How do I fetch only numbers in grep?
Solution 1
You can use grep -E
to access the extended regular expression syntax( Same as egrep)
I have created a testfile with below contents:
>cat testfile
this is some text
with some random lines
again some text
ok now going for numbers (:32)
ok now going for numbers (:12)
ok now going for numbers (:132)
ok now going for numbers (:1324)
Now to grep the numbers alone from the text you can use
>grep -Eo '[0-9]{1,4}' testfile
32
12
132
1324
will be output.
Here "-o" is used to only output the matching segment of the line, rather than the full contents of the line.
The squiggly brackets (e.g. { and }) indicate the number of instances of the match. {1,4} requires that the previous character or character class must occur at least once, but no more than four times.
Hope this helps
Solution 2
You can use RE bracket expression [:digit:]
specified by section 9.3.5 of POSIX standard , in combination with -o
flag to print only matching "words"
$ grep -o '[[:digit:]]*' <<< $'No number in this line\nbut 123 here'
123
Solution 3
You can use Perl style regular expressions as well
grep -Po "\\d+" filename
-P Interpret PATTERNS as Perl-compatible regular expressions (PCREs).
-o Print only the matched (non-empty) parts of a matching line, with each such part on a separate output line.
Solution 4
grep -o
will print only the matching part of the line. Otherwise grep will print any lines with the pattern.
Solution 5
I would use curl to access your file locally or remotely, then I would grep lines with numbers wrapped in (: ) then cut those pieces off and write to file
the accepted answer ignores there could be numbers in the previous lines of the file, it does work for the example data, but what if the file was remote?
Local
curl file:///home/$USER/Public/input.txt | grep -o '(:.*)' | cut -d ":" -f 2 | cut -d ")" -f 1 > output.txt
In this example output.txt
in your current folder will be written over, we are accessing input.txt
from your Public folder.
Remote
curl https://yoursite.com/Public/input.txt | grep -o '(:.*)' | cut -d ":" -f 2 | cut -d ")" -f 1 > output.txt
In this example output.txt
in your current folder will be written over, we are accessing input.txt
from https://yoursite.com/Public/
.
Ant's
Updated on September 18, 2022Comments
-
Ant's over 1 year
I have file like this :
other lines . . . blah blah blah (:34)
I wish to find the occurrence of numbers in the above file. I came up with:
grep [0-9] filename
But that is printing the whole:
blah blah blah (:34)
Rather I want only
34
. Is there any way to do so?-
Admin over 11 yearsIn the future, also check out the man page for grep (or any other program). The man page details the options required for many common uses of the program. e.g.
man grep
-
-
FractalSpace almost 10 yearsNice. Also, to match more than 4 or arbitrary number of digits, use
grep -Eo '[0-9]{1,}' testfile
-
PerlDuck almost 6 yearsWhy use
curl
when a simplecat
would do? -
Stef almost 6 yearsThe question doesn't clearly state that the file is local. this answer allows for both.
-
alec over 3 yearsdeserves way more than 2 upvotes :)
-
Karl Xu about 3 years@FractalSpace when have the constrain of repeating at least one time in
[0-9]{1,}
, the-o
option might not necessary, since repeating at least one time, by definition remove the empty output? -
FractalSpace about 3 yearsIt's been a while, so I have no recollection of what I wrote and why :)
-
Jacob over 2 years@Stef it's inconsequential where the file is or if the input is even a file. This answer has a lot of cruft that just adds confusion.
-
spuder over 2 yearsNote: You may not have
-P
option if runningBSD
grep (e.g. OSX. )