[c++] Refactor SOMAArray and SOMAGroup to inherit from `SOMAObjec…

…t` (#2124) * Use `-mmacosx-version-min=11.0` * Drop test for arch in R `configure` --------- Co-authored-by: Dirk Eddelbuettel <[email protected]>
single-cell-data · Feb 12, 2024 · 57347c0 · 57347c0
1 parent fae2ac1
commit 57347c0
Show file tree

Hide file tree

Showing 15 changed files with 134 additions and 958 deletions.
diff --git a/apis/python/setup.py b/apis/python/setup.py
@@ -202,7 +202,7 @@ def run(self):
     CXX_FLAGS.append(f'-Wl,-rpath,{str(tiledb_dir / "lib")}')
 
 if sys.platform == "darwin":
-    CXX_FLAGS.append("-mmacosx-version-min=10.14")
+    CXX_FLAGS.append("-mmacosx-version-min=11.0")
 
 if os.name == "posix" and sys.platform != "darwin":
     LIB_DIRS.append(str(libtiledbsoma_dir / "lib" / "x86_64-linux-gnu"))

diff --git a/apis/r/configure b/apis/r/configure
@@ -49,7 +49,7 @@ tools/build_libtiledbsoma.sh
 pkgincl="-I../inst/tiledb/include -I../inst/tiledbsoma/include"
 pkglibs="-ltiledb -L../inst/tiledb/lib -ltiledbsoma -L../inst/tiledbsoma/lib"
 rpath="-Wl,-rpath,'\$\$ORIGIN/../tiledb/lib' -Wl,-rpath,'\$\$ORIGIN/../tiledbsoma/lib'"
-macosver=`${R_HOME}/bin/Rscript -e 'if (Sys.info()["machine"] == "x86_64" && Sys.info()["sysname"] == "Darwin") cat("-mmacosx-version-min=10.14") else cat("")'`
+macosver=`${R_HOME}/bin/Rscript -e 'if (Sys.info()["sysname"] == "Darwin") cat("-mmacosx-version-min=11.0") else cat("")'`
 
 sed -e "s|@tiledb_include@|$pkgincl |" \
     -e "s|@tiledb_libs@|$pkglibs|" \

diff --git a/libtiledbsoma/src/soma/soma_array.cc b/libtiledbsoma/src/soma/soma_array.cc
@@ -164,7 +164,7 @@ void SOMAArray::fill_metadata_cache() {
     }
 }
 
-const std::string& SOMAArray::uri() const {
+const std::string SOMAArray::uri() const {
     return uri_;
 };
 
@@ -189,6 +189,7 @@ void SOMAArray::open(
         arr_->close();
         arr_->open(tdb_mode);
     }
+    reset(column_names(), batch_size_, result_order_);
 }
 
 void SOMAArray::close() {

diff --git a/libtiledbsoma/src/soma/soma_array.h b/libtiledbsoma/src/soma/soma_array.h
@@ -43,11 +43,12 @@
 #include "enums.h"
 #include "logger_public.h"
 #include "managed_query.h"
+#include "soma_object.h"
 
 namespace tiledbsoma {
 using namespace tiledb;
 
-class SOMAArray {
+class SOMAArray : public SOMAObject {
    public:
     //===================================================================
     //= public static
@@ -167,17 +168,36 @@ class SOMAArray {
         ResultOrder result_order,
         std::optional<std::pair<uint64_t, uint64_t>> timestamp = std::nullopt);
 
