topic Re: Replace string in place..retain original line. in Operating System - Linux

Replace string in place..retain original line.

OFC_EDM — Mon, 21 Apr 2008 08:45:47 GMT

This question stems from a solution I got from James R Ferguson. Thanks James!

Where the following command replaces a string and backs up the file before the update.

perl -pi.old -e 's/\bDISABLETIME=\d\d\b/DISABLETIME=60/' ./login

I want to customize this further:

In the login file the line may be
#DISABLETIME=20
#DISABLETIME=100
#DISABLETIME=2000

If I find any one or other variations I want to leave the commented out line.

Then add a new line
DISABLETIME=60
after the original line.

Thereby leaving the original value in place in case I need to go back to it.

So it may look like:
#DISABLETIME=200
DISABLETIME=60
After the update is done.

How can the perl command be modified to do this?

Regards

Re: Replace string in place..retain original line.

OFC_EDM — Mon, 21 Apr 2008 09:01:54 GMT

I figured out part of this

perl -pi.old -e 's/\bDISABLETIME=[0-9]*\b/\nDISABLETIME=60/g' ./login.test

Will put in the newline and my new value of DISABLETIME=60.

But it won't retain the original line.

Maybe there's a way to reference the original string in the replacement string?

Re: Replace string in place..retain original line.

Dennis Handly — Mon, 21 Apr 2008 09:13:50 GMT

>In the login file the line may be

These lines start with "#"?

>If I find any one or other variations I want to leave the commented out line.
>Thereby leaving the original value in place in case I need to go back to it.

You need to provide more details. Are the lines already commended out or do you want to comment them out, then add the "=60" line?
Do you only have the one?

You might be able to use sed but it is easier to use awk:
$ awk '
/DISABLETIME=/ {
print $0
print "DISABLETIME=60"
}
{
print $0
}' login > login.new

Re: Replace string in place..retain original line.

OFC_EDM — Mon, 21 Apr 2008 09:24:38 GMT

I'm updating over 100 servers.

Some already have the DISABLETIME=60 in place and some won't.

Some will already have it commented out. And I won't know the value after the = sign.

So I have to handle both those conditions:
Regardless I want the original line to end up being commented out.

And append the new DISABLETIME=60 line immediately after the original

I want to try and stick with the perl statement as it automatically backs up the file.

Another reason is I'm trying to convert myself to Perl :)

Re: Replace string in place..retain original line.

James R. Ferguson — Mon, 21 Apr 2008 11:01:34 GMT

Hi:

# perl -ni.old -e 'print;if m/#\s*DISABLETIME\s*=\s*\d+/) {print "DISABLETIME=60\n"}' file

...which would handle patterns like:

#DISABLETIME=20

# DISABLETIME = 2000

Regards!

...JRF...

Re: Replace string in place..retain original line.

OFC_EDM — Mon, 21 Apr 2008 11:10:12 GMT

Thanks James,

Did I put the code in incorrectly? I get the error below.

perl -ni.old -e 'print;if m/#\s*DISABLETIME\s*=\s*\d+/) {print "DISABLETIME=60\n"}' ./login.test

syntax error at -e line 1, near "if m/#\s*DISABLETIME\s*=\s*\d+/"
syntax error at -e line 1, near ";}"
Execution of -e aborted due to compilation errors.

Would you mind explaining how the command works?

I think the \s* matches all white space from the next to the end of the line (before newline character).

And \d+ is one or more digits.

Whats the m at the beginning?

Am I interpreting the command correctly as follows?:
print the line, match the string, apply condition and then print if condition met?

Thanks again

Re: Replace string in place..retain original line.

James R. Ferguson — Mon, 21 Apr 2008 11:31:55 GMT

Hi (again):

Sorry, mMy error; I dropped the opening parenthesis:

# perl -ni.old -e 'print;if (m/#\s*DISABLETIME\s*=\s*\d+/) {print "DISABLETIME=60\n"}' ./login.test

The '\s*' means zero or more whitespace (blank, tab, newline and carriage-return) characters.

The '\d+' means one or more digits. Note the '+' versus the '*' for one-or-more versus zero-or-more.

