Add test to exercise pipeline on h5path

pathml/core/slide_data.py

@@ -215,7 +215,7 @@ def __init__(
 
  self.masks = pathml.core.Masks(h5manager=self.h5manager, masks=masks)
  self.tiles = pathml.core.Tiles(h5manager=self.h5manager, tiles=tiles)
- self._add_tiles = tiles is None
+ self._add_tiles = tiles is None and not _load_from_h5path
 
  self.tile_size = tile_size
  self._tile_stride = tile_stride
@@ -298,7 +298,9 @@ def run(
  assert isinstance(
  pipeline, pathml.preprocessing.pipeline.Pipeline
  ), f"pipeline is of type {type(pipeline)} but must be of type pathml.preprocessing.pipeline.Pipeline"
- assert self.slide is not None, "cannot run pipeline because self.slide is None"
+ assert (
+ self.slide is not None or not self._add_tiles
+ ), "cannot run pipeline because self.slide is None and no tiles already exist"
 
  shutdown_after = False
 

pathml/preprocessing/transforms.py

@@ -999,8 +999,10 @@ def apply(self, tile):
 
 class DropTileException(Exception):
  """Stops the SlideData from adding a tile to the h5path."""
+
  pass
 
+
 class TissueDetectionHE(Transform):
  """
  Detect tissue regions from H&E stained slide.

tests/integration_tests/test_pipeline_running.py

@@ -188,3 +188,25 @@ def test_pipeline_overlapping_tiles(tmp_path, stride, pad, tile_size):
  )
  expected = AddMean().F(im).astype(np.float16)
  np.testing.assert_equal(readslidedata.tiles[(1000, 1000)].image, expected)
+
+
+@pytest.mark.parametrize("dist", [False, True])
+def test_pipeline_on_h5path(tmp_path, dist, cluster):
+ save_path = str(tmp_path) + str(np.round(np.random.rand(), 8)) + "HE_slide.h5"
+ # Make h5path
+ slide = HESlide("tests/testdata/small_HE.svs")
+ pipeline = Pipeline([BoxBlur(kernel_size=15)])
+ cli = Client(cluster) if dist else None
+ slide.run(pipeline, distributed=dist, client=cli)
+ slide.write(path=save_path)
+ # Load saved h5path and run pipeline
+ h5path_slide = HESlide(save_path, dtype=np.uint8)
+ h5path_slide.run(pipeline, distributed=dist, client=cli)
+ h5path_slide.write(path=save_path)
+
+ if dist:
+ cli.shutdown()
+
+ # test out the dataset
+ dataset = TileDataset(save_path)
+ assert len(dataset) == len(slide.tiles)