+    SOMAArray(const SOMAArray& other)
+        : ctx_(other.ctx_)
+        , config_(other.config_)
+        , uri_(other.uri_)
+        , name_(other.name_)
+        , batch_size_(other.batch_size_)
+        , result_order_(other.result_order_)
+        , metadata_(other.metadata_)
+        , timestamp_(other.timestamp_)
+        , mq_(std::make_unique<ManagedQuery>(
+              other.arr_, other.ctx_, other.name_))
+        , arr_(other.arr_)
+        , first_read_next_(other.first_read_next_)
+        , submitted_(other.submitted_) {
+    }
+
+    SOMAArray(const SOMAObject& other)
+        : SOMAObject(other) {
+    }
+
     SOMAArray() = delete;
-    SOMAArray(const SOMAArray&) = delete;
-    SOMAArray(SOMAArray&&) = default;
+    SOMAArray(SOMAArray&&) = delete;
     ~SOMAArray() = default;
 
     /**
      * @brief Get URI of the SOMAArray.
      *
      * @return std::string URI
      */
-    const std::string& uri() const;
+    const std::string uri() const;
 
     /**
      * @brief Get Ctx of the SOMAArray.
@@ -193,19 +213,6 @@ class SOMAArray {
      */
     std::map<std::string, std::string> config();
 
-    std::optional<std::string> soma_object_type() {
-        auto soma_object_type = this->get_metadata("soma_object_type");
-
-        if (!soma_object_type.has_value())
-            return std::nullopt;
-
-        const char* dtype = (const char*)std::get<MetadataInfo::value>(
-            *soma_object_type);
-        uint32_t sz = std::get<MetadataInfo::num>(*soma_object_type);
-
-        return std::string(dtype, sz);
-    }
-
     /**
      * Open the SOMAArray object.
      *
@@ -709,6 +716,9 @@ class SOMAArray {
     // SOMAArray URI
     std::string uri_;
 
+    // SOMAArray name for debugging
+    std::string_view name_;
+
     // Batch size
     std::string batch_size_;
 

diff --git a/libtiledbsoma/src/soma/soma_collection.cc b/libtiledbsoma/src/soma/soma_collection.cc
@@ -114,19 +114,22 @@ void SOMACollection::set(
 
 std::shared_ptr<SOMAObject> SOMACollection::get(const std::string& key) {
     auto member = group_->get_member(key);
-    std::string soma_object_type = this->type();
+    std::optional<std::string> soma_object_type = this->type();
 
-    if (soma_object_type.compare("SOMACollection") == 0)
+    if (!soma_object_type)
+        throw TileDBSOMAError("Saw invalid SOMA object.");
+
+    if (soma_object_type->compare("SOMACollection") == 0)
         return SOMACollection::open(member.uri(), OpenMode::read);
-    else if (soma_object_type.compare("SOMAExperiment") == 0)
+    else if (soma_object_type->compare("SOMAExperiment") == 0)
         return SOMAExperiment::open(member.uri(), OpenMode::read);
-    else if (soma_object_type.compare("SOMAMeasurement") == 0)
+    else if (soma_object_type->compare("SOMAMeasurement") == 0)
         return SOMAMeasurement::open(member.uri(), OpenMode::read);
-    else if (soma_object_type.compare("SOMADataFrame") == 0)
+    else if (soma_object_type->compare("SOMADataFrame") == 0)
         return SOMADataFrame::open(member.uri(), OpenMode::read);
-    else if (soma_object_type.compare("SOMASparseNDArray") == 0)
+    else if (soma_object_type->compare("SOMASparseNDArray") == 0)
         return SOMASparseNDArray::open(member.uri(), OpenMode::read);
-    else if (soma_object_type.compare("SOMADenseNDArray") == 0)
+    else if (soma_object_type->compare("SOMADenseNDArray") == 0)
         return SOMADenseNDArray::open(member.uri(), OpenMode::read);
 
     throw TileDBSOMAError("Saw invalid SOMA object.");

diff --git a/libtiledbsoma/src/soma/soma_collection.h b/libtiledbsoma/src/soma/soma_collection.h
@@ -153,15 +153,6 @@ class SOMACollection : public SOMAObject {
         return group_->is_open();
     }
 
