Merged in refactor/misc_fixes (pull request #282)

numpy overflow/underflow errors

Approved-by: Randy Taylor

pylinac/core/array_utils.py

@@ -169,10 +169,15 @@ def convert_to_dtype(array: np.array, dtype: type[np.dtype]) -> np.array:
     """
     # original array info
     old_dtype_info = get_dtype_info(array.dtype)
-    relative_values = array.astype(float) / old_dtype_info.max
     # float range is so large that it's better to normalize
     if isinstance(old_dtype_info, np.finfo):
-        relative_values = stretch(relative_values, min=0, max=1)
+        relative_values = stretch(array, min=0, max=1)
+    else:
+        # we have an int-like array.
+        # the float conversion is to avoid integer division
+        # this can help when the values are very small
+        # we will cast back to int later
+        relative_values = array.astype(float) / old_dtype_info.max
     # new array info
     dtype_info = get_dtype_info(dtype)
     dtype_range = dtype_info.max - dtype_info.min

pylinac/core/image.py

@@ -19,7 +19,7 @@
 import matplotlib.pyplot as plt
 import numpy as np
 import pydicom
-import scipy.ndimage.filters as spf
+import scipy.ndimage as spf
 from PIL import Image as pImage
 from PIL.PngImagePlugin import PngInfo
 from PIL.TiffTags import TAGS
@@ -347,8 +347,8 @@ def _is_dicom(path: str | Path | io.BytesIO | ImageLike | np.ndarray) -> bool:
 def _is_image_file(path: str | Path) -> bool:
     """Whether the file is a readable image file via Pillow."""
     try:
-        pImage.open(path)
-        return True
+        with pImage.open(path):
+            return True
     except:
         return False
 
@@ -1051,14 +1051,12 @@ def _parse_compressed_bytes(
             xim, int, num_values=img_width + 1
         )
         diffs = self._get_diffs(lookup_table, xim)
-        for diff, idx in zip(
-            np.asarray(diffs, dtype=np.int16),
-            range(img_width + 1, img_width * img_height),
-        ):
-            left = a[idx - 1]
-            above = a[idx - img_width]
-            upper_left = a[idx - img_width - 1]
-            a[idx] = diff + left + above - upper_left
+        for diff, idx in zip(diffs, range(img_width + 1, img_width * img_height)):
+            # intermediate math can cause overflow errors. Use float for intermediate, then back to int
+            left = float(a[idx - 1])
+            above = float(a[idx - img_width])
+            upper_left = float(a[idx - img_width - 1])
+            a[idx] = np.asarray(diff + left + above - upper_left, dtype=dtype)
         return a.reshape((img_height, img_width))
 
     @staticmethod
@@ -1087,7 +1085,16 @@ def _get_diffs(lookup_table: np.ndarray, xim: BinaryIO):
                 diffs[start + 1 : stop] = vals
             if stop != byte_changes[-1]:
                 diffs[stop] = decode_binary(xim, LOOKUP_CONVERSION[lookup_table[stop]])
-        return diffs
+        return np.asarray(diffs, dtype=float)
+
+    @property
+    def dpmm(self) -> float:
+        """The dots/mm value of the XIM images. The value appears to be in cm in the file."""
+        if self.properties["PixelWidth"] != self.properties["PixelHeight"]:
+            raise ValueError(
+                "The XIM image does not have the same pixel height and width"
+            )
+        return 1 / (10 * self.properties["PixelHeight"])
 
     def save_as(self, file: str, format: str | None = None) -> None:
         """Save the image to a NORMAL format. PNG is highly suggested. Accepts any format supported by Pillow.

pylinac/core/utilities.py

@@ -173,12 +173,12 @@ def decode_binary(
         ssize = struct.calcsize("i") * num_values
         output = np.asarray(struct.unpack("i" * num_values, f.read(ssize)))
         if len(output) == 1:
-            output = int(output)
+            output = int(np.squeeze(output))
     elif dtype == float:
         ssize = struct.calcsize("f") * num_values
         output = np.asarray(struct.unpack("f" * num_values, f.read(ssize)))
         if len(output) == 1:
-            output = float(output)
+            output = float(np.squeeze(output))
     else:
         raise TypeError(f"datatype '{dtype}' was not valid")
 

tests_basic/core/test_image.py

@@ -551,6 +551,16 @@ def test_save_tiff(self):
         png_array = np.asarray(pimg)
         assert_array_almost_equal(png_array, xim.array)
 
+    def test_overflow(self):
+        old_settings = np.geterr()
+        # set overflow to cause errors
+        # shouldn't raise
+        np.seterr(all="raise")
+        xim_path = get_file_from_cloud_test_repo(["IsoCal-kV-08.xim"])
+        XIM(xim_path)
+        # reset to old settings
+        np.seterr(**old_settings)
+
 
 class TestLinacDicomImage(TestCase):
     def test_normal_image(self):

tests_basic/utils.py

@@ -103,9 +103,9 @@ def get_file_from_cloud_test_repo(path: List[str], force: bool = False) -> str:
     """Get a single file from GCP storage. Returns the path to disk it was downloaded to"""
     local_filename = osp.join(osp.dirname(__file__), LOCAL_TEST_DIR, *path)
     if osp.isfile(local_filename) and not force:
-        print(
-            f"Local file found: {local_filename}@{hashlib.md5(open(local_filename, 'rb').read()).hexdigest()}"
-        )
+        with open(local_filename, "rb") as f:
+            file_hash = hashlib.md5(f.read()).hexdigest()
+            print(f"Local file found: {local_filename}@{file_hash}")
         return local_filename
     with access_gcp() as client:
         bucket = client.bucket(GCP_BUCKET_NAME)