Fix: accept empty arrays in struct field lookup (#997)

* Fix: accept empty arrays in struct field lookup

Fixes #992.

Empty `pyarrow` arrays are considered falsy, which caused a `ResolveError` for required fields during scan operations.

* Integration test: empty scan on non-nullable ordered string column

This covers the issue reported in #992 where empty scan queries yielded a `ResolveError`. Specifically, this occurred under the following conditions:
- a table with an ordered, non-nullable string column
- a scan filtering for a non-existing value _within_ the range of the values in that particular column

* Lint (add missing newline)

Author: grobgl

dev/provision.py

@@ -389,3 +389,13 @@
         VALUES (4)
         """
     )
+
+    spark.sql(
+        f"""
+        CREATE OR REPLACE TABLE {catalog_name}.default.test_empty_scan_ordered_str (id string NOT NULL)
+        USING iceberg
+        TBLPROPERTIES ('format-version'='2')
+        """
+    )
+    spark.sql(f"ALTER TABLE {catalog_name}.default.test_empty_scan_ordered_str WRITE ORDERED BY id")
+    spark.sql(f"INSERT INTO {catalog_name}.default.test_empty_scan_ordered_str VALUES 'a', 'c'")

pyiceberg/io/pyarrow.py

@@ -1590,7 +1590,7 @@ def schema_partner(self, partner: Optional[pa.Array]) -> Optional[pa.Array]:
         return partner
 
     def field_partner(self, partner_struct: Optional[pa.Array], field_id: int, _: str) -> Optional[pa.Array]:
-        if partner_struct:
+        if partner_struct is not None:
             # use the field name from the file schema
             try:
                 name = self.file_schema.find_field(field_id).name

tests/integration/test_reads.py

@@ -663,3 +663,11 @@ def another_task() -> None:
 
         table.transaction().set_properties(lock="xxx").commit_transaction()
         assert table.properties.get("lock") == "xxx"
+
+
+@pytest.mark.integration
+@pytest.mark.parametrize("catalog", [pytest.lazy_fixture("session_catalog_hive"), pytest.lazy_fixture("session_catalog")])
+def test_empty_scan_ordered_str(catalog: Catalog) -> None:
+    table_empty_scan_ordered_str = catalog.load_table("default.test_empty_scan_ordered_str")
+    arrow_table = table_empty_scan_ordered_str.scan(EqualTo("id", "b")).to_arrow()
+    assert len(arrow_table) == 0