yet another sed thread - find and replace

[Adunaic]

I have several nagios config files that are looking quite ugly as the tabbing/spacing is all over the place making it hard to read.

My first thought was to have sed remove any groupings of two or more spaces and replace them with a tab ('\t'). Most lines are then:

<tab> string <tab> string

However the strings are of varying length so one or two tabes are needed. Initially I singled out the tabs one by one but there are too many. My current sed script is:

s/ \+ /\t/g
s/\t\+/\t/g
s/\tuse/\tuse\t/g
s/\talias/\talias\t/g
s/\taddress/\taddress\t/g
s/\tmembers/\tmembers\t/g
s/\tname/\tname\t/g
s/\tparents/\tparents\t/g

Is there any way in sed to do something like:

if first string is greater than X characters long put one tab after it, if first string is greater than Y characters long put two tabs after it and so on?

[DBelton]

have you looked at the unexpand command? That is what it is made for

man unexpand

[Adunaic]

That cannot count a string though to decide how many tabs to insert

[DBelton]

actually it does insert multiple tabs based on 8 spaces per tab. You can change that with the --tabs= parameter if you wish.

[Adunaic]

Still not sure that it does what I want (bit confused after reading about sed).

I have however figured out how to get sed to do what I want:

#Remove all groups of 2 or more spaces and replace with a single tab
s/ \+ /\t/g
#Remove all groups of 2 or more tabs and replace with a single tab
s/\t\+/\t/g
#Find any match to tab string tab where the string is 1 - 7 chars long and add two tabs to the end.
s/\(^\t[[:graph:]]\{1,7\}\t\)/\1\t\t/g
#Find any match to tab string tab where the string is 8 - 17 chars long and add one tab to the end.
s/\(^\t[[:graph:]]\{8,17\}\t\)/\1\t/g