multiple sed's

[Igaduma]

Hello Guru's

I read over a file many times to change certain patterns with another, predefined pattern, in this case, an empty "".
Really just removing things I don't need.
But, it involves reading the file in over & over again applying sed to a certain pattern.
Somehow I think there must be a way to do grab many patterns and all replace them with an "" in a more intelligent manner.

So, this is what the file might look like:

$cat input_file
one two three four five six
two three four five seven

If I now have to remove "one" and "seven" from the source file, this is how I do it quick & dirty:

$sed 's/one//g' input_file > output_file
$sed 's/seven//g' output_file > output2_file

Obviously this works, but it involves scanning a file 2 times.
How can one define a range of patterns to be all replaced by a fixed pattern (in this case "") in 'just one go' over the file ?

Thanks,
Iga

 

[feherke]

Hi

Whis is what you want ?

sed 's/one//g;s/seven//' input_file > output_file
[gray]# or[/gray]
sed 's/\(one\|seven\)//' input_file > output_file
[gray]# or[/gray]
sed -r 's/(one|seven)//' input_file > output_file

Note, that if there are more commands, is better to put them into a [tt]sed[/tt] script file, then execute it.

Feherke.

http://rootshell.be/~feherke/

 

[PHV]

Bottom line:
man sed

Hope This Helps, PH.
Want to get great answers to your Tek-Tips questions? Have a look at FAQ219-2884 or FAQ181-2886

 

[Igaduma]

ok, I was a little too quick to
"post it, forget it and receive reply and answers"

But, yes, I must read the man page for sed.
\personal note:RTFM

Thanks for making me humble again

 

[stefanwagner]

just combine multiple commands with semicolons:

sed 's/one//g;s/seven//g' input_file > output_file

seeking a job as java-programmer in Berlin:

http://home.arcor.de/hirnstrom/bewerbung