Patch wk_crs_projjson for GDAL < 3.1

[anthonynorth]

Fixes wk_crs_projjson() for crs objects when GDAL < 3.1.

Closes #212

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (030fcce) 98.85% compared to head (cf58c0e) 98.87%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #214      +/-   ##
==========================================
+ Coverage   98.85%   98.87%   +0.01%     
==========================================
  Files          85       85              
  Lines        6208     6208              
==========================================
+ Hits         6137     6138       +1     
+ Misses         71       70       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[anthonynorth]

Ok that doesn't quite work. {sf} is inserting a NA_STRING, which isn't a character vector.

[paleolimbot]

Good catch that GDAL <3.1 is the problem though...I'll take a closer look soon!

[anthonynorth]

This could actually be an {sf} bug and NA_STRING should be wrapped in Rf_ScalarString(NA_STRING)?

All my attempts to test for this value in R failed. We could create a C util to test for NA_STRING, but not sure we should.

[paleolimbot]

It should definitely be fixed in sf (but if it's not too much trouble to fix we should do it here, too).

[anthonynorth]

Ok. The only workable solution I could come up with required a bit of C to use for the is.na() test.

SEXP wk_c_is_na_string(SEXP x) {
  return Rf_ScalarLogical(TYPEOF(x) == CHARSXP && x == NA_STRING);
}