DataFrame subset_join(DataFrame x, DataFrame y,
const Index& indices_x, const Index& indices_y,
const IntegerVector& by_x, const IntegerVector& by_y,
const IntegerVector& aux_x, const IntegerVector& aux_y,
CharacterVector classes) {
// construct out object
List out(x.ncol() + aux_y.size());
// first the joined columns (all x columns keep their location)
DataFrameJoinVisitors join_visitors(x, y, by_x, by_y, true, false);
for (int i = 0; i < by_x.size(); i++) {
JoinVisitor* v = join_visitors.get(i);
out[by_x[i] - 1] = v->subset(indices_x);
}
// then the auxiliary x columns (all x columns keep their location)
DataFrameSubsetVisitors visitors_x(x, aux_x);
for (int i = 0; i < aux_x.size(); i++) {
SubsetVectorVisitor* const v = visitors_x.get(i);
out[aux_x[i] - 1] = v->subset(indices_x);
}
// then the auxiliary y columns (all y columns keep their relative location)
DataFrameSubsetVisitors visitors_y(y, aux_y);
for (int i = 0, k = x.ncol(); i < visitors_y.size(); i++, k++) {
SubsetVectorVisitor* const v = visitors_y.get(i);
out[k] = v->subset(indices_y);
}
int nrows = indices_x.size();
set_rownames(out, nrows);
set_class(out, classes);
return (SEXP)out;
}
// Based on table()
// dataframe {
// for each column get the num of dims.tfm
// get size of the resulting talbe
// finally tabulate by the bins meaning how many are there
// each value will correspond to its index in the dim array.
// this is just indexing by a set of values, then you go to there and find it.
// }
// [[Rcpp::export]]
Rcpp::IntegerVector table_cpp(const RObject & input, const RObject & columns) {
if(!is<DataFrame>(input)) stop("Must be a data frame.");
DataFrame data = as<DataFrame>(input);
if(!is<CharacterVector>(columns)) stop("Must be character vector.");
CharacterVector cols = as<CharacterVector>(columns);
data = data[cols];
const R_xlen_t ncols = data.ncol();
if (ncols == 0) stop("No columns in data frame.");
const IntegerVector & column = data.at(0);
// There is a single entry for each row
IntegerVector to_tabulate = no_init(column.size());
to_tabulate.fill(1);
// The product of dimensions.
R_xlen_t pd = 1;
IntegerVector dims(ncols);
List dimnames(ncols);
dimnames.names() = data.names();
for (R_xlen_t i = 0; i < ncols; i++) {
const IntegerVector & a = data.at(i);
if(!Rf_isFactor(a)) stop("Not a factor.");
const CharacterVector & factorLevels = a.attr("levels");
R_xlen_t nl = factorLevels.size();
to_tabulate = to_tabulate + pd * (a - 1L);
pd = pd * nl ;
dims.at(i) = nl;
dimnames.at(i) = factorLevels;
}
to_tabulate = na_omit(to_tabulate);
IntegerVector tbl = tabulate_cpp(to_tabulate, pd);
tbl.attr("dim") = dims;
tbl.attr("dimnames") = dimnames;
tbl.attr("class") = "table";
return tbl;
}
