子集DateVector

Question

I'm trying to extract and subset a vector containing date information from a data.frame . I'm able to successfully extract the DateVector from the DataFrame ; however, I receive an error when trying to subset the data.

The below works fine given the /* */ around the DateVector subsets.

Rcpp::cppFunction('
Rcpp::DataFrame test(DataFrame x, StringVector y ) {

  StringVector New = x["string_1"];
  std::string KEY = Rcpp::as<std::string>(y[0]);
  Rcpp::LogicalVector ind(New.size());

  for(int i = 0; i < New.size(); i++){
  ind[i] = (New[i] == KEY);
  }


  Rcpp::StringVector st1 = x["string_1"];
  Rcpp::StringVector Id = x["ID"];
  Rcpp::StringVector NameId = x["NameID"];
  Rcpp::DateVector StDate = x["StartDate"];
  Rcpp::DateVector EtDate = x["EndDate"]; 

  /*
  Rcpp::DateVector StDate_sub = StDate[ind];
  Rcpp::DateVector EtDate_sub = EtDate[ind]; 
  */

  return Rcpp::DataFrame::create(Rcpp::Named("string_1") = st1[ind],
                                 Rcpp::Named("ID") = Id[ind],
                                 Rcpp::Named("NameID") = NameId[ind]/*,
                                 Rcpp::Named("StartDate") = StDate_sub,
                                 Rcpp::Named("EndDate") = EtDate_sub*/
                                 );
}')

There are two notable errors I receive:

error: invalid user-defined conversion from 'Rcpp::LogicalVector {aka Rcpp::Vector<10, Rcpp::PreserveStorage>}' to 'int' [-fpermissive]

Rcpp::DateVector StDate_sub = StDate[ind]

The second is:

no known conversion from 'SEXP' to 'int' file585c1863151c.cpp:23:53: error: conversion from 'Rcpp::Date' to non-scalar type 'Rcpp::DateVector {aka Rcpp::oldDateVector}' requested

Rcpp::DateVector EtDate_sub = EtDate[ind];

I looked at the docs, but couldn't find a way. Sorry, if I missed it. I have a couple of date variables in data.frame . I am using the Rcpp to subset the data set in a nested for loop. Currently, it is taking too much time. I cannot implement it in data.table or dplyr as the subset data set is required fro some processing.

Answer 1

First off, your example is not minimally reproducible as there is no defined data set.

Second, you are making the (heroic?) assumption that assignment by index vector be defined for Date vectors. Appears it may not be.

Third, just looping is trivial. Amended code below. Builds without a hitch, no idea if it run as you supplied no reference data .

#define RCPP_NEW_DATE_DATETIME_VECTORS 1
#include <Rcpp.h>

using namespace Rcpp;

// [[Rcpp::export]]
Rcpp::DataFrame dftest(DataFrame x, StringVector y ) {

  StringVector New = x["string_1"];
  std::string KEY = Rcpp::as<std::string>(y[0]);
  Rcpp::LogicalVector ind(New.size());

  for(int i = 0; i < New.size(); i++){
    ind[i] = (New[i] == KEY);
  }


  Rcpp::StringVector st1 = x["string_1"];
  Rcpp::StringVector Id = x["ID"];
  Rcpp::StringVector NameId = x["NameID"];
  Rcpp::DateVector StDate = x["StartDate"];
  Rcpp::DateVector EtDate = x["EndDate"]; 

  int n = sum(ind);
  Rcpp::DateVector StDate_sub = StDate(n);
  Rcpp::DateVector EtDate_sub = EtDate(n);
  for (int i=0; i<n; i++) {
    StDate_sub[i] = StDate( ind[i] );
    EtDate_sub[i] = EtDate( ind[i] );
  }

  return Rcpp::DataFrame::create(Rcpp::Named("string_1") = st1[ind],
                                 Rcpp::Named("ID") = Id[ind],
                                 Rcpp::Named("NameID") = NameId[ind],
                                 Rcpp::Named("StartDate") = StDate_sub,
                                 Rcpp::Named("EndDate") = EtDate_sub);
}