Shell Programming and Scripting

I'm sure this has been asked and I'm overlooking something entirely simple, but a helping hand would be greatly appreciated.

Here it is:

I have a directory of text files (PHP session files) that are one line each.
I want to read each one and awk for the text in between the first set of quotes. Example:

username|s:6:"sysera";userid|s:25:"01";userstatus|s:13:"administrator";

I want to extract the text sysera out of that file. (It's to create a "current user list" type script).

Here is what I have right now, but I don't know how to make the first set of quotes (" ") the area to extract.

#!/bin/sh
for files in *;
do
cat $files | awk -F, '{print $1 " " " }'
done

Thanks guys!
-Sys

I'd prefer to use sed

sed 's/^[^"]*"\([^"]*\)".*$/\1/' $files

(on sample data, prints sysera)

Cheers
ZB

I shall give that a try. Thank you very much for the help.

-Sys

Edit: It is indeed working very well, thank you.

You can awk it and it's simpler than sed:

Code:

awk 'BEGIN {FS="\""} {print $2}'  test.dat

I'm running into one more issue with this. Is there a way I can get my script to run on files that only contain more than x number of characters? There are some garbage files that do not get cleaned up immediately that are causing the script to generate bad data.

So for example, ignore files that look like this, or have this many characters or less?:

username|s:6:"";userid|s:25:"";userstatus|s:13:"";

Here is the code I have currently(I have not yet had a chance to implement/try the awk version that was suggested):

for files in /.php_sessions/*;
do
sed 's/^[^"]*"\([^"]*\)".*$/\1/' $files >> users_online.txt
echo "
" >> users_online.txt

Thanks guys!

-Sys

Use something like

Code:

username=`sed 's/^[^"]*"\([^"]*\)".*$/\1/' $files`
count=`echo $username | wc -c | awk '{print $1}'`
if [ "$count" -gt "1" ]; then
  echo $username >> output_file
fi

You'll find that even if the username is empty, sed will output a newline, so thats why we test that the username is greater than 1. This stops files with blank usernames being added to the output list. You can use similar techniques to "filter" out other bad parts of the files.

EDIT: If you want to exclude files with less than, say 30 characters (or whatever you choose), then do

Code:

count=`wc -c < $files | awk '{print $1}'`
if [ "$count" -gt "30" ]; then
  # do the sed here
fi

Cheers
ZB

Thank you zazzybob, your examples have been extremely helpful.

*saves the code snippets to his ipod archive of junk to remember*