Regular expression matching in tquery() / PERL?

[ChRauh]

Hi,

Thanks for another extremely useful toolbox! (academic love letters here)

I'm now working on a somewhat complex tree query in which I match certain lemmas with a regular expression (using the "__R" flag).
I realized that base R's TRE regex flavor is used under the hood which, however, doesn't support the negative lookahead that I require in my specific application.

Is it possible to switch to perl = TRUE somewhere?

I have naively tried to set this option within tquery() or annotate_tqueries() downstream but this apparently doesn't work.

Any hints would be greatly appreciated (even if the answer is "not possible".

Many thanks,
Christian

[kasperwelbers]

Hi Christian!

I added a __P flag, for perl regex. So if you just use __P instead of __R it 'should' work.

I say 'should', because I didn't have a good case to test it on, and don't really know much about how TRE and PCRE differ. Under the hood what happens if you use regex is that it first finds all unique values in a column with grepl, and then performs a binary search on these values. If regex is a vector, it just concatenates the regex terms with '|'. If you use __P, it now does the same thing but adds the perl = TRUE argument.

Let me know if it works!

Best,
Kasper