[c++] Unit-test variant-indexed dataframes (#2944)

* [c++] Split out a test fixture for dataframes * fix a failure case * [c++] Unit-test variant-indexed dataframes [WIP] * neaten * code-review feedback [skip_ci]
single-cell-data · Sep 3, 2024 · ded1e42 · ded1e42
1 parent 37b6e50
commit ded1e42
Show file tree

Hide file tree

Showing 4 changed files with 204 additions and 4 deletions.
diff --git a/libtiledbsoma/src/soma/soma_array.h b/libtiledbsoma/src/soma/soma_array.h
@@ -796,6 +796,13 @@ class SOMAArray : public SOMAObject {
      */
     std::optional<TimestampRange> timestamp();
 
+    /**
+     * Exposed for testing purposes.
+     */
+    CurrentDomain get_current_domain() {
+        return _get_current_domain();
+    }
+
    private:
     //===================================================================
     //= private non-static

diff --git a/libtiledbsoma/src/utils/arrow_adapter.cc b/libtiledbsoma/src/utils/arrow_adapter.cc
@@ -868,9 +868,8 @@ ArraySchema ArrowAdapter::tiledb_schema_from_arrow_schema(
                     //   nullptr)
                     //
                     // Fortunately, these are ASCII dims and we can range
-                    // these accordingly. These are minimum and maximum
-                    // values, avoiding the extremes 0x00 and 0xff.
-                    ndrect.set_range(col_name, "\x01", "\xfe");
+                    // these accordingly.
+                    ndrect.set_range(col_name, "", "\xff");
                 } else {
                     const void* buff = index_column_array->children[i]
                                            ->buffers[1];

diff --git a/libtiledbsoma/test/common.cc b/libtiledbsoma/test/common.cc
@@ -237,7 +237,7 @@ static std::unique_ptr<ArrowArray> _create_index_cols_info_array(
                 std::memcpy((void*)dim_array->buffers[1], vsrc, nbytes);
             } else {
                 // domain small; current_domain feature not being used
-                int64_t dom[] = {0, info.dim_max, 1};
+                uint32_t dom[] = {0, (uint32_t)info.dim_max, 1};
                 void* vsrc = (void*)&dom[0];
                 std::memcpy((void*)dim_array->buffers[1], vsrc, nbytes);
             }

diff --git a/libtiledbsoma/test/unit_soma_dataframe.cc b/libtiledbsoma/test/unit_soma_dataframe.cc
@@ -461,3 +461,197 @@ TEST_CASE_METHOD(
 
