// [[Rcpp::export]]
SEXP combine_all( List data ){
int nv = data.size() ;
if( nv == 0 ) stop("combine_all needs at least one vector") ;
// get the size of the output
int n = 0 ;
for( int i=0; i<nv; i++){
n += Rf_length(data[i]) ;
}
// collect
Collecter* coll = collecter( data[0], n ) ;
coll->collect( SlicingIndex(0, Rf_length(data[0])), data[0] ) ;
int k = Rf_length(data[0]) ;
for( int i=1; i<nv; i++){
SEXP current = data[i] ;
int n_current= Rf_length(current) ;
if( coll->compatible(current) ){
coll->collect( SlicingIndex(k, n_current), current ) ;
} else if( coll->can_promote(current) ) {
Collecter* new_coll = promote_collecter(current, n, coll) ;
new_coll->collect( SlicingIndex(k, n_current), current ) ;
new_coll->collect( SlicingIndex(0, k), coll->get() ) ;
delete coll ;
coll = new_coll ;
} else {
std::stringstream msg ;
msg << "incompatible type at index "
<< (i+1)
<< " : "
<< get_single_class(current)
<< ", was collecting : "
<< get_single_class(coll->get())
;
stop( msg.str() ) ;
}
k += n_current ;
}
RObject out = coll->get() ;
delete coll ;
return out ;
}
List rbind__impl( Dots dots ){
int ndata = dots.size() ;
int n = 0 ;
std::vector<DataFrameAble> chunks ;
for( int i=0; i<ndata; i++) {
chunks.push_back( DataFrameAble( dots[i] ) );
n += chunks[i].nrows() ;
}
pointer_vector<Collecter> columns ;
std::vector<String> names ;
int k=0 ;
Function enc2native( "enc2native" ) ;
for( int i=0; i<ndata; i++){
Rcpp::checkUserInterrupt() ;
const DataFrameAble& df = chunks[i] ;
if( !df.size() ) continue ;
int nrows = df.nrows() ;
CharacterVector df_names = enc2native(df.names()) ;
for( int j=0; j<df.size(); j++){
SEXP source = df.get(j) ;
String name = df_names[j] ;
Collecter* coll = 0;
size_t index = 0 ;
for( ; index < names.size(); index++){
if( name == names[index] ){
coll = columns[index] ;
break ;
}
}
if( ! coll ){
coll = collecter( source, n ) ;
columns.push_back( coll );
names.push_back(name) ;
}
if( coll->compatible(source) ){
// if the current source is compatible, collect
coll->collect( SlicingIndex( k, nrows), source ) ;
} else if( coll->can_promote(source) ) {
// setup a new Collecter
Collecter* new_collecter = promote_collecter(source, n, coll ) ;
// import data from this chunk
new_collecter->collect( SlicingIndex( k, nrows), source ) ;
// import data from previous collecter
new_collecter->collect( SlicingIndex(0, k), coll->get() ) ;
// dispose the previous collecter and keep the new one.
delete coll ;
columns[index] = new_collecter ;
} else if( all_na(source) ) {
// do nothing, the collecter already initialized data with the
// right NA
} else if( coll->is_logical_all_na() ) {
Collecter* new_collecter = collecter( source, n ) ;
new_collecter->collect( SlicingIndex(k, nrows), source ) ;
delete coll ;
columns[index] = new_collecter ;
} else {
std::string column_name(name) ;
stop( "incompatible type (data index: %d, column: '%s', was collecting: %s (%s), incompatible with data of type: %s",
(i+1), column_name, coll->describe(), DEMANGLE(*coll), get_single_class(source) );
}
}
k += nrows ;
}
int nc = columns.size() ;
List out(nc) ;
CharacterVector out_names(nc) ;
for( int i=0; i<nc; i++){
out[i] = columns[i]->get() ;
out_names[i] = names[i] ;
}
out.attr( "names" ) = out_names ;
set_rownames( out, n );
out.attr( "class" ) = classes_not_grouped() ;
return out ;
}
//' @export
//' @rdname rbind
// [[Rcpp::export]]
List rbind_all( ListOf<DataFrame> dots ){
int ndata = dots.size() ;
int n = 0 ;
for( int i=0; i<ndata; i++) n += dots[i].nrows() ;
std::vector<Collecter*> columns ;
std::vector<String> names ;
int k=0 ;
for( int i=0; i<ndata; i++){
DataFrame df = dots[i] ;
DataFrameVisitors visitors( df, df.names() ) ;
int nrows = df.nrows() ;
CharacterVector df_names = df.names() ;
for( int j=0; j<df.size(); j++){
SEXP source = df[j] ;
String name = df_names[j] ;
Collecter* coll = 0;
size_t index = 0 ;
for( ; index < names.size(); index++){
if( name == names[index] ){
coll = columns[index] ;
break ;
}
}
if( ! coll ){
coll = collecter( source, n ) ;
columns.push_back( coll );
names.push_back(name) ;
}
if( coll->compatible(source) ){
// if the current source is compatible, collect
coll->collect( SlicingIndex( k, nrows), source ) ;
} else if( coll->can_promote(source) ) {
// setup a new Collecter
Collecter* new_collecter = promote_collecter(source, n, coll ) ;
// import data from this chunk
new_collecter->collect( SlicingIndex( k, nrows), source ) ;
// import data from previous collecter
new_collecter->collect( SlicingIndex(0, k), coll->get() ) ;
// dispose the previous collecter and keep the new one.
