Hi,

I would like to write a script to filter out mutiples patterns string from the apache access log and create a new output file. Below is an example of my old script,

sed -e '/keepalive.html/d' -e '/Yahoo! Slurp/d' -e '/help.yahoo.com\/help\/us\/ysearch\/slurp/d' -e
'/msnbot-media/d' -e '/MSRBOT/d' -e '/Googlebot/d' -e '/+c.root+c.l\[l\-1\].arrow+/d' /usr/log/apache2/access.log > /tmp/www4-test/access-test

But I want to enhance the script by put all those string patterns in file so I can read them from file.

Any sugguestion and example of how to do that ?

Thank you very much.

Why not use grep?

Excellent suggested use of a grep pattern file. If you can make all the patterns "fixed" -- i.e. not regexes -- then you could add the "-F" option. . . .
1st put the patterns in 'patternfile' & check its contents:
Then run the code:

You could also rewrite your sed command as a sed script. Then you can modify the sed script to make changes.

I wonder which method is quicker & if it matters.

matyu,
Could you put one of your logs on a pastebin & post a link to it here. Please, please, please, do not clutter this thread by putting the log here, as it will surely do if it's long enough for those of us who want to play w/ it to have a good sample.

sed method with the "d" flag may be slower, as it has to "search" and "delete" the line.