Very Computer

I have a file containing lines like this:

include/SessionDefs.h,                             1.118,,,
src/screens/apply_joint.trn,                1.7,
src/screens/custapp_appn_review.obj,        1.30,,,,,,,

I would like to get rid of all the trailing commas after the number.
Otherwise,the tabs etc all can stay. Please notr that I have to keep
the comma after the text.

Is there a sed command that would do this? any advice is appreciated.

The way to think about this is this: you need to match a digit followed by
any number of commas, and replace with the original matching digit.
To match the pattern you need is '[0-9],*'
To use the matched digit in the replacement, you need to mark it,
like this: '\([0-9]\),*'.  Then you can use '\1' in the replacement
to get back the digit which matched.
Putting this together, your sed command should be:

sed -e 's/\([0-9]\),*/\1/'

If the pattern you want to replace may occur more than once per line,
then you need the 'g' suffix:

sed -e 's/\([0-9]\),*/\1/g'

Quote:> I have a file containing lines like this:

> include/SessionDefs.h,                             1.118,,,
> src/screens/apply_joint.trn,                1.7,
> src/screens/custapp_appn_review.obj,        1.30,,,,,,,

> I would like to get rid of all the trailing commas after the number.
> Otherwise,the tabs etc all can stay. Please notr that I have to keep
> the comma after the text.

> Is there a sed command that would do this? any advice is appreciated.

daniele> cat data
include/SessionDefs.h,                             1.118,,,
src/screens/apply_joint.trn,                1.7,
src/screens/custapp_appn_review.obj,        1.30,,,,,,,

daniele> sed 's/,\{1,\}$//' data
include/SessionDefs.h,                             1.118
src/screens/apply_joint.trn,                1.7
src/screens/custapp_appn_review.obj,        1.30

daniele>

hi!

Key insight: ',*$'

--

Linux solution for data management and processing.

Yes, so simple I wonder why you didn't just say it,

  sed 's/,*$//' filename

--
David Thompson

Because I'm tired of doing other people's homework.  The best anyone can
do is to point them in the right direction.  They have to do the
walking, I'm afraid.

--

Linux solution for data management and processing.

Quote:>I have a file containing lines like this:

>include/SessionDefs.h,                             1.118,,,
>src/screens/apply_joint.trn,                1.7,
>src/screens/custapp_appn_review.obj,        1.30,,,,,,,

>I would like to get rid of all the trailing commas after the number.
>Otherwise,the tabs etc all can stay. Please notr that I have to keep
>the comma after the text.

>Is there a sed command that would do this? any advice is appreciated.

Hi Sankar,

You can do this in 'sed' with a single 's' substitute command,

s/,*$//

,*   match 0 or more comma characters
$    followed by the end of the line

You could look for 1 or more comma characters but it makes the expression
bulkier;

s/,\{1,\}$//

of course there is no difference because the original code replaces sometimes
nothing at the end of a line (0 comma characters) with nothin' which had no
effect ( possibly the first command is a little slower thou`)

( Both of these substitutions assume that simply removing any trailing commas
  from a line is good enough if not just extend the expression to suit )

btw: You can find complete documentation for 'sed' from the original authors
on my web-site, here;

http://lf.1accesshost.com/UNIX/UNIX.HTM

byefornow
laura

--
alt.fan.madonna |news, interviews, discussion, writings
                |chat, exchange merchandise, meet fans....
                |Get into the groove baby you've got to... check us out!

On 18 Mar 2003 06:40:20 -0800, Sankar
sed 's/,*$//'

--
Ever wonder if taxation without representation might have been cheaper?