grep and digits

gawain · 02-19-2008, 10:55 AM

Hi everybody. The problem should be easy to overcome, but as a matter of fact, I'm still failing.

I'd like to match a fixed sequence of numbers, for instance how can I match the "12" without matching all the sequence of numbers?

text text 12 text text

text text 123 text text

text text 1245 text text

I tried like this: egrep "([\b0-9\b]{2})" myfile.txt

([\b0-9\b]{2})
([\b0-9\b]{2}?)
\b[0-9]{2}\b
\b\d{2}\b

but none of them works. But shouldn't the quantifier {2} limit the sequence to two numbers?

Thanks for any hint

0.o · 02-19-2008, 11:01 AM

What output are you trying get? How should it be formatted?

pwc101 · 02-19-2008, 11:14 AM

Code:

grep -w 12 myfile.txt

rupertwh · 02-19-2008, 11:14 AM

Your second to last version should actually work (if what you want to do is match any two-digit number):

Code:

rw@rodion:~/t$ egrep '\b[0-9]{2}\b' myfile.txt
text text 12 text text

Quote:

Originally Posted by gawain

But shouldn't the quantifier {2} limit the sequence to two numbers?

Yes and no. While it will match two digits, additional digits won't keep it from matching the first two it found.

pixellany · 02-19-2008, 11:31 AM

egrep '12$|12[^0-9]' filename

Matches any occurrence of "12" at the end of a line, or not followed by a digit.

gawain · 02-19-2008, 01:32 PM

Thanks for answering.
What I want to match is the sequence of two numbers: in the example I wrote that's 12, but it could be 34, ... The problem is that the regexes I wrote match every sequence of two numbers, even those which are made up of four numbers

rupertwh at the moment I cannot have a shell ... at disposal: maybe my regex didn't work because I used double quotation marks instead of the 'single ones'?

Thanks again for the help

pixellany · 02-19-2008, 02:34 PM

egrep -o '\b[0-9]{2}\b' num

matches any combo of 2 numbers within a word boundary. The -o option makes it return only the matched patterns, rather than the whole line.

If the 2-digit pattern can be embedded in other characters, then you need something similar to what I showed earlier.

eg:
[^0-9][0-9]{2}[^0-9] (extended regex syntax)
matches any two digits between non-digits. Does not work if the two digits are at beginning or end of the line.

gawain · 02-20-2008, 02:13 AM

Thanks a lot, Pixellany