grep for a variable with boundary conditions

ip_address · 02-15-2013, 05:45 AM

Hello everyone,

How could I grep for a variable curr_line with boundary conditions? Thanks!

# variable named curr_line

Code:

curr_line=$(sed -n "$counter"p "$file")

# want to grep for exact match with boundary conditions

Code:

grep -w "^$curr_line$" "$index_file"

gnashley · 02-15-2013, 05:52 AM

Is what you have there not working?

ip_address · 02-15-2013, 05:54 AM

Code:

grep -w "^$curr_line$" "$index_file"

doesn't work

firstfire · 02-15-2013, 06:45 AM

Hi.

How about -x flag (man grep):

Quote:

-x, --line-regexp
Select only those matches that exactly match the whole line. (-x is specified by POSIX.)

Beware however that $curr_line may contain special symbols such as `.' (any character in regular expressions syntax) which may lead to misbehaviour.

David the H. · 02-17-2013, 05:51 PM

"Doesn't work" tells us nothing. In what way doesn't it work? Do you get any errors? Output? Have you tried any variations? What exactly did you run, and what exactly did you get?

Also, we really need to see an example of the input, including the line you want to match and at least a few of the surrounding ones.

In short, the more information you provide, the more likely you are to get real help.

gnashley · 02-18-2013, 02:09 AM

I remember having problems with such a construct: "^$curr_line$", but using "^$curr_line"$ worked.

mina86 · 02-18-2013, 06:11 AM

Quote:

Originally Posted by ip_address

Code:

curr_line=$(sed -n "$counter"p "$file")

Please don't tell my you are using a counter and a repeated invocation of sed to read a file line by line… If you do, you just changed your script to be O(n²) instead of O(n). Consider instead:

Code:

while read curr_line; do
    # …
done <$file

Quote:

Originally Posted by ip_address

Code:

grep -w "^$curr_line$" "$index_file"

Second dollar sign should be escaped with a slash, ie. "^$curr_line\$".

Finally, in addition to -x you may also be interested in -F switch.