From 1bf7b100cf233a07124b0ced8894de234840ad7b Mon Sep 17 00:00:00 2001
From: John Kerl <kerl.john.r@gmail.com>
Date: Sun, 18 Aug 2024 18:06:33 -0400
Subject: [PATCH] [c++] Resize for variant-indexed `DataFrame` [WIP] [skip ci]

---
 libtiledbsoma/src/soma/soma_array.cc | 37 ++++++++++++++++++++++++++++
 libtiledbsoma/src/soma/soma_array.h  | 16 ++++++++++++
 2 files changed, 53 insertions(+)

diff --git a/libtiledbsoma/src/soma/soma_array.cc b/libtiledbsoma/src/soma/soma_array.cc
index 092af65253..83d43a3226 100644
--- a/libtiledbsoma/src/soma/soma_array.cc
+++ b/libtiledbsoma/src/soma/soma_array.cc
@@ -1307,6 +1307,43 @@ void SOMAArray::resize(const std::vector<int64_t>& newshape) {
     schema_evolution.array_evolve(uri_);
 }
 
+void SOMAArray::resize_soma_joinid_if_dim(
+    const std::vector<int64_t>& newshape) {
+    if (mq_->query_type() != TILEDB_WRITE) {
+        throw TileDBSOMAError(
+            "[SOMAArray::resize] array must be opened in write mode");
+    }
+
+    ArraySchema schema = arr_->schema();
+    Domain domain = schema.domain();
+    unsigned ndim = domain.ndim();
+    if (newshape.size() != 1) {
+        throw TileDBSOMAError(fmt::format(
+            "[SOMAArray::resize]: newshape has dimension count {}; needed 1",
+            newshape.size(),
+            ndim));
+    }
+
+    auto tctx = ctx_->tiledb_ctx();
+    CurrentDomain old_current_domain = ArraySchemaExperimental::current_domain(
+        *tctx, schema);
+    NDRectangle ndrect = old_current_domain.ndrectangle();
+
+    CurrentDomain new_current_domain(*tctx);
+    ArraySchemaEvolution schema_evolution(*tctx);
+
+    for (unsigned i = 0; i < ndim; i++) {
+        if (domain.dimension(i).name() == "soma_joinid") {
+            ndrect.set_range<int64_t>(
+                domain.dimension(i).name(), 0, newshape[0] - 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();
 }
diff --git a/libtiledbsoma/src/soma/soma_array.h b/libtiledbsoma/src/soma/soma_array.h
index 47e272bd6e..dfd9f0d5ef 100644
--- a/libtiledbsoma/src/soma/soma_array.h
+++ b/libtiledbsoma/src/soma/soma_array.h
@@ -619,6 +619,22 @@ class SOMAArray : public SOMAObject {
      */
     void resize(const std::vector<int64_t>& newshape);
 
+    /**
+     * @brief Increases the tiledbsoma shape up to at most the maxshape,
+     * resizing the soma_joinid dimension if it is a dimension.
+     *
+     * While SOMA SparseNDArray and DenseNDArray, along with default-indexed
+     * DataFrame, have int64_t dims, non-default-indexed DataFrame objects need
+     * not: it is only required that they have a dim _or_ an attr called
+     * soma_joinid. If soma_joinid is one of the dims, it will be resized while
+     * the others will be preserved. If soma_joinid is not one of the dims,
+     * nothing will be changed, as nothing _needs_ to be changed.
+     *
+     * @return Throws if the requested shape exceeds the array's create-time
+     * maxshape. Throws if the array does not have current-domain support.
+     */
+    void resize_soma_joinid_if_dim(const std::vector<int64_t>& newshape);
+
     /**
      * @brief Get the number of dimensions.
      *