From 2d8ae0b40bdfb75bcceff0c73d41a52d4bffb5dc Mon Sep 17 00:00:00 2001
From: Alec Koumjian <akoumjian@gmail.com>
Date: Tue, 14 Jan 2025 11:57:48 -0500
Subject: [PATCH] Assume attributes from first value when concatenating only
 empty tables

---
 quivr/concat.py     | 12 ++++++++++--
 test/test_concat.py | 27 +++++++++++++++++++++++++++
 2 files changed, 37 insertions(+), 2 deletions(-)

diff --git a/quivr/concat.py b/quivr/concat.py
index 2aad710..34d73cd 100644
--- a/quivr/concat.py
+++ b/quivr/concat.py
@@ -29,6 +29,9 @@ def concatenate(
     if len(values_list) == 0:
         raise ValueError("No values to concatenate")
 
+    # Note, we don't return immediately if there is only one table,
+    # because we still want to optionally defragment the result.
+
     batches = []
     first_full = False
 
@@ -58,10 +61,15 @@ def concatenate(
             )
 
     if len(batches) == 0:
-        return first_cls.empty()
+        # Return the first table, to preserve the attributes
+        table = first_val.table
+    else:
+        table = pa.Table.from_batches(batches)
 
-    table = pa.Table.from_batches(batches)
+    # We re-initialize the table to optionally validate and create
+    # a unique object
     result = first_cls.from_pyarrow(table=table, validate=validate)
+
     if defrag:
         result = defragment.defragment(result)
     return result
diff --git a/test/test_concat.py b/test/test_concat.py
index d4a9160..dd999b1 100644
--- a/test/test_concat.py
+++ b/test/test_concat.py
@@ -147,3 +147,30 @@ def test_concatenate_same_attrs():
 def test_concatenate_no_values():
     with pytest.raises(ValueError, match="No values to concatenate"):
         qv.concatenate([])
+
+
+def test_concatenate_empty_tables_preserve_attributes():
+    class TableWithAttrs(qv.Table):
+        x = qv.Int64Column()
+        y = qv.Int64Column()
+        name = qv.StringAttribute(default="default")
+        id = qv.IntAttribute(default=0)
+
+    # Create two empty tables with non-default attributes
+    t1 = TableWithAttrs.from_kwargs(x=[], y=[], name="foo", id=1)
+    t2 = TableWithAttrs.from_kwargs(x=[], y=[], name="bat", id=3)
+    
+    # Concatenate them and verify we get an empty table with the same attributes
+    have = qv.concatenate([t1, t2])
+    assert len(have) == 0
+    assert have.name == "foo"  # Not "default"
+    assert have.id == 1  # Not 0
+
+    # Also verify it works when concatenating with a non-empty table
+    # Attributes should be preserved from the non-empty tables
+    t3 = TableWithAttrs.from_kwargs(x=[1], y=[2], name="bar", id=2)
+    have = qv.concatenate([t1, t2, t3])
+    assert len(have) == 1
+    assert have.name == "bar"
+    assert have.id == 2
+