[SOLVED] How to process different version string using sed/cut

robertjinx · 06-01-2016, 02:47 AM

Hello, I'm trying to get the major/minor version from a versions string, but can have different version strings.

For example:

Code:

version="124.0.0.56" 
 or 
version="123.0(123)"

Getting major its kind of easy:

Code:

echo $version | cut -d. -f1

but getting minor which would mean '56' or '123' is difficult or I haven't found a way to do it with a single sed.

I currently use this:

Code:

version="124.0.0.56"
echo "$version" | sed -n 's/\([0-9].*\)\.\([0-9].*\)\.\([0-9].*\)\.\([0-9].*\)/\4/p'

version="123.0(123)"
echo $version | sed -n 's/\([0-9].*\)\.\([0-9].*\)(\([0-9].*\))/\3/p'

Is there a way to unify this 2 sed's into one?

robertjinx · 06-01-2016, 02:59 AM

Sorry, actually there are 3 possible version strings:

Code:

version="124.0.0.56"
 or
version="121.0.0.0.31" 
 or 
version="123.0(123)"

syg00 · 06-01-2016, 03:16 AM

You can use (multiple) -e to run multiple tests on the same record. And yes you could probably scan all in one regex, but it'd be complex to test. Is the time that would be invested worth it ?.

pan64 · 06-01-2016, 04:46 AM

I would split on non-digit characters:

Code:

V=( $(echo $version | sed -r 's/[^0-9]+/ /g') )
echo ${V[0]}

HMW · 06-01-2016, 04:49 AM

My thinking is that it is easier to replace all '.' and '()' chars with whitespace so that we get columns. The last column is the one we want.

Code:

echo "version=\"123.0(123)\"" | sed -r 's/(\.|\(|\))/ /g' | sed -r 's/("$| "$)//' | awk '{ print $NF }'
123

echo "version=\"124.0.0.56\"" | sed -r 's/(\.|\(|\))/ /g' | sed -r 's/("$| "$)//' | awk '{ print $NF }'
56

echo "version=\"121.0.0.0.31\"" | sed -r 's/(\.|\(|\))/ /g' | sed -r 's/("$| "$)//' | awk '{ print $NF }'
31

So, this:

Code:

sed -r 's/(\.|\(|\))/ /g' | sed -r 's/("$| "$)//' | awk '{ print $NF }'

works for all the test cases. Can probably be done better/more efficient.

Best regards,
HMW

HMW · 06-01-2016, 04:51 AM

Quote:

Originally Posted by pan64

I would split on non-digit characters:

Code:

V=( $(echo $version | sed -r 's/[^0-9]+/ /g') )
echo ${V[0]}

^This is great and much better than my approach.

robertjinx · 06-01-2016, 04:58 AM

Yes, all of the ideas are good and thanks for that, but this seems to work:

Code:

$ echo "122.0.0.4" | sed -n -e 's/\([0-9].*\)\.\([0-9].*\)\.\([0-9].*\)\.\([0-9].*\)/\4/p' -e 's/\([0-9].*\)\.\([0-9].*\)(\([0-9].*\))/\3/p'
4
$ echo "122.0.0.0.4" | sed -n -e 's/\([0-9].*\)\.\([0-9].*\)\.\([0-9].*\)\.\([0-9].*\)/\4/p' -e 's/\([0-9].*\)\.\([0-9].*\)(\([0-9].*\))/\3/p'
4
$ echo "121.0(24)" | sed -n -e 's/\([0-9].*\)\.\([0-9].*\)\.\([0-9].*\)\.\([0-9].*\)/\4/p' -e 's/\([0-9].*\)\.\([0-9].*\)(\([0-9].*\))/\3/p'
24

Thanks again guys!

HMW · 06-01-2016, 05:03 AM

Actually, I don't know why I didn't combine my two seds into one, like this:

Code:

echo "version="123.0(123)"" | sed -r 's/(\.|\(|\)|")/ /g' | awk '{ print $NF }'
123

So, this is a lot better:

Code:

sed -r 's/(\.|$|$|")/ /g' | awk '{ print $NF }'

But Pan64's approach is still the best!
HMW

robertjinx · 06-01-2016, 05:35 AM

Quote:

Originally Posted by HMW

Actually, I don't know why I didn't combine my two seds into one, like this:

Code:

echo "version="123.0(123)"" | sed -r 's/(\.|\(|\)|")/ /g' | awk '{ print $NF }'
123

So, this is a lot better:

Code:

sed -r 's/(\.|$|$|")/ /g' | awk '{ print $NF }'

But Pan64's approach is still the best!
HMW

Yep, you got a point.

pan64 · 06-01-2016, 05:52 AM

Quote:

Originally Posted by HMW

Code:

sed -r 's/(\.|\(|\)|")/ /g' | awk '{ print $NF }'

If awk, do not add sed:

Code:

awk ' BEGIN { FS="[^0-9]*" } { print $NF }'