Grep extended regular expressions

tt1187 · 01-20-2014, 03:30 PM

Hello, I have two questions. I apologise if I haven't explained this properly, this is for learning/understanding how regular expressions work.

Could you explain this please;

Quote:

grep "([^()]*)a" file returns any line containing a pair of parentheses that are innermost and are followed by the letter "a". So it matches these lines ;

(hello)a
(aksjdhaksj d ka)a

But not this
x=(y+2(x+1))a

From <http://www.panix.com/~elflord/unix/grep.html>

1. ([^()]*) - My understanding is; this is a paranthesis() around a subexpression. [] brackets matches any one character inside it, which are "()", the ^ is not a character, but a carat to indicate that "(" or ")" have to be at the beginning of the line.

The problem is, it will return (hello), (hi), but will not return )hello, )hello(. The question is, why does it have to be in that order, and why do both have to be included [ie. is my understanding of brackets wrong]?

2. What is the * for (I understand it means match none or any, so test* will match test, testing, tested, tester)? I mean, the expression matches anything inside () that ends with an "a" on the outside - so (testing)a, (whats up)abz, (kk)a. So what's the purpose of the *, and why it is outside the brackets, but inside the paranthesis subexpression?

Thanks for any explanation.

grail · 01-20-2014, 07:26 PM

Ok ... answers are:

1. As we are NOT using extended matching with grep (ie. you need to either use egrep or supply -E), the brackets are literal, so if we start left to right:

Code:

( - - match a left round bracket anywhere in the line

[^()]* - - zero or more characters that are not (^ here means not as opposed to the start of the line, it has this special meaning only when used inside [])

) - - match a right round bracket

a - - match the literal character a

So looking at the example that does not match - - x=(y+2(x+1))a - - the first left round bracket does not encounter its matching right round bracket before matching another left round bracket which
would need to immediately be followed by a right round bracket and an 'a' so grep would move to the next left round bracket to look for a match. And the end of its search it finds a right round bracket
followed by another right round bracket, ie not an 'a', hence no match on the entire line

2. when globbing, ie which is what the bash shell uses for matching, test* will match testing, however when used in a regular expression, the * means zero or more of the preceding character,
so test* will match - - tes, testt, testtttttt, etc

Hope that helps