[c++] Performant DataFrame.shape

libtiledbsoma/src/soma/soma_array.cc

@@ -1196,6 +1196,30 @@ std::vector<int64_t> SOMAArray::maxshape() {
     return _tiledb_domain();
 }
 
+std::optional<int64_t> SOMAArray::_shape_slot_if_soma_joinid_dim() {
+    const std::string dim_name = "soma_joinid";
+
+    if (!arr_->schema().domain().has_dimension(dim_name)) {
+        return std::nullopt;
+    }
+
+    auto current_domain = _get_current_domain();
+    if (current_domain.is_empty()) {
+        return std::nullopt;
+    }
+
+    auto t = current_domain.type();
+    if (t != TILEDB_NDRECTANGLE) {
+        throw TileDBSOMAError("current_domain type is not NDRECTANGLE");
+    }
+
+    NDRectangle ndrect = current_domain.ndrectangle();
+
+    auto range = ndrect.range<int64_t>(dim_name);
+    auto max = range[1] + 1;
+    return std::optional<int64_t>(max);
+}
+
 std::vector<int64_t> SOMAArray::_tiledb_domain() {
     std::vector<int64_t> result;
     auto dimensions = mq_->schema()->domain().dimensions();

libtiledbsoma/src/soma/soma_array.h

@@ -796,6 +796,14 @@ class SOMAArray : public SOMAObject {
      */
     std::optional<TimestampRange> timestamp();
 
+   protected:
+    // For use nominally by SOMADataFrame. This could be moved in its entrety to
+    // SOMADataFrame, but it would entail moving several SOMAArray attributes
+    // from private to protected, which has knock-on effects on the order of
+    // constructor initializers, etc.: in total it's simplest to place this
+    // here, and have SOMADataFrame invoke it.
+    std::optional<int64_t> _shape_slot_if_soma_joinid_dim();
+
    private:
     //===================================================================
     //= private non-static

libtiledbsoma/src/soma/soma_dataframe.cc

@@ -94,4 +94,10 @@ uint64_t SOMADataFrame::count() {
     return this->nnz();
 }
 
+std::vector<int64_t> SOMADataFrame::shape() {
+    std::optional<int64_t> attempt = _shape_slot_if_soma_joinid_dim();
+    int64_t max = attempt.has_value() ? attempt.value() : this->nnz();
+    return std::vector<int64_t>({max});
+}
+
 }  // namespace tiledbsoma

libtiledbsoma/src/soma/soma_dataframe.h

@@ -163,7 +163,19 @@ class SOMADataFrame : public SOMAArray {
      * @return int64_t
      */
     uint64_t count();
+
+    /**
+     * For DataFrame with default indexing, namely, a single int64_t
+     * soma_joinid, return the same as SOMAArray. For DataFrame with soma_joinid
+     * being a dim along with other dims (optional behavior), return the slot
+     * along that dim. For DataFrame with soma_joinid being an attr, not a dim
+     * at all, returns nnz().
+     *
+     * @return int64_t
+     */
+    std::vector<int64_t> shape();
 };
+
 }  // namespace tiledbsoma
 
 #endif  // SOMA_DATAFRAME

libtiledbsoma/test/unit_soma_sparse_ndarray.cc

@@ -71,9 +71,9 @@ TEST_CASE("SOMASparseNDArray: basic") {
         REQUIRE(soma_sparse->ndim() == 1);
         REQUIRE(soma_sparse->nnz() == 0);
 
-        if (use_current_domain) {
-            REQUIRE(soma_sparse->shape() == std::vector<int64_t>{dim_max + 1});
-        } else {
+        auto expect = std::vector<int64_t>{dim_max + 1};
+        REQUIRE(soma_sparse->shape() == std::vector<int64_t>{dim_max + 1});
+        if (!use_current_domain) {
             REQUIRE(
                 soma_sparse->maxshape() == std::vector<int64_t>{dim_max + 1});
         }