NetNews Usenet Archive 1992 #19/NN_1992_19.iso/spool/comp/unix/shell/3809

Text File | 1992-09-03 | 4.2 KB | 123 lines

Path: sparky!uunet!mcsun!ub4b!news.cs.kuleuven.ac.be!blekul11!ffaac09
Organization: K.U.Leuven - Academic Computing Center
Date: Thursday, 3 Sep 1992 22:35:16 +02
From: Paul Bijnens <[e-mail]>
Message-ID: <[e-mail]>
Newsgroups: comp.unix.shell
Subject: Re: Could someone help me with script?
References:  <[e-mail]>
Lines: 112


In article <[e-mail]>,
[e-mail] (Micah Anderson) says:
>
>I run the They Might Be Giants mailing list which has over 260 users on
>it, I am trying to make it much easier for me to subscribe and unsubscibe
>users, I wrote this script to make the addition of users easier, but it
>doesn't always work... comments are in brakets.
>
>#/bin/sh

You probably mean "#!/bin/sh" (with a "!"), but it probably works
without that also.

>#new.exe - adds new users to master list and mails them info files

From the pc-world, he.  I just hate suffixes for executable programs.

>#look for duplicates in new list and remove
>[what I do is go through all my mail and put all the addresses of people
>who want to subscribe into a file called new, I want to check to see that
>I don't have the same address in more than once - the problem with this
>is if there is no duplicate address and the file new.uniq is empty, it just
>sits there... any suggestions?]
>
>sort new > new.sort
>uniq -d new.sort > new.uniq
>grep -v `cat new.uniq` new.sort > new.tmp
>cat new.uniq >> new.tmp
>rm new.uniq
>rm new.sort

Why not just:  "sort -u -o new new"?

>#mail info files to new users
>mail `cat new.tmp` < Welcome
>mail `cat new.tmp` < tmbg.info
>
>#add new users to master list
>cat new.tmp >> /com/mailer/TMBG/USERS

You can combine this into the following step...

>#check for duplicates in master list and remove
>[here I am attempting to do the same thing as above, looking for duplicate
>addresses, if an address is in more than once they get twice as much mail,
>and is a big hassle...]
>
>sort /com/mailer/TMBG/USERS > /com/mailer/TMBG/USERS.sort
>uniq -d /com/mailer/TMBG/USERS.sort > /com/mailer/TMBG/USERS.uniq
>grep -v `cat /com/mailer/TMBG/USERS.uniq` /com/mailer/TMBG/USERS.sort >       p
>USRS.tm
>cat /com/mailer/TMBG/USERS.uniq >> USRS.tmp
>mv USRS.tmp /com/mailer/TMBG/USERS
>rm /com/mailer/TMBG/USERS.sort
>rm USRS.tmp

Add to master list, and remove doubles:

    MASTERLIST=/com/mailer/TMBG/USERS
    sort -u -o $MASTERLIST new $MASTERLIST

(Making the variable just to avoid typing pathnames too much.  Move
the variable to the front of the script, where you can see it easily
and change it once if the masterfile changes.)

>#output info to the log
>[here I would like to also output errors (duplicate addresses) into the file
>uniq -c will give me a 2 infront of any address that is repeated, BUT if
>an address has a 2 in it, I cannot grep for the 2... here to explain a little
>better:
>Ok, I am attempting to log when I have duplicates in a file, so I have this:
>
>uniq -c new.sort|grep 2 > new.log

Hey! "new.sort" is already removed a few lines back...
So, you probably should do the logging earlier.

>and I get something like:
>
> more new.log
>      2 [e-mail]
>      1 [e-mail]
>
>the first one is fine (as there were two of those and I needed to know that)
>but the second one just has a 2 in the address... how can I check just for the
>number in the beginning?]

For the number 2 in the beginning:  "grep '^2' new.sort", but...
what if you have a 3 instead of a 2?
You need something like:   "uniq -c new.sort | awk '$1 > 1' > new.log
Or if you do it in the beginning, before new.sort is made:

    sort new | uniq -c | awk '$1 > 1' > new.log
           (and I think you mean  ... >> new.log   ???)

>date >> new.log
>wc -l new >> new.log
>echo ' ' >> new.log
>cat new >> new.log
>echo '------------------------------'>> new.log
>rm new.tmp

How about this:

   (date; wc -l new; echo; cat new; echo --------------------) >> new.log

>any comments or suggestions to make this better are more than appreciated!
--
Paul Bijnens             -- MS-DOS is the world's most widespread virus.
Linguistics dept., K. University Leuven, Belgium
[e-mail]