[c++] Arrow utils with current-domain option [WIP]

apis/r/tests/testthat/helper-test-data.R

@@ -75,6 +75,6 @@ create_arrow_table <- function(nrows = 10L, factors = FALSE) {
       soma_joinid = bit64::seq.integer64(from = 0L, to = nrows - 1L),
       bar = seq(nrows) + 0.1,
       baz = as.character(seq.int(nrows) + 1000L)
-      # schema = create_arrow_schema()
+      # schema = create_arrow_schema(false)
     )
 }

libtiledbsoma/src/soma/soma_array.cc

@@ -1178,6 +1178,26 @@ uint64_t SOMAArray::nnz_slow() {
 }
 
 std::vector<int64_t> SOMAArray::shape() {
+    // Two reasons for this:
+    // * Transitional, non-monolithic, phased, careful development for the
+    //   new-shape feature
+    // * Even after the new-shape feature is fully released, there will be old
+    //   arrays on disk that were created before this feature existed.
+    // So this is long-term code.
+    auto current_domain = tiledb::ArraySchemaExperimental::current_domain(
+        *ctx_->tiledb_ctx(), arr_->schema());
+    if (current_domain.is_empty()) {
+        return _tiledb_domain();
+    } else {
+        return _tiledb_current_domain();
+    }
+}
+
+std::vector<int64_t> SOMAArray::maxshape() {
+    return _tiledb_domain();
+}
+
+std::vector<int64_t> SOMAArray::_tiledb_domain() {
     std::vector<int64_t> result;
     auto dimensions = mq_->schema()->domain().dimensions();
 
@@ -1253,6 +1273,71 @@ std::vector<int64_t> SOMAArray::shape() {
     return result;
 }
 
+std::vector<int64_t> SOMAArray::_tiledb_current_domain() {
+    std::vector<int64_t> result;
+
+    auto current_domain = tiledb::ArraySchemaExperimental::current_domain(
+        *ctx_->tiledb_ctx(), arr_->schema());
+
+    if (current_domain.is_empty()) {
+        throw TileDBSOMAError(
+            "Internal error: current domain requested for an array which does "
+            "not support it");
+    }
+
+    auto t = current_domain.type();
+    if (t != TILEDB_NDRECTANGLE) {
+        throw TileDBSOMAError("current_domain type is not NDRECTANGLE");
+    }
+
+    NDRectangle ndrect = current_domain.ndrectangle();
+
+    for (auto dimension_name : dimension_names()) {
+        // TODO: non-int64 types for SOMADataFrame extra dims.
+        // This simply needs to be integrated with switch statements as in the
+        // legacy code below.
+        auto range = ndrect.range<int64_t>(dimension_name);
+        result.push_back(range[1] + 1);
+    }
+    return result;
+}
+
+void SOMAArray::resize(const std::vector<int64_t>& newshape) {
+    if (mq_->query_type() != TILEDB_WRITE) {
+        throw TileDBSOMAError(
+            "[SOMAArray::resize] array must be opened in write mode");
+    }
+
+    auto tctx = ctx_->tiledb_ctx();
+    ArraySchema schema = arr_->schema();
+    Domain domain = schema.domain();
+    ArraySchemaEvolution schema_evolution(*tctx);
+    CurrentDomain new_current_domain(*tctx);
+
+    NDRectangle ndrect(*tctx, domain);
+
+    // TODO: non-int64 for DataFrame when it has extra index dims.
+    // This will be via a resize-helper.
+
+    unsigned n = domain.ndim();
+    if ((unsigned)newshape.size() != n) {
+        throw TileDBSOMAError(fmt::format(
+            "[SOMAArray::resize]: newshape has dimension count {}; array has "
+            "{} ",
+            newshape.size(),
+            n));
+    }
+
+    for (unsigned i = 0; i < n; i++) {
+        ndrect.set_range<int64_t>(
+            domain.dimension(i).name(), 0, newshape[i] - 1);
+    }
+
+    new_current_domain.set_ndrectangle(ndrect);
+    schema_evolution.expand_current_domain(new_current_domain);
+    schema_evolution.array_evolve(uri_);
+}
+
 uint64_t SOMAArray::ndim() const {
     return tiledb_schema()->domain().ndim();
 }

libtiledbsoma/src/soma/soma_array.h

@@ -571,13 +571,54 @@ class SOMAArray : public SOMAObject {
     }
 
     /**
-     * @brief Get the capacity of each dimension.
+     * @brief Get the current capacity of each dimension.
+     *
+     * This applies to arrays all of whose dims are of type int64_t: this
+     * includes SOMASparseNDArray and SOMADenseNDArray, and default-indexed
+     * SOMADataFrame.
+     *
+     * At the TileDB-SOMA level we call this "shape". At the TileDB Core
+     * storage level this maps to "current domain".
+     *
+     * Further, we map this single n to the pair (0, n-1) since core permits a
+     * doubly inclusive pair (lo, hi) on each dimension slot.
      *
      * @return A vector with length equal to the number of dimensions; each
-     * value in the vector is the capcity of each dimension.
+     * value in the vector is the capacity of each dimension.
      */
     std::vector<int64_t> shape();
 
+    /**
+     * @brief Get the maximum resizable capacity of each dimension.
+     *
+     * This applies to arrays all of whose dims are of type int64_t: this
+     * includes SOMASparseNDArray and SOMADenseNDArray, and default-indexed
+     * SOMADataFrame.
+     *
+     * At the TileDB-SOMA level we call this "maxshape". At the TileDB Core
+     * storage level this maps to "domain".
+     *
+     * Further, we map this single n to the pair (0, n-1) since core permits a
+     * doubly inclusive pair (lo, hi) on each dimension slot.
+     *
+     * @return A vector with length equal to the number of dimensions; each
+     * value in the vector is the maximum capacity of each dimension.
+     */
+    std::vector<int64_t> maxshape();
+
+    /**
+     * @brief Resize the shape (what core calls "current domain") up to the
+     * maxshape (what core calls "domain").
+     *
+     * This applies to arrays all of whose dims are of type int64_t: this
+     * includes SOMASparseNDArray and SOMADenseNDArray, and default-indexed
+     * SOMADataFrame.
+     *
+     * @return Nothing. Raises an exception if the resize would be a downsize,
+     * which is not supported.
+     */
+    void resize(const std::vector<int64_t>& newshape);
+
     /**
      * @brief Get the number of dimensions.
      *
@@ -762,6 +803,18 @@ class SOMAArray : public SOMAObject {
 
     uint64_t _get_max_capacity(tiledb_datatype_t index_type);
 
+    /**
+     * With old shape: core domain mapped to tiledbsoma shape; core current
+     * domain did not exist.
+     *
+     * With new shape: core domain maps to tiledbsoma maxshape;
+     * core current_domain maps to tiledbsoma shape.
+     *
+     * Here we distinguish between user-side API, and core-side implementation.
+     */
+    std::vector<int64_t> _tiledb_domain();
+    std::vector<int64_t> _tiledb_current_domain();
+
     bool _extend_enumeration(
         ArrowSchema* value_schema,
         ArrowArray* value_array,

libtiledbsoma/src/soma/soma_collection.h

@@ -281,4 +281,4 @@ class SOMACollection : public SOMAGroup {
 };
 }  // namespace tiledbsoma
 
-#endif  // SOMA_COLLECTION
+#endif  // SOMA_COLLECTION