From 3114285e0f197be2c9bee99b56bac7cf7c52582e Mon Sep 17 00:00:00 2001
From: Christopher Harris <xixonia@gmail.com>
Date: Thu, 22 Aug 2024 15:56:15 +0000
Subject: [PATCH] add temporary cudf string copy_range workaround

---
 python/morpheus/morpheus/_lib/src/messages/multi.cpp | 11 ++++++++++-
 python/morpheus/morpheus/messages/multi_message.py   | 11 ++++++++++-
 tests/conftest.py                                    |  5 +++++
 3 files changed, 25 insertions(+), 2 deletions(-)

diff --git a/python/morpheus/morpheus/_lib/src/messages/multi.cpp b/python/morpheus/morpheus/_lib/src/messages/multi.cpp
index 6e42e839d7..72fddfa8ec 100644
--- a/python/morpheus/morpheus/_lib/src/messages/multi.cpp
+++ b/python/morpheus/morpheus/_lib/src/messages/multi.cpp
@@ -397,7 +397,16 @@ void MultiMessageInterfaceProxy::set_meta(MultiMessage& self, pybind11::object c
         }
 
         // Perform the update via slices
-        df.attr("loc")[pybind11::make_tuple(df.attr("index")[row_indexer], columns)] = value;
+        auto is_string_dtype = pybind11::module_::import("cudf.api.types").attr("is_string_dtype");
+        auto series = pybind11::module_::import("cudf").attr("Series");
+
+        if (is_string_dtype(series(value)).cast<bool>()) {
+            df[columns] = pybind11::str();
+            df[columns].attr("iloc")[row_indexer] = value;
+        } else {
+            df.attr("loc")[pybind11::make_tuple(df.attr("index")[row_indexer], columns)] = value;
+        }
+
 
         // Reset the index if we changed it
         if (!saved_index.is_none())
diff --git a/python/morpheus/morpheus/messages/multi_message.py b/python/morpheus/morpheus/messages/multi_message.py
index 44e1bb6cba..eb8f1863bf 100644
--- a/python/morpheus/morpheus/messages/multi_message.py
+++ b/python/morpheus/morpheus/messages/multi_message.py
@@ -291,7 +291,16 @@ def set_meta(self, columns: typing.Union[None, str, typing.List[str]], value):
 
                     saved_index = df.index
                     df.reset_index(drop=True, inplace=True)
-                    df.loc[df.index[row_indexer], columns] = value
+
+                    # TODO: when value is strings, make all values empty strings
+                    from cudf.api.types import is_string_dtype
+
+                    if is_string_dtype(cudf.Series(value)):
+                        df[columns] = ""
+                        df[columns].iloc[row_indexer] = value
+                    else:
+                        df.loc[df.index[row_indexer], columns] = value
+
                     df.set_index(saved_index, inplace=True)
                 else:
                     # Need to determine the boolean mask to use indexes with df.loc
diff --git a/tests/conftest.py b/tests/conftest.py
index 2e4be942ef..8853907bf6 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -25,9 +25,14 @@
 import types
 import typing
 import warnings
+import sys
 from pathlib import Path
 from unittest import mock
 
+import sys
+if "/home/coder" in sys.path:
+    sys.path.remove("/home/coder")
+
 import pytest
 import requests