add logic for resuming after a resume.

test/Wikiq_Unit_Test.py

@@ -1043,3 +1043,43 @@ def test_resume_simple():
     assert_frame_equal(resumed_df, full_df, check_like=True, check_dtype=False)
 
     print(f"Resume simple test passed! Original: {len(full_df)} rows, Resumed: {len(resumed_df)} rows")
+
+
+def test_resume_merge_with_invalid_temp_file():
+    """Test that resume handles invalid/empty temp files gracefully.
+
+    This can happen when a namespace has no records after the resume point,
+    resulting in a temp file that was created but never written to.
+    """
+    import pyarrow.parquet as pq
+    from wikiq.resume import merge_parquet_files, merge_partitioned_namespaces
+    import tempfile
+
+    # Create a valid parquet file
+    with tempfile.TemporaryDirectory() as tmpdir:
+        original_path = os.path.join(tmpdir, "original.parquet")
+        temp_path = os.path.join(tmpdir, "temp.parquet")
+        merged_path = os.path.join(tmpdir, "merged.parquet")
+
+        # Create a valid original file
+        import pyarrow as pa
+        table = pa.table({"articleid": [1, 2, 3], "revid": [10, 20, 30]})
+        pq.write_table(table, original_path)
+
+        # Create an invalid temp file (empty file, not valid parquet)
+        with open(temp_path, 'w') as f:
+            f.write("")
+
+        # merge_parquet_files should return None for invalid temp file
+        result = merge_parquet_files(original_path, temp_path, merged_path)
+        assert result is None, "Expected None when temp file is invalid"
+
+        # Original file should still exist and be unchanged
+        assert os.path.exists(original_path), "Original file should still exist"
+        original_table = pq.read_table(original_path)
+        assert len(original_table) == 3, "Original file should be unchanged"
+
+        # Merged file should not have been created
+        assert not os.path.exists(merged_path), "Merged file should not be created"
+
+    print("Resume merge with invalid temp file test passed!")