editing files in a loop

I am a unix novice, so please forgive me if this sounds silly.

I have to go through about 350 log files on a Unix system, all in the same directory. I decided to write a small script to go through all of the files in that directory and invoke the "vi" command on them.

I did something like this:
find . -type f -print | while read f; do {
vi $f
}
done

That kinda worked, it does invoke vi on the first file. However, it seems to continue, putting all kinds of commands on the vi command line, and it eventually throws me out of vi on the first file.

I need to find a way to get the "vi" command to wait until I can look through the file and quit out of vi before it goes to the next file.

Can anyone help?
thanks.

if you are only reading them (if they'er logs i don't suppose you're doing sth with them) there's some better solution of your problem:
cat *|less

if all you want to do is read them, then less * would do the job fine, but if you want to parse data out of them, then you'll need to script something with sed, grep and awk, or probably more sensibly just write a PERL script. What exactly are you hoping to do with them? just reading log files is an odd thing to do.

christo

Less may work, then...all i have to do is go through and look for any errors, warnings, check row counts, things like that.

Thanks!

cat *|grep error|less

if you can give a detailed example of the data you need, then we can help you to knock up a proper script for the job.. otherwise you'll have to settle with Jabol's cat *|grep error|less. I think he's probably just trying to give you hints tho.

christo

I don't know how big these 350 files are, but even at one line each, your screen will fill several times over. If these are just subsequent versions of the same logfile, then you'll probably not need to know which file contains what error.

example:
http.access.20030806-04.log
http.access.20030806-05.log
http.access.20030806-06.log
etc...

However, if these logs come from many different sources, like

database.access.20030806.log
http.access.Aug2003.log
sudo.3rdquarter.log

...then that many lines will be confusing when you see an error.

If you use grep first, then it'll show you which file contains the found line.

grep error ./*

example:
[15:24:24 zedmelon@kilby bin] grep CMD ./*
./pidmail:CMD=${0##*/}
./pidmailer: echo "$CMD:
./startApache:CMD=${0##*/}
./startApache:EXECMD="/opt/apache/current/bin/httpsdctl.cp startssl"
./startApache: echo "Run $CMD as root, please.
./startApache: echo "$CMD: argument required."
./startApache: $CMD 1 | 2 | 3 | 4 | staging | test | all | h
./startApache: time, $CMD does no redundancy check."
./startApache:ssh $1 $EXECMD
./startApache: for COMPARE in $ACCPTCMD
./startApache: echo "$CMD: illegal option: $CHECKIT"

Of course this assumes that the word "error" appears verbatim in each file. If you have a file that contains "Error", a file that contains "ERR", and a file that contains "ERROR," you can use the -i flag, telling grep to ignore case.

-zedmelon

unix_novice, I must have missed your post before about looking for warnings and row counts (is a row count the number of LINES in the file, or info STORED in the file?).

maybe something like:



(Thanks to Scorpions4ever: my tabs no longer get nuked)

I don't know; that seems horribly inefficient, but I'm brainlocked for anything better, unless there are ways to tell the logs from different applications from other logs, just by filename. Then at the expense of getting more complex, your script can spend less time searching ALL files for EVERY possible error, even errors that will only be encountered in other files.

good luck.
-zedmelon

>> (wish I knew how to format this better, but my tabs get nuked)
Post your code within [ code] and [ /code] tabs (remove the leading space within the square brackets).

Thanks.

I'm leaving now, so you won't get to teach me THREE things in one day.
;)

Have a good one.

-zedmelon

if all the file to be scanned are in the same directory :

grep -E 'error|ERROR|Warning!|ABORT' *

or

ls | xargs -i grep -E 'error|ERROR|Warning!|ABORT' {}

or

ls -l | awk '{print$9}' | xargs -i grep -E 'error|ERROR|Warning!|ABORT' {}

or

just with a loop as suggested by zedmelon

:-)