Merge pull request #284 from bgilbert/filenames

Fix several problems with filename handling

Author: bgilbert

openslide/__init__.py

@@ -26,7 +26,6 @@
 from __future__ import annotations
 
 from io import BytesIO
-from pathlib import Path
 from types import TracebackType
 from typing import Iterator, Literal, Mapping, TypeVar
 
@@ -82,7 +81,7 @@ def __exit__(
         return False
 
     @classmethod
-    def detect_format(cls, filename: str | Path) -> str | None:
+    def detect_format(cls, filename: lowlevel.Filename) -> str | None:
         """Return a string describing the format of the specified file.
 
         If the file format is not recognized, return None."""
@@ -189,23 +188,23 @@ class OpenSlide(AbstractSlide):
     operations on the OpenSlide object, other than close(), will fail.
     """
 
-    def __init__(self, filename: str | Path):
+    def __init__(self, filename: lowlevel.Filename):
         """Open a whole-slide image."""
         AbstractSlide.__init__(self)
         self._filename = filename
-        self._osr = lowlevel.open(str(filename))
+        self._osr = lowlevel.open(filename)
         if lowlevel.read_icc_profile.available:
             self._profile = lowlevel.read_icc_profile(self._osr)
 
     def __repr__(self) -> str:
         return f'{self.__class__.__name__}({self._filename!r})'
 
     @classmethod
-    def detect_format(cls, filename: str | Path) -> str | None:
+    def detect_format(cls, filename: lowlevel.Filename) -> str | None:
         """Return a string describing the format vendor of the specified file.
 
         If the file format is not recognized, return None."""
-        return lowlevel.detect_vendor(str(filename))
+        return lowlevel.detect_vendor(filename)
 
     def close(self) -> None:
         """Close the OpenSlide object."""
@@ -358,7 +357,7 @@ def __repr__(self) -> str:
 class ImageSlide(AbstractSlide):
     """A wrapper for a PIL.Image that provides the OpenSlide interface."""
 
-    def __init__(self, file: str | Path | Image.Image):
+    def __init__(self, file: lowlevel.Filename | Image.Image):
         """Open an image file.
 
         file can be a filename or a PIL.Image."""
@@ -376,7 +375,7 @@ def __repr__(self) -> str:
         return f'{self.__class__.__name__}({self._file_arg!r})'
 
     @classmethod
-    def detect_format(cls, filename: str | Path) -> str | None:
+    def detect_format(cls, filename: lowlevel.Filename) -> str | None:
         """Return a string describing the format of the specified file.
 
         If the file format is not recognized, return None."""
@@ -484,7 +483,7 @@ def set_cache(self, cache: OpenSlideCache) -> None:
         pass
 
 
-def open_slide(filename: str | Path) -> OpenSlide | ImageSlide:
+def open_slide(filename: lowlevel.Filename) -> OpenSlide | ImageSlide:
     """Open a whole-slide or regular image.
 
     Return an OpenSlide object for whole-slide images and an ImageSlide

openslide/lowlevel.py

@@ -2,7 +2,7 @@
 # openslide-python - Python bindings for the OpenSlide library
 #
 # Copyright (c) 2010-2013 Carnegie Mellon University
-# Copyright (c) 2016-2023 Benjamin Gilbert
+# Copyright (c) 2016-2024 Benjamin Gilbert
 #
 # This library is free software; you can redistribute it and/or modify it
 # under the terms of version 2.1 of the GNU Lesser General Public License
@@ -48,6 +48,7 @@
     cdll,
 )
 from itertools import count
+import os
 import platform
 from typing import TYPE_CHECKING, Any, Callable, Protocol, TypeVar, cast
 
@@ -56,7 +57,7 @@
 from . import _convert
 
 if TYPE_CHECKING:
-    # Python 3.10+ for ParamSpec
+    # Python 3.10+
     from typing import ParamSpec, TypeAlias
 
     from _convert import _Buffer
@@ -196,6 +197,28 @@ def from_param(cls, obj: _OpenSlideCache) -> _OpenSlideCache:
         return obj
 
 
+if TYPE_CHECKING:
+    # Python 3.10+
+    Filename: TypeAlias = str | bytes | os.PathLike[Any]
+
+
+class _filename_p:
+    """Wrapper class to convert filename arguments to bytes."""
+
+    @classmethod
+    def from_param(cls, obj: Filename) -> bytes:
+        # fspath and fsencode raise TypeError on unexpected types
+        if platform.system() == 'Windows':
+            # OpenSlide 4.0.0+ requires UTF-8 on Windows
+            obj = os.fspath(obj)
+            if isinstance(obj, str):
+                return obj.encode('UTF-8')
+            else:
+                return obj
+        else:
+            return os.fsencode(obj)
+
+
 class _utf8_p:
     """Wrapper class to convert string arguments to bytes."""
 
