shell script tp compare to liste of names

xpucto · 01-27-2009, 06:35 AM

Hi!

I would like to write a script that compare 2 lists of name:1 list contains all names of a DB, 2 second list contains only the names in the DB of activ users. I would like to compare the two lists in oder to get a list of non active users (users that are in the 1st list but not in the second list).
So if the 1st List looks like this:

Quote:

John
Bill
Helen
Chris
Thomas

and the second like with active user this

Quote:

Helen
Chris

Then the script schould create a 3rd list of inactive users:

Quote:

John
bill
Thomas

I would be very thankful if I could get any help with it.

colucix · 01-27-2009, 06:43 AM

This is a task for the comm command, for example

Code:

comm -3 <(sort list1) <(sort list2)

See man comm for details and look at process substitution in the Advanced Bash Scripting Guide about passing the output of a process as input to another process (as I did to sort the two list in the example above).

AnanthaP · 01-27-2009, 06:44 AM

Using something like awk or perl, you should put the data into "associative arrays" aliass "hashes" with the key value as the name and two flags, one each for occurence in each file.

After scanning both files, the key values flag 2 are the current list, and so the key values with no flag 2 are your exception.

Try it.

End

colucix · 01-27-2009, 07:05 AM

Just seen the editing of your first post. I can suggest an alternative to my previous post, that will preserve the order of the names in the first list:

Code:

grep -f list2 -v list1

This will read patterns from list2 and will exclude them from list1.

xpucto · 01-27-2009, 07:16 AM

Thanks a lot for your help! comm -3 has done the job!

xpucto · 01-27-2009, 07:17 AM

thanks a lot for your help! comm -3 has done the job.

Quote:

grep -f list2 -v list1

even nicer! thanks!

colucix · 01-27-2009, 08:21 AM

A slight correction to myself: the grep command has a side effect, since it consider the names in list2 as patterns and not as real names. That is if you have two similar names like Bill and Billie in list1 and only Bill in list2, both of them in list1 will be excluded. You can avoid this behavior if you use option -w of grep to force the pattern to be considered as a whole word, not just a pattern inside a line:

Code:

grep -w -f list2 -v list1