is it faster to call find before calling captures?

[tsoutsman]

Hi,
The documentation and #389 suggest that you should use find instead of captures as it is more performant. I want to find the line number of a particular regex match and get the capture groups from it. Currently, I'm iterating through the lines of a file and running captures on each line, even though most of them won't contain a match. Should I run find on each line and then only run capture if find returns a match?

I've been looking into the grep crate which is behind ripgrep and that seems to be what it does (at least from the example on this page).

[BurntSushi]

Good question. No, you definitely do not want to run find first, as the regex crate will actually do that for you. In cases where the DFA can't be used, then you would wind up making your total time twice as slow as you would just be running the NFA twice.

The example in the grep crate is doing something different. The searcher only reports lines that match. So you have to run the regex again to find the bounds of each individual match, which is strictly more work. Finding the bounds of the line that matches can be done with less work via the shortest_match method. Indeed, that is what the grep crates do internally.

[tsoutsman]

Thanks!