Problem with parse_expressions() for input with if else

[maelle]

Reading the comments and test I'm not sure what's supposed to happen for this case actually, as it's not the same usage of the curly braces.

What's the expected output?

Why use regular expressions as opposed to parsing code? Then hopefully one could exclude if else statements from the transformations? I'm a bit confused.

[maelle]

Even more concerning, match_brackets() returns NULL:

ex <- c("if (TRUE) {", "    message(\"blop\")", 
"} else {", "    message(\"bla\"", "    ))", "}")
autotest:::match_brackets (ex)
#> NULL
autotest:::parse_expressions (ex)
#> Error in if (qts[[i]][1] > 0) {: missing value where TRUE/FALSE needed

^{Created on 2023-07-11 with reprex v2.0.2}

I'm wondering if we didn't notice this before, because manual pages rarely feature logic?!

[codecov]

Codecov Report

Patch and project coverage have no change.

Comparison is base (1a4e37d) 78.53% compared to head (3776b01) 78.53%.

❗ Current head 3776b01 differs from pull request most recent head 7b06938. Consider uploading reports for the commit 7b06938 to get more accurate results

Additional details and impacted files

@@           Coverage Diff           @@
##             main      #82   +/-   ##
=======================================
  Coverage   78.53%   78.53%           
=======================================
  Files          29       29           
  Lines        4304     4304           
=======================================
  Hits         3380     3380           
  Misses        924      924           

Impacted Files	Coverage Δ
R/text-parsing-fns.R	80.60% <ø> (ø)

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[maelle]

Another package where the problem can be observed https://github.com/mjwoods/RNetCDF

https://github.com/cran/RNetCDF/blob/265b18973cdf1ccd806cc6a4ba68910d79251076/man/utinvcal.nc.Rd#L32

[mpadge]

Thanks @maelle, but note that all of these hand-parsing routines have been completely removed in current dev version in the typetracer branch. Unsure of current timetable for merging that, but once done it will completely change the way autotest works, to use typetracer instead of all this brittle regex code.