Shell Programming and Scripting

Hi all, I am trying to remove quite a lot of numbers for a file I have which looks something along the lines of;

Code:

1,2,3,4,5,6 etc

I have a list of numbers I want removing looking like;

Code:

10000
10987
16572
etc
etc

and have been trying to run;

Code:

for id in `cat list` ; do sed -i -e 's/$id,//g' customers.cfg ; done

However when I come to check the original file nothing has changed. I can run;

Code:

for id in `cat list` ; do echo $id ; done

And that works fine, so it looks like my "sed" syntax is incorrect somewhere down the line.

Can anyone point me in the right direction of getting this working?

Thanks!

Code:

sed -i -e 's/$id,//g'

$id will not be substituted because it is inside the single backticks of sed. Either use " instead or, use additional backticks to separate the variable to the shell like

Code:

sed -i -e 's/'${id}',//g'

Also if your grep has -f you can use a list as input and if it has -v you can exclude, with -w it treats them as words so that it will look like

Code:

grep -fwv list customers.cfg

You would not need the loops etc. around. You can redirect it to a tmp file and mv it to the old name.

Thanks for the prompt reply Zaxxon!

I have amended my run line to look like

Code:

for id in `cat list` ; do sed -i -e 's/'${id}',//g' customers.cfg ; done

However the same happens; the line runs with no problem however nothing is edited within the file.

Also when I try

Code:

grep -fwv

I get

Code:

grep: wv: No such file or directory

Could the problem be the numbers within my list file?

At the moment when I

Code:

cat list

I get a long line of numbers similar to;

Code:

1
2
3
4
5
etc
etc

Would that affect the outcome of the run line?

Even if you got this to work, will it give you the correct result? In your list there are the numbers "1", "2", and "3". It seems that the intent with sed is to match "1,", "2,", and "3,", respectively. So, if the data is "1,2,3,101,102,103,201,202,203,500", the result would be:

After s/1,//g: 2,3,10102,103,20202,203,500
After s/2,//g: 3,1010103,2020203,500
After s/3,//g: 101010202020500

That is what your commands will accomplish. Is this the desired result?

Or should the result after handling "1", "2", an "3" in the list be the removal of the first three digits in the sequence, with the final output being "101,102,103,201,202,203,500" ?

In short, should the items in the list be allowed to match partial numbers in the data? If not, your approach requires modification, because that sed command allows partial matches.

Regards,
Alister

Hi Alister, thanks for the reply. You are right the command would not produce the correct outcome.

I assume the right way (if it worked at all) would be;

Code:

for id in `cat list` ; do sed -i -e 's/,'${id}',//g' customers.cfg ; done

So it actually searches for the number surrounded by 2x commas instead. Is that correct? However that wouldnt pick up the first number in the sequence would it? e.g

Code:

,1,

wouldnt work in the sequence of;

Code:

1,2,3,4

Is there a way around that?

Thanks again

Maybe

Code:

's/[^,]'${id}',//g'

which should mean either begin of line or a comma
(not sure if this will work...)

Actually, that means not-comma followed by the number. It wouldn't help, as that would match a digit and still allow partial matches.

Regards,
Alister