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;
}
DataFrame subset_join(DataFrame x, DataFrame y,
const Index& indices_x, const Index& indices_y,
CharacterVector by_x, CharacterVector by_y,
const std::string& suffix_x, const std::string& suffix_y,
CharacterVector classes) {
if (suffix_x.length() == 0 && suffix_y.length() == 0) {
stop("Cannot use empty string for both x and y suffixes");
}
// first the joined columns
DataFrameJoinVisitors join_visitors(x, y, SymbolVector(by_x), SymbolVector(by_y), false, false);
int n_join_visitors = join_visitors.size();
// then columns from x but not y
CharacterVector all_x_columns = x.names();
std::vector<bool> joiner(all_x_columns.size());
CharacterVector x_columns(all_x_columns.size() - n_join_visitors);
IntegerVector xm = r_match(all_x_columns, by_x);
for (int i = 0, k = 0; i < all_x_columns.size(); i++) {
if (xm[i] == NA_INTEGER) {
joiner[i] = false;
x_columns[k++] = all_x_columns[i];
} else {
joiner[i] = true;
}
}
DataFrameSubsetVisitors visitors_x(x, SymbolVector(x_columns));
int nv_x = visitors_x.size();
// then columns from y but not x
CharacterVector all_y_columns = y.names();
CharacterVector y_columns(all_y_columns.size() - n_join_visitors);
IntegerVector ym = r_match(all_y_columns, by_y);
for (int i = 0, k = 0; i < all_y_columns.size(); i++) {
if (ym[i] == NA_INTEGER) {
y_columns[k++] = all_y_columns[i];
}
}
DataFrameSubsetVisitors visitors_y(y, SymbolVector(y_columns));
int nv_y = visitors_y.size();
// construct out object
int nrows = indices_x.size();
List out(n_join_visitors + nv_x + nv_y);
CharacterVector names(n_join_visitors + nv_x + nv_y);
int index_join_visitor = 0;
int index_x_visitor = 0;
// ---- join visitors
for (int i = 0; i < all_x_columns.size(); i++) {
String col_name = all_x_columns[i];
if (joiner[i]) {
JoinVisitor* v = join_visitors.get(xm[i] - 1);
out[i] = v->subset(indices_x);
index_join_visitor++;
} else {
// we suffix by .x if this column is in y_columns (and if the suffix is not empty)
if (suffix_x.length() > 0) {
while (
(std::find(y_columns.begin(), y_columns.end(), col_name.get_sexp()) != y_columns.end()) ||
(std::find(names.begin(), names.begin() + i, col_name.get_sexp()) != names.begin() + i)
) {
col_name += suffix_x;
}
}
out[i] = visitors_x.get(index_x_visitor)->subset(indices_x);
index_x_visitor++;
}
names[i] = col_name;
}
int k = index_join_visitor + index_x_visitor;
for (int i = 0; i < nv_y; i++, k++) {
String col_name = y_columns[i];
// we suffix by .y if this column is in x_columns (and if the suffix is not empty)
if (suffix_y.length() > 0) {
while (
(std::find(all_x_columns.begin(), all_x_columns.end(), col_name.get_sexp()) != all_x_columns.end()) ||
(std::find(names.begin(), names.begin() + k, col_name.get_sexp()) != names.begin() + k)
) {
col_name += suffix_y;
}
}
out[k] = visitors_y.get(i)->subset(indices_y);
names[k] = col_name;
}
set_class(out, classes);
set_rownames(out, nrows);
out.names() = names;
// out group columns
SymbolVector group_cols_x = get_vars(x);
int n_group_cols = group_cols_x.size();
SymbolVector group_cols(n_group_cols);
IntegerVector group_col_indices = group_cols_x.match_in_table(all_x_columns);
// get updated column names
for (int i = 0; i < n_group_cols; i++) {
int group_col_index = group_col_indices[i];
if (group_col_index != NA_INTEGER) {
group_cols.set(i, names[group_col_index - 1]);
} else {
stop("unknown group column '%s'", group_cols_x[i].get_utf8_cstring());
}
}
set_vars(out, group_cols);
return (SEXP)out;
}