Merge branch 'main' into omri374/dicom/pipenv_lock

presidio-image-redactor/presidio_image_redactor/dicom_image_pii_verify_engine.py

@@ -66,6 +66,11 @@ def verify_dicom_instance(
         """
         instance_copy = deepcopy(instance)
 
+        try:
+            instance_copy.PixelData
+        except AttributeError:
+            raise AttributeError("Provided DICOM instance lacks pixel data.")
+
         # Load image for processing
         with tempfile.TemporaryDirectory() as tmpdirname:
             # Convert DICOM to PNG and add padding for OCR (during analysis)
@@ -86,8 +91,8 @@ def verify_dicom_instance(
         ocr_results = self.ocr_engine.perform_ocr(image)
         analyzer_results = self.image_analyzer_engine.analyze(
             image,
-            ad_hoc_recognizers=[deny_list_recognizer],
             ocr_kwargs=ocr_kwargs,
+            ad_hoc_recognizers=[deny_list_recognizer],
             **text_analyzer_kwargs,
         )
 

presidio-image-redactor/presidio_image_redactor/dicom_image_redactor_engine.py

@@ -1,4 +1,5 @@
 import os
+import uuid
 import shutil
 from copy import deepcopy
 import tempfile
@@ -49,18 +50,24 @@ def redact(
         :return: DICOM instance with redacted pixel data.
         """
         # Check input
-        if type(image) != pydicom.dataset.FileDataset:
+        if type(image) not in [pydicom.dataset.FileDataset, pydicom.dataset.Dataset]:
             raise TypeError("The provided image must be a loaded DICOM instance.")
+        try:
+            image.PixelData
+        except AttributeError:
+            raise AttributeError("Provided DICOM instance lacks pixel data.")
+
         instance = deepcopy(image)
 
         # Load image for processing
         with tempfile.TemporaryDirectory() as tmpdirname:
             # Convert DICOM to PNG and add padding for OCR (during analysis)
             is_greyscale = self._check_if_greyscale(instance)
             image = self._rescale_dcm_pixel_array(instance, is_greyscale)
-            self._save_pixel_array_as_png(image, is_greyscale, "tmp_dcm", tmpdirname)
+            image_name = str(uuid.uuid4())
+            self._save_pixel_array_as_png(image, is_greyscale, image_name, tmpdirname)
 
-            png_filepath = f"{tmpdirname}/tmp_dcm.png"
+            png_filepath = f"{tmpdirname}/{image_name}.png"
             loaded_image = Image.open(png_filepath)
             image = self._add_padding(loaded_image, is_greyscale, padding_width)
 
@@ -72,8 +79,8 @@ def redact(
         )
         analyzer_results = self.image_analyzer_engine.analyze(
             image,
-            ad_hoc_recognizers=[deny_list_recognizer],
             ocr_kwargs=ocr_kwargs,
+            ad_hoc_recognizers=[deny_list_recognizer],
             **text_analyzer_kwargs,
         )
 
@@ -225,8 +232,11 @@ def _check_if_greyscale(instance: pydicom.dataset.FileDataset) -> bool:
         :return: FALSE if the Photometric Interpretation is RGB.
         """
         # Check if image is grayscale using the Photometric Interpretation element
-        color_scale = instance[0x0028, 0x0004].value
-        is_greyscale = color_scale != "RGB"
+        try:
+            color_scale = instance.PhotometricInterpretation
+        except AttributeError:
+            color_scale = None
+        is_greyscale = (color_scale in ["MONOCHROME1", "MONOCHROME2"])
 
         return is_greyscale
 
@@ -243,7 +253,10 @@ def _rescale_dcm_pixel_array(
         """
         # Normalize contrast
         if "WindowWidth" in instance:
-            image_2d = apply_voi_lut(instance.pixel_array, instance)
+            if is_greyscale:
+                image_2d = apply_voi_lut(instance.pixel_array, instance)
+            else:
+                image_2d = instance.pixel_array
         else:
             image_2d = instance.pixel_array
 
@@ -814,6 +827,11 @@ def _redact_single_dicom_image(
         # Load instance
         instance = pydicom.dcmread(dst_path)
 
