In actual practice, there are only a double-handful of regex features that are "incredibly useful," but those features are "incredibly useful!"

Okay, okay, it is "a language of chicken-scratches." That much is absolutely true.

Plus, when you are using regexes in shell scripts, you have to "backslash-escape" many symbols, such as periods and asterisks, so that the shell won't try to interpret them, itself.

It just takes a little getting used to.

However: if someone sees you using them, you must remember to drop your voice and to mutter phrases such as "Abracadabra." (Keeping a jar of newt's ears nearby is also a good idea, along with a jar of Orange Smoke.) No one must ever know that it is actually relatively easy to use!

It could've been worse. I could have said 'love' and 'regex'

On the serious side, most of my regex are simple ones. A few complex ones here and there, some regex I understand. I picked a lot regex expressions from commandlinefu.com.

1 members found this post helpful.

How revolting!

I'm sorry, you feel that way :/

Peace

22 Years in IT and I suck at RegEx.
I survive by asking those smarter than myself here, after I've tried a few examples on my own.

If I just spent about a week on it, it may even become un-spaghetti

5 minutes after this post, I'm sure I'll need help with my fail2ban regexes

I'm not an expert either. I started with the basics of regex and most of my scripts are using basic regex. It's almost impossible to decipher and remember those regex you see on the web. Sed and Awk deals with regex and have books dedicated to that subject.

Hya,

Now, I know that I am not the only one penguin who suffers from regex. Penguins can dive deep, so they can practice.

Shell escape and single/double quotes make things complicated together with '\|' and '|' in different versions of regex. (in addition to shell * and regex *)

Sigh,

cheers

I'm not a fanatic of regular expressions. I know simple bash regular expressions. Like

rm *
ls *.[ma]*
ls file??.txt
cp file.txt{,.bak}

And maybe a few other bash regular expressions.

What's hard for me is the regular expressions used by sed, awk, python, perl and other scripting languages. I have to google it.

Hya

Yes, meta-notation is another headache.

Now, may be I know the difference between regex and glob (??shell file name substitution).

cheers

By the way, does anybody know how *? behaves in sed regex?

I was sure that it behaves lazy way, but it does not. The "?" looks like to behave as an ordinary char.

I need to figure out how *\? behaves.

cheers

would be nice to give a full example (the command you want to understand)

Hya,

My motivation for post #25 was not on specific grammar or syntax. I am lost among deep confusion of regex sea.

Thank you very much for your listening to my SIGH.

cheers

In a basic regular expression, "?" is an ordinary character.

In "*\?" (or "*?" in an extended regular expression), the "\?" or "?" is basically a no-op. It makes the preceding atom optional, but the "*" already says it could appear zero or more times.

Here's an interesting variant:From experiment, that is interpreted as meaning that the character "c" must appear 2 or 3 times or not at all, so the strings "abxy", "abccxy", and "abcccxy" match, but strings with other numbers of the character "c" do not.

What I see is a bit different. see http://www.myezapp.com/apps/dev/regexp/show.ws and/or http://www.regexr.com/.
? will modify only the greediness.
Anyway I suggest to use these online regexp testers to understand how does it work (but it is not valid for sed, it has its own syntax, use sed -r instead)

I have tried the following /tmp/test:
and executed the following:
I cannot explain why

Note that I did use "sed -r", and FWIW, "grep -E" behaves exactly the same way, regardless of what the online regexp testers say.