sed confusion

vkmgeek · 07-02-2008, 08:34 AM

[root@localhost ~]# cat t1
viral
hello
hi
root@localhost ~]#
root@localhost ~]# cat t1 | sed 's/\(.*\)./\1t/'
virat
hellt
ht
[root@localhost ~]# cat t1 | sed 's/\(*\)./\1t/'
viral
hello
hi
[root@localhost ~]#

I am confused. There is only one difference in last two commands and that is one dot is extra cat t1 | sed 's/\("."*\)./\1t/' in second last command.

To me that "." (dot) should not have any impact on output, yet it has...

How ?? Why ??

syg00 · 07-02-2008, 08:46 AM

Post deleted - didn't contribute like I had hoped.
(BTW, you don't need the "cat" but that's a different discussion)

nx5000 · 07-02-2008, 09:03 AM

Try it on this:

viral
hello
hi
he*llo

n* means zero or more of n
* with nothing before means *
n\* means n folowed by *

webaware · 07-02-2008, 09:04 AM

A '.' matches any single character. A '*' matches the preceding element zero or more times. Thus, '.*' matches any single character followed by zero or more 'any single character's.

Do a 'man perlre' for a quick primer on regular expressions. It's Perl-specific, but close enough to what sed uses.

pixellany · 07-02-2008, 09:07 AM

I am missing the question.

I would assume that ".*" and "*" do not mean the same thing, but then I don't know what "*" means by itself.

What is the desired result?

pixellany · 07-02-2008, 09:12 AM

Quote:

Originally Posted by nx5000

* with nothing before means *

I didn't know that!!

Is it true of any of the "modifier" characters?---eg "+" or "?"

If "*" by itself is read as the literal value, then how is "\*" read?

Mr. C. · 07-02-2008, 04:50 PM

Quantifiers require an atom preceeding them to be quantifiers. Otherwise, they are non-special. \* follows the order of precidence that \ quotes a character, so the * is quoted to a literal *.