def it_knows_its_scaled_slides(self, request, tmpdir):
tmp_path_ = tmpdir.mkdir("myslide")
slide1 = instance_mock(request, Slide)
slide2 = instance_mock(request, Slide)
slideset = SlideSet(tmp_path_, os.path.join(tmp_path_, "processed"),
[])
slides = method_mock(request, SlideSet, "__iter__")
slides.return_value = [slide1, slide2]
slideset.scaled_images(32, 2)
slide1.scaled_image.assert_called_once_with(32)
slide2.scaled_image.assert_called_once_with(32)
def it_knows_its_thumbnails(self, request, tmpdir):
tmp_path_ = tmpdir.mkdir("myslide")
thumbnail_ = property_mock(request, Slide, "thumbnail")
slide1 = Slide("foo/bar", "proc")
slide2 = Slide("foo/bar", "proc")
slideset = SlideSet(tmp_path_, os.path.join(tmp_path_, "processed"),
[])
slides = method_mock(request, SlideSet, "__iter__")
slides.return_value = [slide1, slide2]
slideset.thumbnails()
assert thumbnail_.call_args_list == [call(), call()]
def it_knows_the_slides_dimensions(self, tmpdir):
tmp_path_ = tmpdir.mkdir("myslide")
image = PILIMG.RGBA_COLOR_500X500_155_249_240
image.save(os.path.join(tmp_path_, "mywsi.svs"), "TIFF")
image2 = PILIMG.RGBA_COLOR_50X50_155_0_0
image2.save(os.path.join(tmp_path_, "mywsi2.svs"), "TIFF")
slideset = SlideSet(tmp_path_, "proc", [".svs"])
slides_dimensions = slideset._slides_dimensions
expected_value = [
{
"slide": "mywsi",
"width": 500,
"height": 500,
"size": 250000
},
{
"slide": "mywsi2",
"width": 50,
"height": 50,
"size": 2500
},
]
assert dict_list_eq(slides_dimensions, expected_value) is True
def it_knows_its_total_slides(self, request, Slide_):
slides = property_mock(request, SlideSet, "slides")
slides.return_value = [Slide_ for _ in range(4)]
slideset = SlideSet("the/path", "proc", [".svs"])
total_slides = slideset.total_slides
assert total_slides == 4
def it_constructs_its_sequence_of_slides_to_help(self, tmpdir):
tmp_path_ = tmpdir.mkdir("myslide")
image = PILIMG.RGBA_COLOR_500X500_155_249_240
image.save(os.path.join(tmp_path_, "mywsi.svs"), "TIFF")
image2 = PILIMG.RGBA_COLOR_50X50_155_0_0
image2.save(os.path.join(tmp_path_, "mywsi2.svs"), "TIFF")
slideset = SlideSet(tmp_path_, "proc", [".svs"])
expected_slides = [
Slide(os.path.join(tmp_path_, _path), "proc")
for _path in os.listdir(tmp_path_)
]
slides = slideset.__iter__()
for i, slide in enumerate(slides):
np.testing.assert_array_almost_equal(
slide.resampled_array(), expected_slides[i].resampled_array())
def it_knows_its_total_slides(self, request, Slide_):
slides = method_mock(request, SlideSet, "__iter__")
slides.return_value = [Slide_ for _ in range(4)]
slideset = SlideSet("the/path", "proc", [".svs"])
total_slides = len(slideset)
assert total_slides == 4
def it_knows_its_slides(self, tmpdir):
tmp_path_ = tmpdir.mkdir("myslide")
image = PILIMG.RGBA_COLOR_500X500_155_249_240
image.save(os.path.join(tmp_path_, "mywsi.svs"), "TIFF")
slideset = SlideSet(tmp_path_, "proc", [".svs"])
assert len(slideset) == 1
slideset = SlideSet(None, "proc", [".svs"])
assert len(slideset) == 0
with pytest.raises(FileNotFoundError) as err:
slideset = SlideSet("fake/path", "proc", [".svs"])
list(slideset)
assert isinstance(err.value, FileNotFoundError)
assert err.value.errno == errno.ENOENT
def and_it_is_exaclty_what_expected(self, tmpdir):
tmp_path_ = tmpdir.mkdir("myslide")
image = PILIMG.RGBA_COLOR_500X500_155_249_240
image.save(os.path.join(tmp_path_, "mywsi.svs"), "TIFF")
slideset = SlideSet(tmp_path_, "proc", [".svs"])
slide = slideset[0]
np.testing.assert_array_almost_equal(slide.resampled_array(),
slideset[0].resampled_array())
def it_can_constructs_slides(self, request, tmpdir):
tmp_path_ = tmpdir.mkdir("myslide")
slides_ = tuple(instance_mock(request, Slide) for _ in range(10))
_slides_ = property_mock(request, SlideSet, "slides")
_slides_.return_value = slides_
slideset = SlideSet(tmp_path_, os.path.join(tmp_path_, "b"), [".svs"])
slides = slideset.slides
_slides_.assert_called_once_with()
assert len(slides) == 10
def it_constructs_from_args(self, request):
_init_ = initializer_mock(request, SlideSet)
_slides_path = "/foo/bar/"
_processed_path = "/foo/bar/wsislides/processed"
_valid_extensions = [".svs", ".tiff"]
slideset = SlideSet(_slides_path, _processed_path, _valid_extensions)
_init_.assert_called_once_with(ANY, _slides_path, _processed_path,
_valid_extensions)
assert isinstance(slideset, SlideSet)
def it_knows_its_slides_dimensions_list(self, tmpdir):
tmp_path_ = tmpdir.mkdir("myslide")
