insert zero before single numbers using sed (or awk)

chess · 10-21-2008, 11:05 AM

I have been banging my head and googling for about a week trying various things and cannot come up with a solution. So, I would really appreciate any assistance.

I need to add a zero before a single number surrounded by period. For example:

10.4.256.14.3 would need to be converted to 10.04.256.14.03
3.4 would need to be converted to 3.04

I cannot just add a zero before all numbers, only the 'single' digits surrounded by periods.

Here is what I have so far:

Code:

#!/bin/bash
ONE=10.4.256.14.3
TWO=$(echo $ONE | sed -e 's/\.\([0-9]\)\./\.0\1\./g')
echo $TWO

The output is:

10.04.236.14.3

It works for the first single number, the '4' but not for the second single number, the '3' at the end. I know the problem is the use of the \1 but I am using that because I need to 'keep' the number matched by [0-9] so I can stick it back in after inserting the 0. I hope this makes sense.

Thanks for the help.

pwc101 · 10-21-2008, 11:14 AM

Assuming the number of characters remains constant in each "column":

Code:

echo 10.4.256.14.3 | awk -F"." '{printf "%2i.%02i.%3i.%2i.%02i\n", $1,$2,$3,$4,$5}'

I imagine colucix or billymayday will come along with a much more elegant solution

chess · 10-21-2008, 11:19 AM

Yeah, unfortunately, they don't. That's why I used the two examples, one with 5 'columns' and one with 2. There is no pattern to either the number of columns or the number of digits in any particular column.

AuroraCA · 10-21-2008, 11:30 AM

In sed and awk '^' indicates beginning of line and '$' indicatesend of line are recognized.

The following would replace your end of line condition.

Code:

#!/bin/bash
ONE=10.4.256.14.3
TWO=$(echo $ONE | sed -e {
's/\.\([0-9]\)\./\.0\1\./g')
's/\.\([0-9]\)$/\.0\1\')
}
echo $TWO

Note: there is no g after the second replacement statement since there can only be one eol condition per line.

forrestt · 10-21-2008, 11:45 AM

I think you meant:

Code:

#!/bin/bash
ONE=10.4.256.14.3
TWO=$(echo $ONE | sed -e 's/\.\([0-9]\)\./\.0\1\./g' -e 's/\.\([0-9]\)$/\.0\1/')
echo $TWO

HTH

Forrest

AuroraCA · 10-21-2008, 11:50 AM

Sorry if I mislead you. I'm use to doing substitutions from a file.

Thanks for the correction.

keefaz · 10-21-2008, 12:03 PM

Maybe with perl:

Code:

$ echo 10.4.256.14.3 | perl -pe  's/\.(\d[^0-9])/.0$1/g'
10.04.256.14.03
$ echo 3.4 | perl -pe  's/\.(\d[^0-9])/.0$1/g'
3.04

Kenhelm · 10-21-2008, 12:03 PM

This uses GNU sed.
\b represents a word boundary.

Code:

echo 10.4.256.14.3 | sed 's/\.\([0-9]\)\b/.0\1/g'
10.04.256.14.03

chess · 10-21-2008, 01:16 PM

Thanks, everybody, for the excellent help. Several of these solutions seem to work just fine. I'm testing out Kenhelm's solution in my script and so far so good.

Thanks, again!

syg00 · 10-21-2008, 04:53 PM

Nice answer - does what was asked for. Small nit - does the OP really want ".0." converted to ".00." ???
Maybe, maybe doesn't care.

Easy to fix.

chess · 10-22-2008, 08:06 AM

Quote:

Originally Posted by syg00

Nice answer - does what was asked for. Small nit - does the OP really want ".0." converted to ".00." ???
Maybe, maybe doesn't care.

Easy to fix.

No, in this case it does not matter if .0. is converted to .00. -- but point well taken, sometimes it matters a great deal. :-)

Thanks again for all the input and suggestions. It's really appreciated.