From fd21999860a1829677797b4bef33050c344eeb8a Mon Sep 17 00:00:00 2001
From: Igor <igor.niebylski@gmail.com>
Date: Mon, 23 Jun 2025 09:35:09 +0200
Subject: [PATCH 1/3] evaluate stream_dataframe before concat and check if its
 empty

---
 quasardb/pandas/__init__.py | 11 +++++++----
 tests/test_pandas.py        | 18 ++++++++++++++++++
 2 files changed, 25 insertions(+), 4 deletions(-)

diff --git a/quasardb/pandas/__init__.py b/quasardb/pandas/__init__.py
index 031e2d42..b24285b0 100644
--- a/quasardb/pandas/__init__.py
+++ b/quasardb/pandas/__init__.py
@@ -344,10 +344,13 @@ def read_dataframe(conn: quasardb.Cluster, table, **kwargs):
         )
         kwargs["batch_size"] = 2**16
 
-    # Note that this is *lazy*, dfs is a generator, not a list -- as such, dataframes will be
-    # fetched on-demand, which means that an error could occur in the middle of processing
-    # dataframes.
-    dfs = stream_dataframe(conn, table, **kwargs)
+    # stream_dataframe is *lazy* - is a generator, not a list.
+    # if result of stream_dataframe is empty this could result in ValueError on pd.concat()
+    # we need to evaluate the generator first, and then concatenate if result set is not empty.
+    dfs = list(stream_dataframe(conn, table, **kwargs))
+
+    if len(dfs) == 0:
+        return pd.DataFrame()
 
     return pd.concat(dfs)
 
diff --git a/tests/test_pandas.py b/tests/test_pandas.py
index f9a2b0d2..d2bcdeb7 100644
--- a/tests/test_pandas.py
+++ b/tests/test_pandas.py
@@ -665,3 +665,21 @@ def test_retries(
             )
 
         assert seen == True
+
+
+def test_read_dataframe_empty_table_sc16881(qdbd_connection, table_name):
+    """
+    If results of qdbpd.read_dataframe are empty it would raise ValueError on pd.concat().
+    It should return an empty DataFrame instead.
+    """
+    table = qdbd_connection.ts(table_name)
+
+    table_config = [
+        quasardb.ColumnInfo(quasardb.ColumnType.Double, "d"),
+    ]
+
+    table.create(table_config)
+
+    df = qdbpd.read_dataframe(qdbd_connection, table)
+
+    assert df.empty

From 69679dcb78df684e5dff25161ac13fb51738c4dd Mon Sep 17 00:00:00 2001
From: Igor <igor.niebylski@gmail.com>
Date: Mon, 23 Jun 2025 09:42:48 +0200
Subject: [PATCH 2/3] add better test description

---
 tests/test_pandas.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/tests/test_pandas.py b/tests/test_pandas.py
index d2bcdeb7..25c505f6 100644
--- a/tests/test_pandas.py
+++ b/tests/test_pandas.py
@@ -669,8 +669,9 @@ def test_retries(
 
 def test_read_dataframe_empty_table_sc16881(qdbd_connection, table_name):
     """
-    If results of qdbpd.read_dataframe are empty it would raise ValueError on pd.concat().
-    It should return an empty DataFrame instead.
+    Ensures qdbpd.read_dataframe returns an empty DataFrame when the table exists but contains no data.
+
+    Previously this raised ValueError due to pd.concat() on an empty result set.
     """
     table = qdbd_connection.ts(table_name)
 

From 257993c47f34064658794e888f07b5ca9c0fde2e Mon Sep 17 00:00:00 2001
From: Igor <igor.niebylski@gmail.com>
Date: Tue, 24 Jun 2025 12:46:32 +0200
Subject: [PATCH 3/3] simplify

---
 quasardb/pandas/__init__.py | 21 ++++++++++++++-------
 1 file changed, 14 insertions(+), 7 deletions(-)

diff --git a/quasardb/pandas/__init__.py b/quasardb/pandas/__init__.py
index b24285b0..224d0a9f 100644
--- a/quasardb/pandas/__init__.py
+++ b/quasardb/pandas/__init__.py
@@ -344,16 +344,23 @@ def read_dataframe(conn: quasardb.Cluster, table, **kwargs):
         )
         kwargs["batch_size"] = 2**16
 
-    # stream_dataframe is *lazy* - is a generator, not a list.
-    # if result of stream_dataframe is empty this could result in ValueError on pd.concat()
-    # we need to evaluate the generator first, and then concatenate if result set is not empty.
-    dfs = list(stream_dataframe(conn, table, **kwargs))
+    # Note that this is *lazy*, dfs is a generator, not a list -- as such, dataframes will be
+    # fetched on-demand, which means that an error could occur in the middle of processing
+    # dataframes.
+    dfs = stream_dataframe(conn, table, **kwargs)
 
-    if len(dfs) == 0:
+    # if result of stream_dataframe is empty this could result in ValueError on pd.concat()
+    # as stream_dataframe is a generator there is no easy way to check for this condition without evaluation
+    # the most simple way is to catch the ValueError and return an empty DataFrame
+    try:
+        return pd.concat(dfs, copy=False)
+    except ValueError as e:
+        logger.error(
+            "Error while concatenating dataframes. This can happen if result set is empty. Returning empty dataframe. Error: %s",
+            e,
+        )
         return pd.DataFrame()
 
-    return pd.concat(dfs)
-
 
 def _extract_columns(df, cinfos):
     """