@@ -350,14 +373,14 @@ def decorator(fn: Callable[_P, _T]) -> _Func[_P, _T]:
 
 
 try:
-    detect_vendor: _Func[[str], str] = _func(
-        'openslide_detect_vendor', c_char_p, [_utf8_p], _check_string
+    detect_vendor: _Func[[Filename], str] = _func(
+        'openslide_detect_vendor', c_char_p, [_filename_p], _check_string
     )
 except AttributeError:
     raise OpenSlideVersionError('3.4.0')
 
-open: _Func[[str], _OpenSlide] = _func(
-    'openslide_open', c_void_p, [_utf8_p], _check_open
+open: _Func[[Filename], _OpenSlide] = _func(
+    'openslide_open', c_void_p, [_filename_p], _check_open
 )
 
 close: _Func[[_OpenSlide], None] = _func(
@@ -452,7 +475,7 @@ def read_icc_profile(slide: _OpenSlide) -> bytes | None:
     'openslide_get_property_names', POINTER(c_char_p), [_OpenSlide], _check_name_list
 )
 
-get_property_value: _Func[[_OpenSlide, str], str] = _func(
+get_property_value: _Func[[_OpenSlide, str | bytes], str] = _func(
     'openslide_get_property_value', c_char_p, [_OpenSlide, _utf8_p]
 )
 
@@ -464,7 +487,7 @@ def read_icc_profile(slide: _OpenSlide) -> bytes | None:
 )
 
 _get_associated_image_dimensions: _Func[
-    [_OpenSlide, str, _Pointer[c_int64], _Pointer[c_int64]], None
+    [_OpenSlide, str | bytes, _Pointer[c_int64], _Pointer[c_int64]], None
 ] = _func(
     'openslide_get_associated_image_dimensions',
     None,
@@ -473,46 +496,54 @@ def read_icc_profile(slide: _OpenSlide) -> bytes | None:
 
 
 @_wraps_funcs([_get_associated_image_dimensions])
-def get_associated_image_dimensions(slide: _OpenSlide, name: str) -> tuple[int, int]:
+def get_associated_image_dimensions(
+    slide: _OpenSlide, name: str | bytes
+) -> tuple[int, int]:
     w, h = c_int64(), c_int64()
     _get_associated_image_dimensions(slide, name, byref(w), byref(h))
     return w.value, h.value
 
 
-_read_associated_image: _Func[[_OpenSlide, str, _Pointer[c_uint32]], None] = _func(
-    'openslide_read_associated_image', None, [_OpenSlide, _utf8_p, POINTER(c_uint32)]
+_read_associated_image: _Func[[_OpenSlide, str | bytes, _Pointer[c_uint32]], None] = (
+    _func(
+        'openslide_read_associated_image',
+        None,
+        [_OpenSlide, _utf8_p, POINTER(c_uint32)],
+    )
 )
 
 
 @_wraps_funcs([get_associated_image_dimensions, _read_associated_image])
-def read_associated_image(slide: _OpenSlide, name: str) -> Image.Image:
+def read_associated_image(slide: _OpenSlide, name: str | bytes) -> Image.Image:
     w, h = get_associated_image_dimensions(slide, name)
     buf = (w * h * c_uint32)()
     _read_associated_image(slide, name, buf)
     return _load_image(buf, (w, h))
 
 
-get_associated_image_icc_profile_size: _Func[[_OpenSlide, str], int] = _func(
+get_associated_image_icc_profile_size: _Func[[_OpenSlide, str | bytes], int] = _func(
     'openslide_get_associated_image_icc_profile_size',
     c_int64,
     [_OpenSlide, _utf8_p],
     minimum_version='4.0.0',
 )
 
-_read_associated_image_icc_profile: _Func[[_OpenSlide, str, _Pointer[c_char]], None] = (
-    _func(
-        'openslide_read_associated_image_icc_profile',
-        None,
-        [_OpenSlide, _utf8_p, POINTER(c_char)],
-        minimum_version='4.0.0',
-    )
+_read_associated_image_icc_profile: _Func[
+    [_OpenSlide, str | bytes, _Pointer[c_char]], None
+] = _func(
+    'openslide_read_associated_image_icc_profile',
+    None,
+    [_OpenSlide, _utf8_p, POINTER(c_char)],
+    minimum_version='4.0.0',
 )
 