The 'm' in this context is optional, but means "match". With a forward slash '/' as a delimiter, 'm' is inferred. Other delimiters can be used for clarity and that's when the leading 'm' is really needed. I could have written:

# perl -ni.old -e 'print;if (m{#\s*DISABLETIME\s*=\s*\d+}) {print "DISABLETIME=60\n"}' ./login.test

...that is, used '{' and '}' as the matching delimiters. This is very useful when you want to avoid the "leaning toothpick syndrome", like:

# if ( m/\/usr\/bin\/sh/ ) ...

...becomes:

# if ( m{/usr/bin/sh} ) ...

Regards!

...JRF...

Re: Replace string in place..retain original line.

OFC_EDM — Tue, 22 Apr 2008 09:48:51 GMT

James,

I've hit a snag where some login files for some systems have previous commented out values.

Which results in multiple line matches.

I just want to insert my DISABLETIME=60 after the LAST match

Example:

# DISABLETIME=20
# DISABLETIME=100
DISABLETIME=60 <---Insert here

Unfortunately my code does this:
# DISABLETIME=20
DISABLETIME=60 <---Inserts here
# DISABLETIME=100
DISABLETIME=60 <---and Inserts here

I run the following:
# Find any lines uncommented and comment out
perl -pi.old -e 's/^DISABLETIME=\s*/#DISABLETIME=/g' ${dwnld_dir}/login.${TARGET}

# Find LAST commented line and insert new line
perl -ni -e 'print;if (m/#\s*DISABLETIME\s*=\s*\d+/) {print "DISABLETIME=60\n"}' ${dwnld_dir}/login.${TARGET}

Any ideas on how to improve the code?

Re: Replace string in place..retain original line.

James R. Ferguson — Tue, 22 Apr 2008 11:52:20 GMT

Hi (again):

Ok, given a configuration file like this:

# cat ./myconfig
this is a some config file...
# DISABLETIME=20
# DISABLETIME= 200
#DISABLETIME=1000
DISABLETIME=60
this is more of the file
and this is still more of the file

...Use this script:

# cat ./myfilter
#!/usr/bin/perl -i.old
use strict;
use warnings;
my $lines = do { local $/ = undef; <> };
$lines =~ s/( \#\s*DISABLETIME\s*=\d+.+?$ ) (?! \s*\#\s*DISABLETIME ) /$1\nDISABLETIME=60/smx;
print $lines;
1;

...and thus:

# ./myfilter ./myconfig

...yields a modified './myconfig' [with a backup as '*.old'] of:

this is a some config file...
# DISABLETIME=20
# DISABLETIME= 200
#DISABLETIME=1000
DISABLETIME=60
this is more of the file
and this is still more of the file

Regards!

...JRF...

Re: Replace string in place..retain original line.

OFC_EDM — Tue, 22 Apr 2008 12:19:58 GMT

Thanks (again) James.

But I think I have a perl issue.
I try to run the code and get this.
I assume it means I don't have a module called warnings.pm on the system.
And I can't update the perl install there.

=>./myfilter ./retrievedfiles/login.iecs
Can't locate warnings.pm in @INC (@INC contains: /usr/perl5/5.00503/sun4-solaris /usr/perl5/5.00503 /usr/perl5/site_perl/5.005/sun4-solaris /usr/perl5/site_perl/5.005 .) at ./myfilter line 3.
BEGIN failed--compilation aborted at ./myfilter line 3.

Re: Replace string in place..retain original line.

OFC_EDM — Tue, 22 Apr 2008 12:21:18 GMT

I took out the

use warnings;

and it works

Re: Replace string in place..retain original line.

James R. Ferguson — Tue, 22 Apr 2008 12:36:55 GMT

Hi (again):

> But I think I have a perl issue...

> Can't locate warnings.pm in @INC (@INC contains: /usr/perl5/5.00503/sun4-solaris /usr/perl5/5.00503 ...

Yup, you REALLY NEED a newer version of Perl. The 'warnings' pragma appeared in the core of 5.8.

Perl 5.8.8 would be excellent and Perl 5.10 has been out for a few months.

In the meantime, if you don't have the 'warnings' pragma, use the '-w' switch. It is not a robust but is far, far better than nothing!

# perl -w

Regards!

...JRF...

Re: Replace string in place..retain original line.

OFC_EDM — Tue, 22 Apr 2008 13:05:00 GMT

Thanks James.

You must have got a 1000 points from me this week.

Hope I can return the favour one day!

Cheers

Re: Replace string in place..retain original line.

OFC_EDM — Tue, 22 Apr 2008 13:05:45 GMT

See above for solution