From a8364a13fa9a2cdaceffb10cf1cd986d3bdb38c6 Mon Sep 17 00:00:00 2001
From: Agis Kounelis <kounelisagis@gmail.com>
Date: Thu, 11 Apr 2024 23:18:50 +0300
Subject: [PATCH] Wrap tiledb_array_consolidate_fragments from pybind11

---
 tiledb/cc/array.cc         |  8 ++++++++
 tiledb/tests/cc/test_cc.py | 32 ++++++++++++++++++++++++++++++++
 2 files changed, 40 insertions(+)

diff --git a/tiledb/cc/array.cc b/tiledb/cc/array.cc
index 5fd55e9b73..7ace98ac5e 100644
--- a/tiledb/cc/array.cc
+++ b/tiledb/cc/array.cc
@@ -83,6 +83,14 @@ void init_array(py::module &m) {
       // TODO non_empty_domain_var
 
       .def("query_type", &Array::query_type)
+      .def("consolidate_fragments", [](Array &self, const Context &ctx, const std::vector<std::string> &fragment_uris, Config *config) {
+             std::vector<const char *> c_strings;
+             c_strings.reserve(fragment_uris.size());
+             for(const auto& str : fragment_uris) {
+               c_strings.push_back(str.c_str());
+             }
+             ctx.handle_error(tiledb_array_consolidate_fragments(ctx.ptr().get(), self.uri().c_str(), c_strings.data(), fragment_uris.size(), config->ptr().get()));
+           })
       .def("consolidate_metadata",
            py::overload_cast<const Context &, const std::string &,
                              tiledb_encryption_type_t, const std::string &,
diff --git a/tiledb/tests/cc/test_cc.py b/tiledb/tests/cc/test_cc.py
index 4bd6e0813c..e75f067bad 100644
--- a/tiledb/tests/cc/test_cc.py
+++ b/tiledb/tests/cc/test_cc.py
@@ -7,6 +7,7 @@
 import tiledb
 import tiledb.cc as lt
 from tiledb.datatypes import DataType
+from tiledb.main import PyFragmentInfo
 
 
 def test_config():
@@ -160,6 +161,37 @@ def test_array():
     arr.close()
 
 
+def test_consolidate_fragments():
+    uri = tempfile.mkdtemp()
+    ctx = lt.Context()
+    config = lt.Config()
+
+    tiledb.from_numpy(uri, np.random.rand(4)).close()
+
+    with tiledb.open(uri, "w") as A:
+        A[:] = np.random.rand(4)
+
+    with tiledb.open(uri, "w") as A:
+        A[:] = np.random.rand(4)
+
+    schema = tiledb.ArraySchema.load(uri, ctx=ctx)
+    fragment_info = PyFragmentInfo(uri, schema, False, ctx)
+
+    assert fragment_info.get_num_fragments() == 3
+
+    uris = fragment_info.get_uri()
+    # get fragment name form alone, not the full path(s) (the part of each uri after the last /)
+    # https://github.com/TileDB-Inc/TileDB-Py/pull/1946
+    uris = [uri.split("/")[-1] for uri in uris]
+
+    arr = lt.Array(ctx, uri, lt.QueryType.WRITE)
+    arr.consolidate_fragments(ctx, uris, config)
+    arr.close()
+
+    fragment_info = PyFragmentInfo(uri, schema, False, ctx)
+    assert fragment_info.get_num_fragments() == 1
+
+
 def test_array_config():
     uri = tempfile.mkdtemp()