Show only N matches at end of file

[jungle-boogie]

Hello,

I would like to know if I can use rg to show the last N lines of a match.
Effectively something like

rg 'ripgrep is great' myfile.log | tail

I'm familiar with options -A, -B, and -C but I don't want ripgrep is great printed hundreds of times, only the last N times.

Is there an option with rg to achieve this?

Thanks!

[jungle-boogie]

Why do I want this?

2466 matches
2466 matched lines
1 files contained matches
1 files searched
277011 bytes printed
5170124 bytes searched
2.181747 seconds spent searching
2.186922 seconds

I don't care about 2455 of those matches; I only want to know the status of the last ten lines in my file.

[BurntSushi]

There is no way, no.

To answer the next natural question: no, I don't think I will add it. In order to know if a match is part of the last N matches, you have to know where the last match is. In order to know where the last match is, you need to search the entire input. Thus, by necessity, implementing this feature requires buffering the last N matches in memory and only printing them once the search has completed. ripgrep isn't really set up to do that.

Instead, I suggest using tail: rg foo | tail -n 10 will print only the last ten matches.

[jungle-boogie]

Hi @BurntSushi,

Understood. I had a feeling your response would be to not add it into rg, because of the reasons you cited. Makes sense to me.

I asked the question because rg can already do so much, I didn't know if I was overlooking a one of the flags.

I'll use tail to accomplish reviewing last ten lines.

Thanks for rg and I await the next great release!

[krzentner]

For anyone who comes here still looking for a way to do this for several files, it can approximately be done on unix systems using --pre, tac, and --max-count like so (as long as you don't have a multi-line pattern and don't care about line numbers):

rg --pre tac -m 10 <PATTERN>

This will run tac on every input file, which maybe isn't ideal, but that can be worked around in exchange for running rg twice (usually faster):

rg --pre tac -m 10 <PATTERN> $(rg -l <PATTERN>)

or, if you have a very large number of matching files:

rg -l <PATTERN> | xargs rg --pre tac -m 10 <PATTERN>

If you do have a multi-line pattern or care about line numbers, you'll probably want to just use the shell:

rg -l <PATTERN> | while read f; do rg <PATTERN> $f | tail -n 10; done

Thanks for making such a great tool, @BurntSushi.