delete coll ;
columns[index] = new_collecter ;
} else {
std::stringstream msg ;
std::string column_name(name) ;
msg << "incompatible type ("
<< "data index: "
<< (i+1)
<< ", column: '"
<< column_name
<< "', was collecting: "
<< coll->describe()
<< " ("
<< DEMANGLE(*coll)
<< ")"
<< ", incompatible with data of type: "
<< type_name(source) ;
stop( msg.str() ) ;
}
}
k += nrows ;
}
int nc = columns.size() ;
List out(nc) ;
CharacterVector out_names(nc) ;
for( int i=0; i<nc; i++){
out[i] = columns[i]->get() ;
out_names[i] = names[i] ;
}
out.attr( "names" ) = out_names ;
delete_all( columns ) ;
set_rownames( out, n );
out.attr( "class" ) = "data.frame" ;
return out ;
}
List rbind__impl( Dots dots, SEXP id = R_NilValue ){
int ndata = dots.size() ;
int n = 0 ;
DataFrameAbleVector chunks ;
std::vector<int> df_nrows ;
int k=0 ;
for( int i=0; i<ndata; i++) {
SEXP obj = dots[i] ;
if( Rf_isNull(obj) ) continue ;
chunks.push_back( obj ) ;
int nrows = chunks[k].nrows() ;
df_nrows.push_back(nrows) ;
n += nrows ;
k++ ;
}
ndata = chunks.size() ;
pointer_vector<Collecter> columns ;
std::vector<String> names ;
k=0 ;
Function enc2native( "enc2native" ) ;
for( int i=0; i<ndata; i++){
Rcpp::checkUserInterrupt() ;
const DataFrameAble& df = chunks[i] ;
if( !df.size() ) continue ;
int nrows = df.nrows() ;
CharacterVector df_names = enc2native(df.names()) ;
for( int j=0; j<df.size(); j++){
SEXP source = df.get(j) ;
String name = df_names[j] ;
Collecter* coll = 0;
size_t index = 0 ;
for( ; index < names.size(); index++){
if( name == names[index] ){
coll = columns[index] ;
break ;
}
}
if( ! coll ){
coll = collecter( source, n ) ;
columns.push_back( coll );
names.push_back(name) ;
}
if( coll->compatible(source) ){
// if the current source is compatible, collect
coll->collect( SlicingIndex( k, nrows), source ) ;
} else if( coll->can_promote(source) ) {
// setup a new Collecter
Collecter* new_collecter = promote_collecter(source, n, coll ) ;
// import data from this chunk
new_collecter->collect( SlicingIndex( k, nrows), source ) ;
// import data from previous collecter
new_collecter->collect( SlicingIndex(0, k), coll->get() ) ;
// dispose the previous collecter and keep the new one.
delete coll ;
columns[index] = new_collecter ;
} else if( all_na(source) ) {
// do nothing, the collecter already initialized data with the
// right NA
} else if( coll->is_logical_all_na() ) {
Collecter* new_collecter = collecter( source, n ) ;
new_collecter->collect( SlicingIndex(k, nrows), source ) ;
delete coll ;
columns[index] = new_collecter ;
} else {
std::string column_name(name) ;
stop(
"Can not automatically convert from %s to %s in column \"%s\".",
coll->describe(), get_single_class(source), column_name
) ;
}
}
k += nrows ;
}
int nc = columns.size() ;
int has_id = Rf_isNull(id) ? 0 : 1;
List out(nc + has_id) ;
CharacterVector out_names(nc + has_id) ;
for( int i=0; i<nc; i++){
out[i + has_id] = columns[i]->get() ;
out_names[i + has_id] = names[i] ;
}
// Add vector of identifiers if .id is supplied
if (!Rf_isNull(id)) {
CharacterVector df_names = dots.names() ;
CharacterVector id_col = no_init(n) ;
CharacterVector::iterator it = id_col.begin() ;
for (int i=0; i<ndata; ++i) {
std::fill( it, it + df_nrows[i], df_names[i] ) ;
it += df_nrows[i] ;
}
out[0] = id_col ;
out_names[0] = Rcpp::as<std::string>(id) ;
}
out.attr( "names" ) = out_names ;
set_rownames( out, n ) ;
// infer the classes and extra info (groups, etc ) from the first (#1692)
if( ndata ){
const DataFrameAble& first = chunks[0] ;
if( first.is_dataframe() ){
DataFrame df = first.get() ;
out.attr("class") = df.attr("class") ;
if( df.inherits("grouped_df") ){
out.attr("vars") = df.attr("vars") ;
out = GroupedDataFrame(out).data() ;
}
} else {
out.attr( "class" ) = classes_not_grouped() ;
}
} else {
out.attr( "class" ) = classes_not_grouped() ;
}
return out ;
}