 
 @_wraps_funcs(
     [get_associated_image_icc_profile_size, _read_associated_image_icc_profile]
 )
-def read_associated_image_icc_profile(slide: _OpenSlide, name: str) -> bytes | None:
+def read_associated_image_icc_profile(
+    slide: _OpenSlide, name: str | bytes
+) -> bytes | None:
     size = get_associated_image_icc_profile_size(slide, name)
     if size == 0:
         return None

tests/fixtures/😐.png

@@ -45,7 +45,7 @@ def test_lowlevel_available(self):
             if getattr(attr, '__module__', None) == '__future__':
                 continue
             # ignore random imports
-            if hasattr(ctypes, name) or name in ('count', 'platform'):
+            if hasattr(ctypes, name) or name in ('count', 'os', 'platform'):
                 continue
             self.assertTrue(
                 hasattr(attr, 'available'),

tests/fixtures/😐.svs

@@ -1,7 +1,7 @@
 #
 # openslide-python - Python bindings for the OpenSlide library
 #
-# Copyright (c) 2016-2023 Benjamin Gilbert
+# Copyright (c) 2016-2024 Benjamin Gilbert
 #
 # This library is free software; you can redistribute it and/or modify it
 # under the terms of version 2.1 of the GNU Lesser General Public License
@@ -19,6 +19,7 @@
 
 from __future__ import annotations
 
+import sys
 import unittest
 
 from PIL import Image
@@ -44,6 +45,21 @@ def test_open_image(self):
                 self.assertEqual(osr.dimensions, (300, 250))
                 self.assertEqual(repr(osr), 'ImageSlide(%r)' % img)
 
+    @unittest.skipUnless(
+        sys.getfilesystemencoding() == 'utf-8',
+        'Python filesystem encoding is not UTF-8',
+    )
+    def test_unicode_path(self):
+        path = file_path('😐.png')
+        for arg in path, str(path):
+            self.assertEqual(ImageSlide.detect_format(arg), 'PNG')
+            self.assertEqual(ImageSlide(arg).dimensions, (300, 250))
+
+    def test_unicode_path_bytes(self):
+        arg = str(file_path('😐.png')).encode('UTF-8')
+        self.assertEqual(ImageSlide.detect_format(arg), 'PNG')
+        self.assertEqual(ImageSlide(arg).dimensions, (300, 250))
+
     def test_operations_on_closed_handle(self):
         with Image.open(file_path('boxes.png')) as img:
             osr = ImageSlide(img)

tests/test_base.py

@@ -1,7 +1,7 @@
 #
 # openslide-python - Python bindings for the OpenSlide library
 #
-# Copyright (c) 2016-2023 Benjamin Gilbert
+# Copyright (c) 2016-2024 Benjamin Gilbert
 #
 # This library is free software; you can redistribute it and/or modify it
 # under the terms of version 2.1 of the GNU Lesser General Public License
@@ -60,12 +60,27 @@ def test_open(self):
         self.assertRaises(
             OpenSlideUnsupportedFormatError, lambda: OpenSlide('setup.py')
         )
-        self.assertRaises(OpenSlideUnsupportedFormatError, lambda: OpenSlide(None))
-        self.assertRaises(OpenSlideUnsupportedFormatError, lambda: OpenSlide(3))
+        self.assertRaises(ArgumentError, lambda: OpenSlide(None))
+        self.assertRaises(ArgumentError, lambda: OpenSlide(3))
         self.assertRaises(
             OpenSlideUnsupportedFormatError, lambda: OpenSlide('unopenable.tiff')
         )
 
+    @unittest.skipUnless(
+        sys.getfilesystemencoding() == 'utf-8',
+        'Python filesystem encoding is not UTF-8',
+    )
+    def test_unicode_path(self):
+        path = file_path('😐.svs')
+        for arg in path, str(path):
+            self.assertEqual(OpenSlide.detect_format(arg), 'aperio')
+            self.assertEqual(OpenSlide(arg).dimensions, (16, 16))
+
+    def test_unicode_path_bytes(self):
+        arg = str(file_path('😐.svs')).encode('UTF-8')
+        self.assertEqual(OpenSlide.detect_format(arg), 'aperio')
+        self.assertEqual(OpenSlide(arg).dimensions, (16, 16))
+
     def test_operations_on_closed_handle(self):
         osr = OpenSlide(file_path('boxes.tiff'))
         props = osr.properties