     soma_dataframe->close();
 }
+
+TEST_CASE_METHOD(
+    VariouslyIndexedDataFrameFixture,
+    "SOMADataFrame: variant-indexed dataframe dim:sjid attr:str,u32") {
+    auto use_current_domain = GENERATE(false, true);
+    std::ostringstream section;
+    section << "- use_current_domain=" << use_current_domain;
+    SECTION(section.str()) {
+        set_up(
+            std::make_shared<SOMAContext>(),
+            "mem://unit-test-variant-indexed-dataframe-1");
+
+        std::vector<helper::DimInfo> dim_infos(
+            {i64_dim_info(use_current_domain)});
+        std::vector<helper::AttrInfo> attr_infos(
+            {str_attr_info(), u32_attr_info()});
+
+        // Create
+        create(dim_infos, attr_infos);
+
+        // Check current domain
+        auto soma_dataframe = open(OpenMode::read);
+
+        CurrentDomain current_domain = soma_dataframe->get_current_domain();
+        if (!use_current_domain) {
+            REQUIRE(current_domain.is_empty());
+        } else {
+            REQUIRE(!current_domain.is_empty());
+            REQUIRE(current_domain.type() == TILEDB_NDRECTANGLE);
+            NDRectangle ndrect = current_domain.ndrectangle();
+
+            std::array<int64_t, 2> i64_range = ndrect.range<int64_t>(
+                dim_infos[0].name);
+            REQUIRE(i64_range[0] == (int64_t)0);
+            REQUIRE(i64_range[1] == (int64_t)dim_infos[0].dim_max);
+        }
+
+        soma_dataframe->close();
+
+        write_generic_data();
+    }
+}
+
+TEST_CASE_METHOD(
+    VariouslyIndexedDataFrameFixture,
+    "SOMADataFrame: variant-indexed dataframe dim:sjid,u32 attr:str") {
+    auto use_current_domain = GENERATE(false, true);
+    std::ostringstream section;
+    section << "- use_current_domain=" << use_current_domain;
+    SECTION(section.str()) {
+        set_up(
+            std::make_shared<SOMAContext>(),
+            "mem://unit-test-variant-indexed-dataframe-2");
+
+        std::vector<helper::DimInfo> dim_infos(
+            {i64_dim_info(use_current_domain),
+             u32_dim_info(use_current_domain)});
+        std::vector<helper::AttrInfo> attr_infos({str_attr_info()});
+
+        // Create
+        create(dim_infos, attr_infos);
+
+        // Check current domain
+        auto soma_dataframe = open(OpenMode::read);
+
+        CurrentDomain current_domain = soma_dataframe->get_current_domain();
+        if (!use_current_domain) {
+            REQUIRE(current_domain.is_empty());
+        } else {
+            REQUIRE(!current_domain.is_empty());
+            REQUIRE(current_domain.type() == TILEDB_NDRECTANGLE);
+            NDRectangle ndrect = current_domain.ndrectangle();
+
+            std::array<int64_t, 2> i64_range = ndrect.range<int64_t>(
+                dim_infos[0].name);
+            REQUIRE(i64_range[0] == (int64_t)0);
+            REQUIRE(i64_range[1] == (int64_t)dim_infos[0].dim_max);
+
+            std::array<uint32_t, 2> u32_range = ndrect.range<uint32_t>(
+                dim_infos[0].name);
+            REQUIRE(u32_range[0] == (uint32_t)0);
+            REQUIRE(u32_range[1] == (uint32_t)dim_infos[0].dim_max);
+        }
+
+        soma_dataframe->close();
+
+        // Write
+        write_generic_data();
+    }
+}
+
+TEST_CASE_METHOD(
+    VariouslyIndexedDataFrameFixture,
+    "SOMADataFrame: variant-indexed dataframe dim:sjid,str attr:u32") {
+    auto use_current_domain = GENERATE(false, true);
+    std::ostringstream section;
+    section << "- use_current_domain=" << use_current_domain;
+    SECTION(section.str()) {
+        set_up(
+            std::make_shared<SOMAContext>(),
+            "mem://unit-test-variant-indexed-dataframe-3");
+
+        std::vector<helper::DimInfo> dim_infos(
+            {i64_dim_info(use_current_domain),
+             str_dim_info(use_current_domain)});
+        std::vector<helper::AttrInfo> attr_infos({u32_attr_info()});
+
+        // Create
+        create(dim_infos, attr_infos);
+
+        // Check current domain
+        auto soma_dataframe = open(OpenMode::read);
+
+        CurrentDomain current_domain = soma_dataframe->get_current_domain();
+        if (!use_current_domain) {
+            REQUIRE(current_domain.is_empty());
+        } else {
+            REQUIRE(!current_domain.is_empty());
+            REQUIRE(current_domain.type() == TILEDB_NDRECTANGLE);
+            NDRectangle ndrect = current_domain.ndrectangle();
+
+            std::array<int64_t, 2> i64_range = ndrect.range<int64_t>(
+                dim_infos[0].name);
+            REQUIRE(i64_range[0] == (int64_t)0);
+            REQUIRE(i64_range[1] == (int64_t)dim_infos[0].dim_max);
+
+            std::array<std::string, 2> str_range = ndrect.range<std::string>(
+                dim_infos[1].name);
+            // Can we write empty strings in this range?
+            REQUIRE(str_range[0] <= "");
+            REQUIRE(str_range[1] >= "");
+            // Can we write ASCII values in this range?
+            REQUIRE(str_range[0] < " ");
+            REQUIRE(str_range[1] > "~");
+        }
+
+        soma_dataframe->close();
+
+        // Write
+        write_generic_data();
+    }
+}
+
+TEST_CASE_METHOD(
+    VariouslyIndexedDataFrameFixture,
+    "SOMADataFrame: variant-indexed dataframe dim:str,u32 attr:sjid") {
+    auto use_current_domain = GENERATE(false, true);
+    std::ostringstream section;
+    section << "- use_current_domain=" << use_current_domain;
+    SECTION(section.str()) {
+        set_up(
+            std::make_shared<SOMAContext>(),
+            "mem://unit-test-variant-indexed-dataframe-4");
+
+        std::vector<helper::DimInfo> dim_infos(
+            {str_dim_info(use_current_domain),
+             u32_dim_info(use_current_domain)});
+        std::vector<helper::AttrInfo> attr_infos({i64_attr_info()});
+
+        // Create
+        create(dim_infos, attr_infos);
+
+        // Check current domain
+        auto soma_dataframe = open(OpenMode::read);
+
+        CurrentDomain current_domain = soma_dataframe->get_current_domain();
+        if (!use_current_domain) {
+            REQUIRE(current_domain.is_empty());
+        } else {
+            REQUIRE(!current_domain.is_empty());
+            REQUIRE(current_domain.type() == TILEDB_NDRECTANGLE);
+            NDRectangle ndrect = current_domain.ndrectangle();
+
+            std::array<std::string, 2> str_range = ndrect.range<std::string>(
+                dim_infos[0].name);
+            // Can we write empty strings in this range?
+            REQUIRE(str_range[0] <= "");
+            REQUIRE(str_range[1] >= "");
+            // Can we write ASCII values in this range?
+            REQUIRE(str_range[0] < " ");
+            REQUIRE(str_range[1] > "~");
+
+            std::array<uint32_t, 2> u32_range = ndrect.range<uint32_t>(
+                dim_infos[1].name);
+            REQUIRE(u32_range[0] == (uint32_t)0);
+            REQUIRE(u32_range[1] == (uint32_t)dim_infos[1].dim_max);
+        }
+
+        soma_dataframe->close();
+
+        // Write
+        write_generic_data();
+    }
+}