Add tests for handling empty DataFrames and zero-row queries

Author: kosiew

python/tests/test_dataframe.py

@@ -2664,3 +2664,70 @@ def trigger_interrupt():
 
     # Make sure the interrupt thread has finished
     interrupt_thread.join(timeout=1.0)
+
+
+def test_show_from_empty_rows(capsys):
+    """Create a DataFrame with a valid schema but zero rows and call show().
+
+    This verifies that showing an empty-but-schema'd DataFrame does not panic
+    and prints a helpful message instead.
+    """
+    ctx = SessionContext()
+
+    # Create a RecordBatch with the correct schema but zero rows
+    batch = pa.RecordBatch.from_arrays(
+        [pa.array([], type=pa.int64()), pa.array([], type=pa.string())],
+        names=["a", "b"],
+    )
+
+    df = ctx.create_dataframe([[batch]])
+
+    # Call show() and capture stdout
+    df.show()
+    captured = capsys.readouterr()
+
+    # Expect either an explicit no-rows message or an empty DataFrame print
+    out = captured.out.lower()
+    assert "dataframe has no rows" in out or "dataframe()" in out
+
+
+def test_select_where_no_rows(capsys):
+    """Create a DataFrame a:[1,2,3], filter with a>4 to produce zero rows and call show().
+
+    This verifies that a query returning zero rows does not trigger a panic and
+    instead prints a helpful message.
+    """
+    ctx = SessionContext()
+
+    batch = pa.RecordBatch.from_arrays(
+        [pa.array([1, 2, 3], type=pa.int64())], names=["a"]
+    )
+
+    df = ctx.create_dataframe([[batch]])
+
+    df_filtered = df.filter(column("a") > literal(4))
+
+    df_filtered.show()
+    captured = capsys.readouterr()
+    out = captured.out.lower()
+
+    assert "dataframe has no rows" in out or "dataframe()" in out
+
+
+def test_sql_select_where_no_rows(capsys):
+    """Register a table 't' with a:[1,2,3], run SQL that returns no rows, and call show().
+
+    Ensures SQL path that returns zero rows doesn't panic when showing results.
+    """
+    ctx = SessionContext()
+
+    batch = pa.RecordBatch.from_arrays([pa.array([1, 2, 3], type=pa.int64())], names=["a"])
+    ctx.register_record_batches("t", [[batch]])
+
+    df = ctx.sql("SELECT * FROM t WHERE a > 4")
+
+    df.show()
+    captured = capsys.readouterr()
+    out = captured.out.lower()
+
+    assert "dataframe has no rows" in out or "dataframe()" in out