CODEWITHSUNDEEP
rrcpp
dww
dww

Reputation: 31452

Using Rcpp sugar on data.frame names

I would like to pass a data.frame as argument to Rcpp function with optional columns. The c++ function then needs to test whether the columns exist or not. If I use the sugar function any as in the following example, I get a compilation error.

cppFunction(
  'double test(DataFrame test_data) {
      double x=NA_REAL;
      CharacterVector colnames = CharacterVector::create("foo");
      CharacterVector df_names = test_data.names();
      if (any(df_names == colnames)) x = 1.0;
      return(x);
    }')

invalid use of incomplete type class Rcpp::sugar::forbidden_conversion`

I know I can test the character values one by one in a loop, like this (which works as expected):

cppFunction(
  'double test(DataFrame test_data) {
      double x=NA_REAL;
      CharacterVector colnames = CharacterVector::create("foo");
      CharacterVector df_names = test_data.names();
      for (int i=0; i<df_names.length(); i++) {
        if (df_names[i] == colnames[i]) x = 1.0;
      }
      return(x);
    }')
test(data.frame(bar=3))
# [1] NA
test(data.frame(foo=3))
# [1] 1

But, I would like to use a vectorised "sugar" version if this is possible. What am I doing wrong, and how do I do that?

Upvotes: 0

Views: 116

Answers (1)

Waldi
Waldi

Reputation: 41220

You could add is_true to return a boolean:

library(Rcpp)
cppFunction(
  'double test(DataFrame test_data) {
      double x=NA_REAL;
      CharacterVector colnames = CharacterVector::create("foo");
      CharacterVector df_names = test_data.names();
      if (is_true(any(df_names == colnames))) x = 1.0;
      return(x);
    }')

Upvotes: 5

Related Questions