Hi!

Trying my new knowledge on using pipes and grep on some ls filelisting output and notice grep really works strange.

If I try ls -l | egrep -i '*te*' it doesn't just find files containing the 'te'-string, it lists almost every file in my home-folder. Just check a little sample of the large output. Doesn't make much sence.
-rw-rw-r-- 1 seb seb 138454 maj 6 16:21 alien-night.1024.jpg
-rw-rw-r-- 1 seb seb 939871 maj 8 13:49 autoconf2.5-2.59-6mdk.src.rpm
-rw-r--r-- 1 seb seb 16 maj 25 14:03 bort
drwx------ 3 seb seb 4096 maj 24 16:39 Desktop/
-rw-rw-r-- 1 seb seb 2306461 maj 8 13:45 gtk+-1.2.10-41mdk.src.rpm
drwxrwxr-x 6 seb seb 4096 maj 25 20:40 JavaApplication1/
drwxrwxr-x 6 root root 4096 maj 25 12:37 rss-kde-0.3/
-rw-rw-r-- 1 seb seb 4858642 maj 25 12:28 rss-kde-0.3.tar.bz2
-rw-r--r-- 1 seb seb 200424 maj 3 23:22 stargate.jpg

While trying ls -l | egrep -i '*test*' it just finds the files containing the 'test'-string as follows.
$ ls -l | egrep -i '*test*'
-rw-r--r-- 1 seb seb 1039 maj 25 20:17 test.class
-rw-r--r-- 1 seb seb 428 maj 25 20:17 test.java
-rw-r--r-- 1 seb seb 415 maj 25 20:15 test.java~

Removing the ' ' around makes it find nothing at all.

Exchanging * for . makes it work....although my Unix-book says * is any number of characters and . is just one character, as usually is the case, but they also use it with grep/egrep.

So, is Linux different from Unix/Solaris, or am I somehow using this wrong (even though I almost copy the syntax in my Unix course-book).

Thanks

Read your book again

.* is any number of any characters

But to get back to your example
ls -l *test*
will do exactly what you expect.



Cheers,
Tink

Nope, just '*' in the book and not '.*'

And yes, I know I dont have to use grep for filtering filenames out, I'm just trying to learn grep, and learning grep on ls-output was a good idea I thought.
Anyway the '^' and '$' works for matching start/end of the string, so ls -l | egrep -i '^-' for example finds all files and skips directorys
Cant do that with only ls sadly.

So, I guess Unix/Solaris works different from Linux then....or my Cisco-book has some errors in it

Problem remains ...

* is a QUANTIFIER for grep, not a QUALIFIER.

if you use
egrep -i '*te*'
the first astrix has NOTHING to have any number
of, therefore will give you EVERYTHING.

If you did
egrep -i 'te*'
it would show you any line that has a t
followed by ANY NUMBER of e's.

If you use
egrep -i '.*te.*'
it will give you the expected result.


Cheers,
Tink

Ah, ok, I think I understand. And I guess the book is right too then

But I dont think the first .* is needed in this example.
egrep -i '.*te.*'
Cause grep returns the row if it finds 'te' anywhere...it doesn't care about start/end of lines unless you specify with ^ or $. Right?

Well well, very nice to have this cleared up a bit...thanks a lot
I was going nuts over here trying to understand why the * sometimes worked and sometimes totally messed everything up

Yep - it's not needed. I just threw it in because of
your use of *te

One thing I should have mentioned in the
example above is that te* (matches t followed
by ANY number of es) also includes "all ts
followed by ZERO es", in other words, anything
with a t in it ;}


Cheers,
Tink