From c7708406ab8d2c2e8e0b253e99c07053dc6b8aac Mon Sep 17 00:00:00 2001
From: Patrick Hoefler <61934744+phofl@users.noreply.github.com>
Date: Tue, 19 Dec 2023 10:59:31 +0100
Subject: [PATCH] Fixup concat dtypes for empty dataframes (#602)

---
 dask_expr/_concat.py           |  7 ++++++-
 dask_expr/tests/test_concat.py | 11 +++++++++++
 2 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/dask_expr/_concat.py b/dask_expr/_concat.py
index 1301434b..d0ae12f9 100644
--- a/dask_expr/_concat.py
+++ b/dask_expr/_concat.py
@@ -46,9 +46,14 @@ def _frames(self):
 
     @functools.cached_property
     def _meta(self):
+        # ignore DataFrame without columns to avoid dtype upcasting
         meta = make_meta(
             methods.concat(
-                [meta_nonempty(df._meta) for df in self._frames],
+                [
+                    meta_nonempty(df._meta)
+                    for df in self._frames
+                    if df.ndim < 2 or len(df._meta.columns) > 0
+                ],
                 join=self.join,
                 filter_warning=False,
                 axis=self.axis,
diff --git a/dask_expr/tests/test_concat.py b/dask_expr/tests/test_concat.py
index c1cd54f7..505c5363 100644
--- a/dask_expr/tests/test_concat.py
+++ b/dask_expr/tests/test_concat.py
@@ -140,6 +140,17 @@ def test_concat_index(df, pdf):
     assert query._name == expected._name
 
 
+def test_concat_dataframe_empty():
+    df = lib.DataFrame({"a": [100, 200, 300]}, dtype="int64")
+    empty_df = lib.DataFrame([], dtype="int64")
+    df_concat = lib.concat([df, empty_df])
+
+    ddf = from_pandas(df, npartitions=1)
+    empty_ddf = from_pandas(empty_df, npartitions=1)
+    ddf_concat = concat([ddf, empty_ddf])
+    assert_eq(df_concat, ddf_concat)
+
+
 def test_concat_after_merge():
     pdf1 = lib.DataFrame(
         {"x": range(10), "y": [1, 2, 3, 4, 5] * 2, "z": ["cat", "dog"] * 5}