CODEWITHSUNDEEP
regexbashshellsed
intumwa
intumwa

Reputation: 354

sed replace anything after nth occurence

I am new in Bash scripting and have been searching for a proper answer to how can I use sed to replace anything after the nth occurence (not to replace the nth occurence).

For example, if I want to change anything that comes after the 2nd space, I'd have the following sentence as input: Today is a good day and the following sentence as result: Today is a friday

Any ideas?

Upvotes: 4

Views: 2754

Answers (4)

Jotne
Jotne

Reputation: 41460

Why not just use awk

echo "Today is a good day" | awk '{print $1,$2,"a friday"}'
Today is a friday

This will keep the first two fields, and replace the rest of the line.

Upvotes: 1

Charles Hethcoat
Charles Hethcoat

Reputation: 1

As I understand the question being asked, this doesn't really even need sed or awk, if you're using Unix 7th Edition shell or a newer shell based on it like sh or bash. Just use an unfortunately obscure feature of the set command:

$ set Today is a good day
$ echo $1 $2 "Friday"
Today is Friday

The original question asked about patterns, but the example offered has the pattern being just a single space. The second occurrence of the space implicitly follows the second argument. You can adapt it to "nth occurrence" using a little shell arithmetic, I imagine.

This set trick has been around a long time, even though I just recently discovered it. An example of its use appears on page 136 of Kernighan and Pike's "The UNIX Programming Environment" (1984). It works the same on Bash, and I would guess it must be present on any 7th Ed. shell derivative.

Upvotes: 0

anubhava
anubhava

Reputation: 786289

You can use:

s='Today is a good day'
echo "$s" | sed 's/^\(\([^[:space:]]\+[[:space:]]\+\)\{3\}\)[^[:space:]]\+[[:space:]]\+/\1fri/'
Today is a friday

Simplified using -r:

cho "$s" | sed -r 's/^((\S+\s+){3})\S+\s+/\1fri/'
Today is a friday

Upvotes: 2

Wintermute
Wintermute

Reputation: 44073

I would say

 echo 'Today is a good day.' | sed 's/ /&\n/3; s/\n.*/friday/'

This consists of two commmands:

s/ /&\n/3       # inserts a \n after the third space
s/\n.*/friday/  # replaces \n and everything that comes after it with "friday"

This uses the fact that \n will never appear in a line -- something that makes \n great as a marker in sed scripts. & refers to the matched portion of the line.

Upvotes: 0

Related Questions