+        try:
+            instance.PixelData
+        except AttributeError:
+            raise AttributeError("Provided DICOM file lacks pixel data.")
+
         # Load image for processing
         with tempfile.TemporaryDirectory() as tmpdirname:
             # Convert DICOM to PNG and add padding for OCR (during analysis)
@@ -830,8 +848,8 @@ def _redact_single_dicom_image(
         )
         analyzer_results = self.image_analyzer_engine.analyze(
             image,
-            ad_hoc_recognizers=[deny_list_recognizer],
             ocr_kwargs=ocr_kwargs,
+            ad_hoc_recognizers=[deny_list_recognizer],
             **text_analyzer_kwargs,
         )
 

presidio-image-redactor/tests/test_dicom_image_pii_verify_engine.py

@@ -1,5 +1,6 @@
 """Unit tests for dicom_image_pii_verify_engine
 """
+from copy import deepcopy
 import pydicom
 
 from presidio_image_redactor import (
@@ -121,6 +122,28 @@ def test_verify_dicom_instance_happy_path(
     assert mock_analyze.call_count == 1
     assert mock_verify.call_count == 1
 
+def test_verify_dicom_instance_exception(
+    mock_engine: DicomImagePiiVerifyEngine,
+    get_mock_dicom_instance: pydicom.dataset.FileDataset,
+):
+    """Test exceptions for DicomImagePiiVerifyEngine.verify_dicom_instance
+    Args:
+        mock_engine (DicomImagePiiVerifyEngine): Instantiated engine.
+        get_mock_dicom_instance (pydicom.dataset.FileDataset): Loaded DICOM.
+    """
+    with pytest.raises(Exception) as exc_info:
+        # Arrange
+        padding_width = 25
+        test_instance = deepcopy(get_mock_dicom_instance)
+        del test_instance.PixelData
+        expected_error_type = AttributeError
+
+        # Act
+        _, _, _ = mock_engine.verify_dicom_instance(test_instance, padding_width)
+
+        # Assert
+        assert expected_error_type == exc_info.typename
+
 
 # ------------------------------------------------------
 # DicomImagePiiVerifyEngine.eval_dicom_instance()

presidio-image-redactor/tests/test_dicom_image_redactor_engine.py

@@ -41,7 +41,7 @@ def mock_engine():
             Path(TEST_DICOM_PARENT_DIR),
             [
                 Path(TEST_DICOM_PARENT_DIR, "0_ORIGINAL.dcm"),
-                Path(TEST_DICOM_PARENT_DIR, "0_ORIGINAL_compressed.dcm"),
+                Path(TEST_DICOM_PARENT_DIR, "0_ORIGINAL_no_pixels.dcm"),
                 Path(TEST_DICOM_PARENT_DIR, "RGB_ORIGINAL.dcm"),
                 Path(TEST_DICOM_DIR_2, "1_ORIGINAL.DCM"),
                 Path(TEST_DICOM_DIR_2, "2_ORIGINAL.dicom"),
@@ -1208,7 +1208,8 @@ def test_DicomImageRedactorEngine_redact_happy_path(
         (Path(TEST_DICOM_PARENT_DIR), "TypeError"),
         ("path_here", "TypeError"),
         (np.random.randint(255, size=(64, 64)), "TypeError"),
-        (Image.fromarray(np.random.randint(255, size=(400, 400),dtype=np.uint8)), "TypeError")
+        (Image.fromarray(np.random.randint(255, size=(400, 400),dtype=np.uint8)), "TypeError"),
+        (pydicom.dcmread(Path(TEST_DICOM_PARENT_DIR, "0_ORIGINAL_no_pixels.dcm")), "AttributeError")
     ],
 )
 def test_DicomImageRedactorEngine_redact_exceptions(
@@ -1343,6 +1344,7 @@ def save_as(self, dst_path: str):
     [
         (Path(TEST_DICOM_PARENT_DIR), "FileNotFoundError"),
         (Path("nonexistentfile.extension"), "FileNotFoundError"),
+        (Path(TEST_DICOM_PARENT_DIR, "0_ORIGINAL_no_pixels.dcm"), "AttributeError")
     ],
 )
 def test_DicomImageRedactorEngine_redact_single_dicom_image_exceptions(