
> it.
> I am wanting to finally sort through these and delete spurious thumbnails, 
> internet cached images etc. and wondered if there is a convenient way of for 
> instance filtering just full sized pictures out of the folders, or deleting 
> files below a certain size... I am Xfce based so thunar scripts an option...
> Any help with this greatly appreciated- trying to sort out some memorable 
> photo's for my Daughter leaving home.

How good is your shell-fu?

You can use the jpeginfo command to get the size:

  jpeginfo *.jpg > /tmp/foo

then edit /tmp/foo and look for the ones that are not thumbnails.

A quick example output:

  IMG_20180116_193807.jpg 4160 x 3120 24bit Exif  N 3832530
  j.jpg  495 x 811  24bit Exif  N  104293
  loaf.jpg 2000 x 2694 24bit JFIF  P  753672

So the first is a camera image, the 2nd - an odd size, but probably 
something scaled for the web, the last another (probably) processed image. 
Thumbnails might be 128x or smaller, so you can manually get rid of them 
by sifting through the file...

However something like:

  sort -k2 -nr /tmp/foo > /tmp/foo2

will reverse sort the file on the 2nd field (the X size) and write it to 
/tmp/foo2 - using the example above yields:

  IMG_20180116_193807.jpg 4160 x 3120 24bit Exif  N 3832530
  loaf.jpg 2000 x 2694 24bit JFIF  P  753672
  j.jpg  495 x 811  24bit Exif  N  104293

... you can then manually edit the file and cull the lines with X size 
smaller than something you determine (this sorts largest at the top, so go 
to the end of the file) - or futher edit commands into the file, to turn 
the file into a script with e.g. rm commands at the start if the files you 
want to delete.

And so on.

You can get more clever using extra stuff like the awk command which is 
better at finding a <= number thing, but then you're into diminishing 
returns for a one-off task. Faced with a few directories of a few 1000 
random files then this is how I'd do it.

Gordon

--
The Mailing List for the Devon & Cornwall LUG
https://mailman.dclug.org.uk/listinfo/list
FAQ: http://www.dcglug.org.uk/listfaq