isSubset.cpp

isSubset.cpp

#include <Rcpp.h>
#include "string"

using namespace Rcpp;

// [[Rcpp::export]]
DataFrame isSubset(DataFrame  data) {
  
  int n=data.nrows();
  StringVector lhs;
  StringVector rhs;
  try{  
   StringVector lhs=data["lhs"];
   StringVector rhs=data["rhs"];
  }
  catch(std::exception e){
   Rcpp::Rcout<<"Invalid column names were passed, function returns empty DataFrame" <<std::endl;
  };
  
  std::string *arrayLhs=new  std::string [n];
  std::string *arrayRhs=new  std::string [n];
  
  
  for(int i=0;i<n;i++){
    arrayLhs[i]=Rcpp::as<std::string> (lhs(i));
    arrayRhs[i]=Rcpp::as<std::string> (rhs(i));
  }  
  
  
  NumericVector hasSubset=NumericVector(n);
  
  for(int i=0;i<n;i++){
    for(int j=i+1;j<n;j++){
      
      if((arrayLhs[i].find(arrayLhs[j])!= std::string::npos) & arrayRhs[i]==arrayRhs[j]){
        hasSubset[i]=1;
      }
      else if((arrayLhs[j].find(arrayLhs[i])!= std::string::npos) & arrayRhs[i]==arrayRhs[j]){
        hasSubset[j]=1;
        
      }
    }  
  }
  
  delete [] arrayLhs;
  delete [] arrayRhs;
  
  
  data["hasSubset"]=hasSubset;
  
  return data;  
}