UNIX for Advanced & Expert Users

I am learning regex fundamentals on my own and when I try to use \w for characters (i.e. [a-zA-Z]), or \d for digits ([0-9]) it doesnt work even though I see in greps man page that \w should be the same as [[:alphanum:]]...

[[:alphanum:]] and those types of syntactic character classes do work for me, its just the shorthand \w \W and \d \D dont. I am using mac os x which should be similar to freebsd with bash shell.

?

---------- Post updated at 02:36 PM ---------- Previous update was at 02:32 PM ----------

I meant [[:alnum:]] not alphanum. Where can I get a list of character classes that work besides he man page? I am using: grep (GNU grep) 2.5.1

---------- Post updated at 03:30 PM ---------- Previous update was at 02:36 PM ----------

I am becoming increasingly confused about regex and bash.. the regular expressions that work with vi differ from the ones that work with bash? can someone break it down to me where pattern matching and regex differ?

There are multiple variants of regular expression. There are the original POSIX basic regex, with a very simple syntax. Then there's the POSIX extended regex, which is basically basic POSIX regex with a simpler usage. Both support character classes like [:alnum:], [:blank:] and the like, as a means of saving characters & improving readability.

A lot more powerful are Perl Compatible Regular Expressions (PCRE), as copied by almost any modern regex engine (Java, Ruby, PHP, .Net, ...). That introduced backslash sequences as a means for writing character classes even shorter, eg \d for any single digit, \W for any non-word character, ...

Now you're saying that the regex don't work for you. Could you give an example of what's your input, and what command line you've tried? Maybe the issue isn't with the regex, but the way it's used.

Well is seems you're saying that \w is a post grep regex character class, and used for PERL but not grep? I would try something like

Code:

#echo hello |grep \w+

and get no results.

---------- Post updated at 04:44 PM ---------- Previous update was at 04:38 PM ----------

Code:

 echo hello |grep [[:lower:]]
hello


echo hello |grep [a-z]
hello

These work. I expected

Code:

echo 123 |grep \d

to give me 123 back

I guess \w \W \d \D dont work with grep, they dont seem to work in vim either.

I just went through the man page for grep, and it seems that grep, in it's basic incarnation, does not support the sequences introduced by Perl. That is, until you add the -P switch:

Code:

$ echo '123' | grep '\d'
$ echo '123' | grep -P '\d'
123

And no, vim usually isn't compiled with support for PCREs, at least on no system I've ever check for it.

Wow, even after going back and looking for it I couldnt find the -p switch, I actually had to grep the grep manual for it! Thanks, that's what I was looking for!

---------- Post updated at 09:37 PM ---------- Previous update was at 12:53 AM ----------

I got some errors using the -P switch, but I found the \w character class works well when double quoted and with the -E switch for extended regex.

---------- Post updated at 09:40 PM ---------- Previous update was at 09:37 PM ----------

Code:

:~ localadmin$ echo "Hello 123 abc <># .. asdf" |grep -Eo "\w+"
Hello
123
abc
asdf


STLM011:~ localadmin$ echo "Hello 123 abc <># .. asdf" |grep -o "\w+"
STLM011:~ localadmin$ 

STLM011:~ localadmin$ echo "Hello 123 abc <># .. asdf" |grep -oE \w+
STLM011:~ localadmin$ 

STLM011:~ localadmin$ echo "Hello 123 abc <># .. asdf" |grep -oE "\W+"
 
 
 <># .. 
STLM011:~ localadmin$