test: add test case for non-ASCII characters (#7)

* test: add test case for non-ASCII characters

* fix: non-ASCII table names fail on Windows

Author: DeflateAwning

.github/workflows/ci.yml

@@ -28,9 +28,15 @@ jobs:
           - os: macos-latest
             python-version: "3.12"
             polars-version: "1.31"
+          - os: macos-latest # Another mac variant.
+            python-version: "3.11"
+            polars-version: "1.34"
           - os: windows-latest
             python-version: "3.12"
             polars-version: "1.31"
+          - os: windows-latest # Another Windows variant.
+            python-version: "3.13"
+            polars-version: "1.34"
 
           # Python variations
           - os: ubuntu-latest

README.md

@@ -5,7 +5,7 @@ A library for reading tables from an Access database into Polars dataframes, usi
 
 A tiny, `subprocess`-based tool for reading a 
 [MS Access](https://products.office.com/en-us/access) 
-database (`.rdb` or `.accdb`) as a [Python Polars Dataframe](https://docs.pola.rs).
+database (`.mdb`, `.accdb`, `.rdb`) as a [Python Polars Dataframe](https://docs.pola.rs).
 
 ## Installation
 

src/polars_access_mdbtools/__init__.py

@@ -41,7 +41,7 @@ def list_table_names(db_path: str | Path) -> list[str]:
         subprocess.check_output(  # noqa: S603
             ["mdb-tables", "--single-column", _path_to_cmd_str(db_path)],  # noqa: S607
         )
-        .decode()
+        .decode(locale.getpreferredencoding())
         .replace("\r\n", "\n")
         .strip()
     )
@@ -106,12 +106,10 @@ def _extract_data_type_definitions(defs_str: str) -> dict[str, str]:
 def _read_table_mdb_schema(
     db_path: str | Path,
     table_name: str,
-    encoding: str = "utf-8",
 ) -> dict[str, str]:
     """Read the schema of a given database into a dictionary of the mdb-schema output.
 
     :param db_path: The MS Access database file.
-    :param encoding: The schema encoding.
     :return: a dictionary of `{column_name: access_data_type}`
     """
     cmd = [
@@ -135,7 +133,7 @@ def _read_table_mdb_schema(
             raise ValueError(msg) from e
         raise
 
-    cmd_output = cmd_output.decode(encoding)
+    cmd_output = cmd_output.decode(locale.getpreferredencoding())
     lines = cmd_output.splitlines()
     schema_ddl = "\n".join(line for line in lines if line and not line.startswith("-"))
 
@@ -203,8 +201,7 @@ def read_table(
         Otherwise, raise an error on unhandled SQL data types.
     :return: a `pl.DataFrame`
     """
-    schema_encoding = "utf-8"
-    mdb_schema = _read_table_mdb_schema(db_path, table_name, schema_encoding)
+    mdb_schema = _read_table_mdb_schema(db_path, table_name)
     pl_schema_target = _convert_mdb_schema_to_polars_schema(
         mdb_schema,
         implicit_string=implicit_string,

tests/conftest.py

@@ -43,3 +43,26 @@ def sample_db_1() -> Iterator[Path]:
 
         yield db_path  # Provide file path to test.
         # Tempdir automatically cleaned up on exit.
+
+
+@pytest.fixture
+def sample_db_2() -> Iterator[Path]:
+    """Give another sample access database file for tests.
+
+    Download the Access Sample .mdb file to a temporary directory,
+    yield its path for tests, and delete it afterward.
+    """
+    url = "https://github.com/el3um4s/mdbtools/raw/refs/heads/main/src/__tests__/test%202.mdb"
+    with tempfile.TemporaryDirectory() as temp_dir_str:
+        db_path = Path(temp_dir_str) / "file_example_MDB_250kB.mdb"
+        response = requests.get(url, stream=True, timeout=10)
+        response.raise_for_status()
+
+        db_path.write_bytes(response.content)
+        assert (
+            _sha256_checksum(db_path)
+            == "560bfd44ad5a6efbab4c86622c92a7071eda9d73c3b453e4bba227d82d725fec"
+        ), "Downloaded file checksum does not match expected value."
+
+        yield db_path  # Provide file path to test.
+        # Tempdir automatically cleaned up on exit.

tests/test_list_table_names.py

@@ -21,3 +21,16 @@ def test_list_table_names_db_1(sample_db_1: Path) -> None:
         "tblFileList",
         "USysRibbons",
     ]
+
+
+def test_list_table_names_db_2(sample_db_2: Path) -> None:
+    table_names = list_table_names(sample_db_2)
+    assert isinstance(table_names, list)
+    assert table_names == [
+        "Colors",
+        "Colors Table Two",
+        "Colors-Table Others",
+        "Dictionary",
+        "Users",
+        "Colors 1°à",
+    ]

tests/test_read_table.py

@@ -62,4 +62,30 @@ def test_reading_specific_table_1a(sample_db_1: Path) -> None:
         },
     )
     assert_frame_equal(df, df_expected)
-    assert df.schema == df_expected.schema
+
+
+def test_reading_specific_table_2a(sample_db_2: Path) -> None:
+    """Test reading a specific table and checking its schema.
+
+    Table contains string columns with non-ASCII characters.
+    """
+    df = read_table(sample_db_2, table_name="Colors-Table Others")
+
+    df_expected = pl.DataFrame(
+        [
+            {"Colors": "Red", "Value": 10, "Second Value": 5, "Others-A": "à"},
+            {"Colors": "Green", "Value": 5, "Second Value": 3, "Others-A": "1a"},
+            {"Colors": "Blue", "Value": 16, "Second Value": 4, "Others-A": "ò"},
+            {"Colors": "Black", "Value": 1, "Second Value": 3, "Others-A": "2°"},
+            {"Colors": "Yellow", "Value": 12, "Second Value": 3, "Others-A": "Y"},
+            {"Colors": "White", "Value": 10, "Second Value": 1, "Others-A": "W"},
+            {"Colors": "Others", "Value": 0, "Second Value": 0, "Others-A": "A"},
+        ],
+        schema={
+            "Colors": pl.String,
+            "Value": pl.Int64,
+            "Second Value": pl.Int64,
+            "Others-A": pl.String,
+        },
+    )
+    assert_frame_equal(df, df_expected)