test(spark): handle remote skipping when asserting facts about local files

Author: cpcloud

ibis/backends/pyspark/tests/test_import_export.py

@@ -8,6 +8,7 @@
 from pandas.testing import assert_frame_equal
 
 from ibis.backends.pyspark.datatypes import PySparkSchema
+from ibis.conftest import IS_SPARK_REMOTE
 
 
 @pytest.mark.parametrize(
@@ -76,6 +77,9 @@ def test_to_parquet_dir(con_streaming, tmp_path):
     assert len(df) == 5
 
 
+@pytest.mark.skipif(
+    IS_SPARK_REMOTE, reason="Spark remote does not support assertions about local paths"
+)
 def test_to_parquet_read_parquet(con, tmp_path):
     # No Partitions
     t_out = con.table("awards_players")

ibis/backends/tests/test_export.py

@@ -207,6 +207,8 @@ def test_to_pyarrow_batches_memtable(con):
 
 
 def test_table_to_parquet(tmp_path, backend, awards_players):
+    if backend.name() == "pyspark" and IS_SPARK_REMOTE:
+        pytest.skip("writes to remote output directory")
     outparquet = tmp_path / "out.parquet"
     awards_players.to_parquet(outparquet)
 
@@ -258,6 +260,8 @@ def test_table_to_parquet_writer_kwargs(version, tmp_path, backend, awards_playe
     awards_players.to_parquet(outparquet, version=version)
 
     if backend.name() == "pyspark":
+        if IS_SPARK_REMOTE:
+            pytest.skip("writes to remote output directory")
         # Pyspark will write more than one parquet file under outparquet as directory
         parquet_files = sorted(outparquet.glob("*.parquet"))
         df = (
@@ -349,6 +353,8 @@ def test_memtable_to_file(tmp_path, con, ftype, monkeypatch):
     getattr(con, f"to_{ftype}")(memtable, outfile)
 
     if con.name == "pyspark" and ftype == "parquet":
+        if IS_SPARK_REMOTE:
+            pytest.skip("writes to remote output directory")
         assert outfile.is_dir()
     else:
         assert outfile.is_file()