-    /**
-     * Return the constant "SOMACollection".
-     *
-     * @return std::string
-     */
-    const std::string type() const {
-        return "SOMACollection";
-    }
-
     /**
      * Get the SOMACollection URI.
      */

diff --git a/libtiledbsoma/src/soma/soma_dataframe.cc b/libtiledbsoma/src/soma/soma_dataframe.cc
@@ -30,10 +30,6 @@
  *   This file defines the SOMADataFrame class.
  */
 
-#include <filesystem>
-
-#include <tiledb/tiledb>
-#include "array_buffers.h"
 #include "soma_dataframe.h"
 
 namespace tiledbsoma {
@@ -84,96 +80,28 @@ std::unique_ptr<SOMADataFrame> SOMADataFrame::open(
         mode, uri, ctx, column_names, result_order, timestamp);
 }
 
-//===================================================================
-//= public non-static
-//===================================================================
-
-SOMADataFrame::SOMADataFrame(
-    OpenMode mode,
-    std::string_view uri,
-    std::shared_ptr<Context> ctx,
-    std::vector<std::string> column_names,
-    ResultOrder result_order,
-    std::optional<std::pair<uint64_t, uint64_t>> timestamp) {
-    std::string array_name = std::filesystem::path(uri).filename().string();
-    array_ = std::make_shared<SOMAArray>(
-        mode,
-        uri,
-        array_name,  // label used when debugging
-        ctx,
-        column_names,
-        "auto",  // batch_size,
-        result_order,
-        timestamp);
-}
-
-void SOMADataFrame::open(
-    OpenMode mode, std::optional<std::pair<uint64_t, uint64_t>> timestamp) {
-    array_->open(mode, timestamp);
-}
-
-void SOMADataFrame::close() {
-    array_->close();
-}
-
 bool SOMADataFrame::exists(std::string_view uri) {
     try {
-        auto soma_dataframe = SOMADataFrame::open(uri, OpenMode::read);
-        auto soma_object_type = soma_dataframe->get_metadata(
-            "soma_object_type");
-
-        if (!soma_object_type.has_value())
-            return false;
-
-        const char* dtype = (const char*)std::get<MetadataInfo::value>(
-            *soma_object_type);
-
-        uint32_t sz = std::get<MetadataInfo::num>(*soma_object_type);
-
-        return std::string(dtype, sz) == "SOMADataFrame";
-    } catch (std::exception& e) {
+        return "SOMADataFrame" == SOMAObject::open(uri, OpenMode::read)->type();
+    } catch (TileDBSOMAError& e) {
         return false;
     }
 }
 
-bool SOMADataFrame::is_open() const {
-    return array_->is_open();
-}
-
-const std::string SOMADataFrame::uri() const {
-    return array_->uri();
-}
-
-std::shared_ptr<Context> SOMADataFrame::ctx() {
-    return array_->ctx();
-}
-
-std::map<std::string, std::string> SOMADataFrame::config() {
-    return array_->config();
-}
+//===================================================================
+//= public non-static
+//===================================================================
 
 std::unique_ptr<ArrowSchema> SOMADataFrame::schema() const {
-    return array_->arrow_schema();
-}
-
-std::shared_ptr<ArraySchema> SOMADataFrame::tiledb_schema() const {
-    return array_->tiledb_schema();
+    return this->arrow_schema();
 }
 
 const std::vector<std::string> SOMADataFrame::index_column_names() const {
-    return array_->dimension_names();
-}
-
-int64_t SOMADataFrame::count() const {
-    return array_->nnz();
-}
-
-std::optional<std::shared_ptr<ArrayBuffers>> SOMADataFrame::read_next() {
-    return array_->read_next();
+    return this->dimension_names();
 }
 
-void SOMADataFrame::write(std::shared_ptr<ArrayBuffers> buffers) {
-    array_->write(buffers);
+uint64_t SOMADataFrame::count() {
+    return this->nnz();
 }
 
 }  // namespace tiledbsoma