I know you are new to scripting but you really need to do some trialling of your own.
What I mean is, try running the following:
At the command prompt and see what actually happens??

The way scripting works is we are trying to implement a solution that would be run by hand and take a long time to do manually.
So essentially anything you can do on the command line is then used in your script as an entry.

Hence your question:
My answer would be, how do you run it on the command line?

cat alertpause | wc -l will print 'how many lines in alertpause log files' then in my case it would be printed 4, I still didn't get it in echo alertlog, cause when I ran 'cat alertpause | wc -l | echo alertlog' it will just print alertlog, without showing how many lines in alertpause.

Yeah, I'm currently new in bash scripting
now, I tried using tail but the printed log was didn't like what I expected.

just type python smsgateway.py "cell phone number" "message"
here python "message" parameter should read from alertpause then sent the last information about attack log files.

I believe that /var/log/snort always logging a new alert if there is an intrusion or an attack, I wonder how to make the earlier bash script to running in daemon mode, it will logging every new file in /var/log/snort then printed out in new file called alertpause. This is the real goal of my idea.

Now I'm using swatch to tail every new file in /var/log/snort/alert, and it works great, but the log file is too much complicated and I need a new file a little bit easier to read so I used the the previous bash script, and it works great again..thanx...but now I'm going deeper and can't sleep to think about my curiosity. Here, I got a new idea to start tailing alertpause with swatch but it nothing happens, cause I should running bash parsetext.sh every time to make itu logged on alertpause..

I wonder how to make it keep always read /var/log/snort/alert automatically which the last line of the alert will be printed out in alertpause, so the alertpause will be kept on updating the log, each time snort got an attack/intrusion ?

It would be easier if you run Snort with "-A fast" as this creates single line alerts. As an exercise it's definitely nice to code a alerting app but why try to reinvent the wheel when there's apps around? Or don't Snort.org, Freshmeat or Sourceforge list one that seems usable to you?

yes that's good idea, I'm sorry if my mind still growing like a young kid in here, but I tried to learn the basic idea to do more development on my network system. I have read a lot about snort conf @ snort.org, and I haven't opened yet the freshmeat..thx for the idea..

Cheers.