image = PILIMG.RGBA_COLOR_500X500_155_249_240
image.save(os.path.join(tmp_path_, "mywsi.svs"), "TIFF")
image2 = PILIMG.RGBA_COLOR_50X50_155_0_0
image2.save(os.path.join(tmp_path_, "mywsi2.svs"), "TIFF")
slideset = SlideSet(tmp_path_, "proc", [".svs"])
_slides_dimensions_list = slideset._slides_dimensions_list
assert sorted(_slides_dimensions_list) == sorted([(500, 500),
(50, 50)])
def it_knows_its_slides(self, tmpdir):
tmp_path_ = tmpdir.mkdir("myslide")
image = PILImageMock.DIMS_500X500_RGBA_COLOR_155_249_240
image.save(os.path.join(tmp_path_, "mywsi.svs"), "TIFF")
slideset = SlideSet(tmp_path_, "proc", [".svs"])
slides = slideset.slides
assert len(slides) == 1
slideset = SlideSet(None, "proc", [".svs"])
slides = slideset.slides
assert len(slides) == 0
with pytest.raises(FileNotFoundError) as err:
slideset = SlideSet("fake/path", "proc", [".svs"])
slides = slideset.slides
assert isinstance(err.value, FileNotFoundError)
assert str(
err.value) == "[Errno 2] No such file or directory: 'fake/path'"
def extract_random_tiles(
dataset_dir: str,
processed_path: str,
tile_size: Tuple[int, int],
n_tiles: int,
level: int,
seed: int,
check_tissue: bool,
) -> None:
"""Save random tiles extracted from WSIs in `dataset_dir` into
`processed_path`/tiles/
Parameters
----------
dataset_dir : str
Path were the WSIs are saved
processed_path : str
Path where to store the tiles (will be concatenated with /tiles)
tile_size : Tuple[int, int]
width and height of the cropped tiles
n_tiles : int
Maximum number of tiles to extract
level : int
Magnification level from which extract the tiles
seed : int
Seed for RandomState
check_tissue : bool
Whether to check if the tile has enough tissue to be saved
"""
slideset = SlideSet(dataset_dir, processed_path, valid_extensions=[".svs"])
for slide in tqdm(slideset.slides):
prefix = f"{slide.name}_"
random_tiles_extractor = RandomTiler(
tile_size=tile_size,
n_tiles=n_tiles,
level=level,
seed=seed,
check_tissue=check_tissue,
prefix=prefix,
)
random_tiles_extractor.extract(slide)
def it_knows_its_min_size_slide(self, _slides_dimensions_prop):
_slides_dimensions_prop.return_value = [
{
"slide": "mywsi",
"width": 500,
"height": 100,
"size": 250000
},
{
"slide": "mywsi2",
"width": 600,
"height": 50,
"size": 2500
},
]
slideset = SlideSet("fake/path", "proc", [".svs"])
_min_size_slide = slideset._min_size_slide
assert _min_size_slide == {"slide": "mywsi2", "size": 2500}
def it_generates_slides_stats(self, total_slides_prop, tmpdir):
total_slides_prop.return_value = 2
tmp_path_ = tmpdir.mkdir("myslide")
image = PILIMG.RGBA_COLOR_500X500_155_249_240
image.save(os.path.join(tmp_path_, "mywsi.svs"), "TIFF")
image2 = PILIMG.RGBA_COLOR_50X50_155_0_0
image2.save(os.path.join(tmp_path_, "mywsi2.svs"), "TIFF")
slideset = SlideSet(tmp_path_, os.path.join("proc"), [".svs"])
slides_stats = slideset.slides_stats
assert slides_stats == {
"no_of_slides": 2,
"max_width": {
"slide": "mywsi",
"width": 500
},
"max_height": {
"slide": "mywsi",
"height": 500
},
"max_size": {
"slide": "mywsi",
"size": 250000
},
"min_width": {
"slide": "mywsi2",
"width": 50
},
"min_height": {
"slide": "mywsi2",
"height": 50
},
"min_size": {
"slide": "mywsi2",
"size": 2500
},
"avg_width": 275.0,
"avg_height": 275.0,
"avg_size": 126250.0,
}
def it_constructs_its_sequence_of_slides_to_help(self, request, Slide_,
tmpdir):
slides_path = tmpdir.mkdir("mypath")
for i in range(4):
open(os.path.join(slides_path, f"myfile{i}.svs"), "a")
slides_ = tuple(instance_mock(request, Slide) for _ in range(4))
Slide_.side_effect = iter(slides_)
slide_set = SlideSet(
slides_path=slides_path,
processed_path=os.path.join(slides_path, "processed"),
valid_extensions=[".svs"],
)
slides = tuple(slide_set.slides)
assert sorted(Slide_.call_args_list) == sorted([
call(
os.path.join(slides_path, f"myfile{i}.svs"),
os.path.join(slides_path, "processed"),
) for i in range(4)
])
assert slides == slides_
def it_knows_its_avg_size_slide(self, _slides_dimensions_prop,
total_slides_prop):
total_slides_prop.return_value = 2
_slides_dimensions_prop.return_value = [
{
"slide": "mywsi",
"width": 500,
"height": 100,
"size": 250000
},
{
"slide": "mywsi2",
"width": 50,
"height": 50,
"size": 2500
},
]
slideset = SlideSet("fake/path", "proc", [".svs"])
_avg_size_slide = slideset._avg_size_slide
assert _avg_size_slide == 126250.0
assert (_avg_size_slide == sum(
d["size"] for d in _slides_dimensions_prop.return_value) / 2)
