How to compare files in 2 dirs and write results on a file

fpp666 · 07-12-2009, 03:55 PM

Hi, I need to write a script using ksh that compares files from two directories.

I need to do a comparison between historic output files from a program and the new output files of the same program, to verify that changes made to the code did not affect the overall result of it.

The original files and the new files have the same name, but the new files have an additional .rdy at the end. Other than that, the files are the same, but have to be compared to the corresponding original file.

The result of this has to be written in a file, for an easier verification.

Hope you can give me a hand, thanks in advance!
Fernando

corbintechboy · 07-12-2009, 04:11 PM

Check out the command diff.

Sergei Steshenko · 07-12-2009, 05:53 PM

Quote:

Originally Posted by corbintechboy

Check out the command diff.

And/or 'cmp'.

fpp666 · 07-13-2009, 08:44 AM

Ok, I've been reading and I think I get how diff works now, but I have a doubt about how to diff a bunch of files.

Should I just list all the diff commands for the files I need to compare one after another, or maybe put them on a list and get the script to pick them up from there?

Sorry about all the questions, I'm really new to UNIX.

Thanks!

colucix · 07-13-2009, 08:52 AM

If the files had the same name, you could have recursively compared the content of two directories using option -r of diff. In your case you can use a loop like this:

Code:

#!/bin/bash
dir_old=/path/to/dir/containing/original/files
dir_new=/path/to/dir/containing/new/files
cd $dir_old
for name in *
do
  echo diff "$dir_old/$name" "$dir_new/$name.rdy" >> logfile
  diff "$name" "$dir_new/$name.rdy" >> logfile
done

chrism01 · 07-14-2009, 01:58 AM

See also 'comm' http://linux.die.net/man/1/comm

PMP · 07-14-2009, 02:26 AM

use diff -r
It will handle sub directories as well

Code:

 diff -r dir1 dir2 > files.diff

fpp666 · 07-14-2009, 10:53 AM

Thanks for the responses, guys!

One thing I forgot to mention: The original output files are all together in the same directory, but the newly generated files are separated in sub directories (e.g. newfiles/switch1, newfiles/switch2).

How can I make the script look recursively in one side of the diff, and always on the same dir for the other half of the diff sentence.

I tried what Colucix suggested, using one of the subdirs as new_dir, but I get a huge output dir, because the script tries to compare the complete set of original files (about 120) with the 3 that are located in that subdir.

I really appreciate your help, cheers!

colucix · 07-14-2009, 11:31 AM

Quote:

Originally Posted by fpp666

One thing I forgot to mention: The original output files are all together in the same directory, but the newly generated files are separated in sub directories (e.g. newfiles/switch1, newfiles/switch2).

Keep it simple and put the new files in one directory and keep the original names without the suffix .rdy? In that case you can use diff -r.

If you like to make things complicate...

you have to use the find command to locate where the corresponding new file is. For example something like this:

Code:

#!/bin/bash
dir_old=/path/to/dir/containing/original/files
dir_new=/path/to/dir/containing/subdirs/with/new/files
cd $dir_old
for name in *
do
  newfile=$(find $dir_new -name ${name}.rdy)
  echo diff "$dir_old/$name" "$newfile" >> logfile
  diff "$name" "$newfile" >> logfile
done

fpp666 · 07-15-2009, 12:27 PM

Excellent advice, guys, I really appreciate it!

Cheers!
Fernando