Update test_preserve_leading_zeros.py

Author: dxdc

pandas/tests/io/parser/test_preserve_leading_zeros.py

@@ -3,10 +3,6 @@
 import pytest
 
 
-@pytest.mark.xfail(
-    condition=getattr(all_parsers, "engine", "") == "pyarrow",
-    reason="pyarrow engine strips leading zeros even with dtype=str",
-)
 def test_leading_zeros_preserved_with_dtype_str(all_parsers):
     """
     Ensure that all parser engines preserve leading zeros when dtype=str is passed.
@@ -29,43 +25,30 @@ def test_leading_zeros_preserved_with_dtype_str(all_parsers):
         dtype=str,
     )
 
-    # Verify leading zeros are preserved in col2
-    assert result.loc[0, "col2"] == "000388907", (
-        f"Engine {engine_name}: Leading zeros lost in col2, row 0. Got: {result.loc[0, 'col2']}"
-    )
-    assert result.loc[2, "col2"] == "000023607", (
-        f"Engine {engine_name}: Leading zeros lost in col2, row 2. Got: {result.loc[2, 'col2']}"
-    )
-
-    # Verify leading zeros are preserved in col4
-    assert result.loc[0, "col4"] == "0150", (
-        f"Engine {engine_name}: Leading zeros lost in col4, row 0. Got: {result.loc[0, 'col4']}"
-    )
-    assert result.loc[2, "col4"] == "0205", (
-        f"Engine {engine_name}: Leading zeros lost in col4, row 2. Got: {result.loc[2, 'col4']}"
-    )
-
-    # Verify all columns are string type
-    assert result.dtypes["col1"] == "object", (
-        f"Engine {engine_name}: col1 should be string type, got {result.dtypes['col1']}"
-    )
-    assert result.dtypes["col2"] == "object", (
-        f"Engine {engine_name}: col2 should be string type, got {result.dtypes['col2']}"
-    )
-    assert result.dtypes["col3"] == "object", (
-        f"Engine {engine_name}: col3 should be string type, got {result.dtypes['col3']}"
-    )
-    assert result.dtypes["col4"] == "object", (
-        f"Engine {engine_name}: col4 should be string type, got {result.dtypes['col4']}"
-    )
-
-    # Verify shape
-    assert result.shape == (4, 4), (
-        f"Engine {engine_name}: Expected shape (4, 4), got {result.shape}"
-    )
-
-    # Verify column names
-    expected_columns = ["col1", "col2", "col3", "col4"]
-    assert list(result.columns) == expected_columns, (
-        f"Engine {engine_name}: Expected columns {expected_columns}, got {list(result.columns)}"
-    )
+    try:
+        assert result.loc[0, "col2"] == "000388907", (
+            f"{engine_name} lost zeros in col2 row 0"
+        )
+        assert result.loc[2, "col2"] == "000023607", (
+            f"{engine_name} lost zeros in col2 row 2"
+        )
+        assert result.loc[0, "col4"] == "0150", (
+            f"{engine_name} lost zeros in col4 row 0"
+        )
+        assert result.loc[2, "col4"] == "0205", (
+            f"{engine_name} lost zeros in col4 row 2"
+        )
+
+        for col in ["col1", "col2", "col3", "col4"]:
+            assert result.dtypes[col] == "object", (
+                f"{engine_name} wrong dtype for {col}"
+            )
+
+        assert result.shape == (4, 4)
+        assert list(result.columns) == ["col1", "col2", "col3", "col4"]
+    except AssertionError as exc:
+        if engine_name == "pyarrow":
+            # Known issue: pyarrow engine strips leading zeros even with dtype=str.
+            pytest.xfail(f"failed assertions: {exc}")
+        else:
+            raise