\{a,b\} regular expressions

sycamorex · 10-18-2008, 08:28 AM

Hi all,
How can I match exactly 2 occurences of zero. It's supposed to be:
0\{2\}, but as you can see it doesn't work as it supposed to. It seems that
it doesn't make any difference between 0\{2,\} - which should find 2 or more occurences of zero and 0\{2\} which is supposed to find exactly 2 occurences of zero.
What am I doing wrong? All the examples find at least 2 zeros. I tried using single quotes (the same results) and no quotes (doesn't work).

Code:

[xtd8865@centbox reg_exp]$ cat wl | grep "0\{2,\}"
100
1000
10000
100000
[xtd8865@centbox reg_exp]$ cat wl | grep "0\{2\}"
100
1000
10000
100000
[xtd8865@centbox reg_exp]$ cat wl | grep "0\{2,3\}"
100
1000
10000
100000

thanks

jschiwal · 10-18-2008, 08:41 AM

I'll use egrep to dispense with the backslashes.

egrep '([^0]|^)00([^0]|$)' wl

This will also match ^00$.

openSauce · 10-18-2008, 08:47 AM

That looks like correct behaviour to me. All those lines do contain substrings which are exactly two zeroes - your regex doesn't say that such a substring can't be followed by more zeroes.

'[^0]00[^0]' will match strings of two zeroes with any character other than 0 before and after them.

What you want is probably '(^|[^0])00($|[^0])', which will also match strings of two zeroes at the beginning or end of a line. But note that to use the '|' character with grep, you need to turn on extended mode by calling egrep or using the -e switch. Or you can still use grep's basic mode, but escape the | with a backslash.

Code:

egrep '(^|[^0])00([^0]|$)' temp.txt
100
100b

sycamorex · 10-18-2008, 09:16 AM

thanks guys for your examples. They work fine.

Quote:

That looks like correct behaviour to me. All those lines do contain substrings which are exactly two zeroes - your regex doesn't say that such a substring can't be followed by more zeroes.

Quote:

\{m,n\} Matches the preceding element at least m and not more than n times (from wikipedia)

man grep:

Quote:

{n} The preceding item is matched exactly n times.
{n,} The preceding item is matched n or more times.
{n,m} The preceding item is matched at least n times, but not more than m times.

In my examples everything works as if it were {n,}.
If they just find a substring n and do not care what follows, then what is the point of introducing 3 syntactic variations of the same thing.

thanks

openSauce · 10-18-2008, 09:24 AM

In other situations, it would make a difference which one you used.

Code:

$ egrep 'ab{2}a' file
abba

$ egrep 'ab{2,}a' file
abba
abbba
abbbba

$ egrep 'ab{2,3}a' file
abba
abbba

sycamorex · 10-18-2008, 09:40 AM

hmmm, it's getting a bit funny.
Is there anything peculiar in my situation that it won't work as it's supposed to?

Code:

[xtd8865@centbox reg_exp]$ cat wl | egrep '0{2,3}'
100
1000
10000
100000
[xtd8865@centbox reg_exp]$ cat wl | egrep '0{2}'
100
1000
10000
100000
[xtd8865@centbox reg_exp]$ cat wl | egrep '0{2,}'
100
1000
10000
100000
[xtd8865@centbox reg_exp]$

It's just a text file containing the following lines:

Code:

ErV · 10-18-2008, 10:36 AM

Quote:

Originally Posted by sycamorex

hmmm, it's getting a bit funny.
Is there anything peculiar in my situation that it won't work as it's supposed to?

Code:

[xtd8865@centbox reg_exp]$ cat wl | egrep '0{2,3}'

If you want to match "00" placed on separate line, try
try egrep '^0{2,3}$'.

If you want with text that ends with "00" use egrep '0{2,3}$'.

If you want to match "00" in a word, try something like
egrep '[^0]+0{2,3}[^0]+'

openSauce · 10-18-2008, 12:42 PM

Quote:

Originally Posted by sycamorex

hmmm, it's getting a bit funny.
Is there anything peculiar in my situation that it won't work as it's supposed to?

But it is working as it's supposed to!

What makes you think 0{2} doesn't match "10000"? It does, in fact it matches it 3 times, in the 2nd, 3rd, and 4th positions.

It might help to realise that:

0{2} is equivalent to 00
0{2,3} is equivalent to 000?, and also equivalent to (00)|(000)
0{2,} is equivalent to 000*

does that make sense now? Do you see why my examples with abbba etc did what you expected, and yours didn't?

Regexes can take a bit of getting used to

sycamorex · 10-18-2008, 01:59 PM

Fair enough. Thanks for the explanation. It has made the thing clearer to me.

jschiwal · 10-18-2008, 06:26 PM

The "00" pattern is in "0000000" so you will have a match. In order to match exactly 2 zeroes, you need to take the neighbors that you allow into account. [] is used to produce a set of characters for a match. If the first character is ^, then the meaning is negated, and you want to match any character not in the set. [^0] means any character but "0". Outside the square brackets, ^ means the beginning of a line. $ means the end of a line. [^0]0{2,3}[^0] matches a non-zero character followed by 2 or 3 zeroes, followed by a non-zero character. That will match 4006 or 220005 but not 10000.

sycamorex · 10-18-2008, 06:38 PM

I get it now. Initially I thought that by writing {2,3} it will automatically ensure that the pattern of 2-3 zeroes will not be followed by any other zero (or another match of 2-3 zeros). You live, you learn.