[r] Push attrnames down to C++ (#3121)

* [r] Push `attrnames` down to C++ * more
single-cell-data · Oct 3, 2024 · 342673e · 342673e
1 parent dd0b823
commit 342673e
Show file tree

Hide file tree

Showing 7 changed files with 56 additions and 12 deletions.
diff --git a/apis/r/R/RcppExports.R b/apis/r/R/RcppExports.R
@@ -214,6 +214,10 @@ c_dimnames <- function(uri, ctxxp) {
     .Call(`_tiledbsoma_c_dimnames`, uri, ctxxp)
 }
 
+c_attrnames <- function(uri, ctxxp) {
+    .Call(`_tiledbsoma_c_attrnames`, uri, ctxxp)
+}
+
 resize <- function(uri, new_shape, ctxxp) {
     invisible(.Call(`_tiledbsoma_resize`, uri, new_shape, ctxxp))
 }

diff --git a/apis/r/R/SOMADenseNDArray.R b/apis/r/R/SOMADenseNDArray.R
@@ -80,11 +80,11 @@ SOMADenseNDArray <- R6::R6Class(
     ) {
       private$check_open_for_read()
 
-      dims <- self$dimensions()
-      attr <- self$attributes()
+      ndim <- self$ndim()
+      attrnames <- self$attrnames()
 
-      stopifnot("Array must have two dimensions" = length(dims) == 2,
-                "Array must contain column 'soma_data'" = all.equal("soma_data", names(attr)))
+      stopifnot("Array must have two dimensions" = ndim == 2,
+                "Array must contain column 'soma_data'" = all.equal("soma_data", attrnames))
 
       if (is.null(coords)) {
         ned <- self$non_empty_domain()

diff --git a/apis/r/R/TileDBArray.R b/apis/r/R/TileDBArray.R
@@ -282,12 +282,7 @@ TileDBArray <- R6::R6Class(
     #' @description Retrieve attribute names (lifecycle: maturing)
     #' @return A character vector with the array's attribute names
     attrnames = function() {
-      vapply(
-        self$attributes(),
-        FUN = tiledb::name,
-        FUN.VALUE = vector("character", 1L),
-        USE.NAMES = FALSE
-      )
+      c_attrnames(self$uri, private$.soma_context)
     },
 
     #' @description Retrieve the names of all columns, including dimensions and

diff --git a/apis/r/src/RcppExports.cpp b/apis/r/src/RcppExports.cpp
@@ -476,6 +476,18 @@ BEGIN_RCPP
     return rcpp_result_gen;
 END_RCPP
 }
+// c_attrnames
+Rcpp::CharacterVector c_attrnames(const std::string& uri, Rcpp::XPtr<somactx_wrap_t> ctxxp);
+RcppExport SEXP _tiledbsoma_c_attrnames(SEXP uriSEXP, SEXP ctxxpSEXP) {
+BEGIN_RCPP
+    Rcpp::RObject rcpp_result_gen;
+    Rcpp::RNGScope rcpp_rngScope_gen;
+    Rcpp::traits::input_parameter< const std::string& >::type uri(uriSEXP);
+    Rcpp::traits::input_parameter< Rcpp::XPtr<somactx_wrap_t> >::type ctxxp(ctxxpSEXP);
+    rcpp_result_gen = Rcpp::wrap(c_attrnames(uri, ctxxp));
+    return rcpp_result_gen;
+END_RCPP
+}
 // resize
 void resize(const std::string& uri, Rcpp::NumericVector new_shape, Rcpp::XPtr<somactx_wrap_t> ctxxp);
 RcppExport SEXP _tiledbsoma_resize(SEXP uriSEXP, SEXP new_shapeSEXP, SEXP ctxxpSEXP) {
@@ -734,6 +746,7 @@ static const R_CallMethodDef CallEntries[] = {
     {"_tiledbsoma_has_current_domain", (DL_FUNC) &_tiledbsoma_has_current_domain, 2},
     {"_tiledbsoma_ndim", (DL_FUNC) &_tiledbsoma_ndim, 2},
     {"_tiledbsoma_c_dimnames", (DL_FUNC) &_tiledbsoma_c_dimnames, 2},
+    {"_tiledbsoma_c_attrnames", (DL_FUNC) &_tiledbsoma_c_attrnames, 2},
     {"_tiledbsoma_resize", (DL_FUNC) &_tiledbsoma_resize, 3},
     {"_tiledbsoma_resize_soma_joinid", (DL_FUNC) &_tiledbsoma_resize_soma_joinid, 3},
     {"_tiledbsoma_tiledbsoma_upgrade_shape", (DL_FUNC) &_tiledbsoma_tiledbsoma_upgrade_shape, 3},

diff --git a/apis/r/src/rinterface.cpp b/apis/r/src/rinterface.cpp
@@ -374,6 +374,21 @@ Rcpp::CharacterVector c_dimnames(
     return retval;
 }
 
+// [[Rcpp::export]]
+Rcpp::CharacterVector c_attrnames(
+    const std::string& uri, Rcpp::XPtr<somactx_wrap_t> ctxxp) {
+    auto sr = tdbs::SOMAArray::open(OpenMode::read, uri, ctxxp->ctxptr);
+    auto lib_retval = sr->attribute_names();
+    sr->close();
+
+    size_t n = lib_retval.size();
+    Rcpp::CharacterVector retval(n);
+    for (size_t i = 0; i < n; i++) {
+        retval[i] = lib_retval[i];
+    }
+    return retval;
+}
+
 // [[Rcpp::export]]
 void resize(
     const std::string& uri,

diff --git a/libtiledbsoma/src/soma/soma_array.cc b/libtiledbsoma/src/soma/soma_array.cc
@@ -1022,6 +1022,16 @@ bool SOMAArray::has_dimension_name(const std::string& name) const {
     return false;
 }
 
+std::vector<std::string> SOMAArray::attribute_names() const {
+    std::vector<std::string> result;
+    auto schema = tiledb_schema();
+    unsigned n = schema->attribute_num();
+    for (unsigned i = 0; i < n; i++) {
+        result.push_back(schema->attribute(i).name());
+    }
+    return result;
+}
+
 void SOMAArray::write(bool sort_coords) {
     if (mq_->query_type() != TILEDB_WRITE) {
         throw TileDBSOMAError("[SOMAArray] array must be opened in write mode");

diff --git a/libtiledbsoma/src/soma/soma_array.h b/libtiledbsoma/src/soma/soma_array.h
@@ -284,9 +284,9 @@ class SOMAArray : public SOMAObject {
     uint64_t ndim() const;
 
     /**
-     * @brief Get the name of each dimensions.
+     * @brief Get the name of each dimension.
      *
-     * @return std::vector<std::string> Name of each dimensions.
+     * @return std::vector<std::string> Name of each dimension.
      */
     std::vector<std::string> dimension_names() const;
 
@@ -297,6 +297,13 @@ class SOMAArray : public SOMAObject {
      */
     bool has_dimension_name(const std::string& name) const;
 
+    /**
+     * @brief Get the name of each attribute.
+     *
+     * @return std::vector<std::string> Name of each attribute.
+     */
+    std::vector<std::string> attribute_names() const;
+
     /**
      * @brief Set the dimension slice using one point
      *