unit tests for png-extraction

.gitignore

@@ -6,3 +6,9 @@ target/
 **/*.csv
 __pycache__
 **/*.pyc 
+htmlcov
+.coverage
+coverage.xml
+/tests/data/tmp
+/tests/data/**/*.dcm
+!/tests/data/**/no-img.dcm

requirements-dev.txt

@@ -0,0 +1,13 @@
+requests
+pymongo
+schedule
+pydicom
+pynetdicom
+image
+numpy
+pandas
+pillow
+pypng
+pytest
+pytest-mock
+pytest-cov

tests/README.md

@@ -0,0 +1,34 @@
+# Testing Framework for Niffler Modules
+
+## Setup
+
+Install the requirements from `<repo-home>/requirements-dev.txt`
+
+```
+pip install -r requirements-dev.txt
+```
+
+Add the test data in `<repo-home>/tests/data/<module-name>/input` for respective tests.
+
+### PNG Extraction Data Setup
+
+Test data in `<repo-home>/tests/data/png-extraction/input`.
+
+For unit tests, add a valid dcm file, with name `test-img.dcm`.
+
+## Running Tests
+
+Initialize the required data, and run tests from `<repo-home>`.
+
+```bash
+pytest ./tests
+```
+
+For coverage report, run
+
+```bash
+pytest ./tests --cov=./modules --cov-report=html
+```
+
+and open the `<repo-home>/htmlcov/index.html`
+

tests/conftest.py

@@ -0,0 +1,15 @@
+import os
+import pytest
+from pathlib import Path
+
+
+def create_dirs(*args):
+    for dir in args:
+        if not os.path.exists(dir):
+            os.makedirs(dir)
+
+
+def pytest_configure():
+    pytest.data_dir = Path.cwd() / 'tests' / 'data'
+    pytest.out_dir = Path.cwd() / 'tests' / 'data' / 'tmp' / 'niffler-tests'
+    pytest.create_dirs = create_dirs

tests/unit/test_png_extraction.py

@@ -0,0 +1,217 @@
+import glob
+import pytest
+import sys
+import time
+
+from pathlib import Path, PurePath
+from pytest_mock import MockerFixture
+
+# Import Niffler Module
+niffler_modules_path = Path.cwd() / 'modules'
+sys.path.append(str(niffler_modules_path / 'png-extraction'))
+import ImageExtractor
+
+import pydicom
+import pandas as pd
+
+
+@pytest.fixture
+def mock_pydicom_config_data_element_callback(mocker: MockerFixture):
+    return mocker.patch.object(pydicom.config, 'data_element_callback')
+
+
+@pytest.fixture
+def mock_pydicom_config_data_element_callback_kwargs(mocker: MockerFixture):
+    return mocker.patch.object(pydicom.config, 'data_element_callback_kwargs')
+
+
+@pytest.fixture
+def mock_logger(mocker: MockerFixture):
+    return mocker.patch('ImageExtractor.logging')
+
+
+class TestGetPath:
+    dicom_home = "/mock/path/to/dicom/home"
+
+    def test_get_path_zero_depth(self):
+        depth = 0
+        dcm_path = ImageExtractor.get_path(depth, self.dicom_home)
+        assert dcm_path == f"{self.dicom_home}/*.dcm"
+
+    def test_get_path_some_depth(self):
+        depth = 3
+        dcm_path = ImageExtractor.get_path(depth, self.dicom_home)
+        assert dcm_path == f"{self.dicom_home}{''.join(['/*']*depth)}/*.dcm"
+
+
+class TestFixMismatch:
+    with_VRs = ['PN', 'DS', 'IS']
+
+    def test_fix_mismatch(self, mock_pydicom_config_data_element_callback, mock_pydicom_config_data_element_callback_kwargs):
+        ImageExtractor.fix_mismatch(with_VRs=self.with_VRs)
+        assert pydicom.config.data_element_callback is ImageExtractor.fix_mismatch_callback
+        assert pydicom.config.data_element_callback_kwargs['with_VRs'] == self.with_VRs
+
+
+class TestExtractHeaders:
+    valid_test_dcm_file = 0, str(
+        pytest.data_dir / 'png-extraction' / 'input' / 'test-img.dcm')
+    invalid_test_dcm_file = 0, str(
+        pytest.data_dir / 'png-extraction' / 'input' / 'no-img.dcm')
+
+    def test_no_image(self):
+        headers = ImageExtractor.extract_headers(
+            self.invalid_test_dcm_file)
+        assert headers['has_pix_array'] is False
+
+    def test_valid_image(self):
+        headers = ImageExtractor.extract_headers(self.valid_test_dcm_file)
+        assert headers['has_pix_array'] is True
+
+    # TODO large dcm files
+
+
+class TestGetTuples:
+    test_dcm_file = str(
+        pytest.data_dir / 'png-extraction' / 'input' / 'test-img.dcm')
+    test_valid_plan = pydicom.dcmread(test_dcm_file, force=True)
+
+    def test_correct_output(self):
+        first_key = self.test_valid_plan.dir()[0]
+        tuple_list = ImageExtractor.get_tuples(self.test_valid_plan)
+        assert tuple_list[0][0] == first_key
+
+    # TODO hasattr error
+    # TODO large dcm files
+
+
+class TestExtractImages:
+
+    # TODO Write suitable assertions
+
+    test_dcm_file = str(
+        pytest.data_dir / 'png-extraction' / 'input' / 'test-img.dcm')
+    invalid_test_dcm_file = str(
+        pytest.data_dir / 'png-extraction' / 'input' / 'no-img.dcm')
+
+    def setup_method(self):
+        header_list = [ImageExtractor.extract_headers(
+            (0, self.test_dcm_file))]
+        self.file_data = pd.DataFrame(header_list)
+        self.index = 0
+        self.invalid_file_data = pd.DataFrame([
+            {
+                'some_col_1': 'Dummy Col1 Value',
+                'some_col_2': 'Dummy Col2 Value',
+                'file': self.invalid_test_dcm_file
+            }
+        ])
+        out_dir = pytest.out_dir / 'png-extraction/outputs/TestExtractImages'
+        self.png_destination = f"{str(out_dir)}/extracted-images/"
+        self.failed = f"{str(out_dir)}/failed-dicom/"
+        pytest.create_dirs(out_dir, self.png_destination, self.failed)
+
+    def test_is16bit(self):
+        flattened_to_level = "patient"
+        is16Bit = "True"
+        out_img = ImageExtractor.extract_images(
+            self.file_data,
+            self.index,
+            self.png_destination,
+            flattened_to_level,
+            self.failed,
+            is16Bit
+        )
+        assert out_img[0].startswith(self.test_dcm_file)
+
+    def test_not_is16bit(self):
+        flattened_to_level = "patient"
+        is16Bit = "False"
+        out_img = ImageExtractor.extract_images(
+            self.file_data,
+            self.index,
+            self.png_destination,
+            flattened_to_level,
+            self.failed,
+            is16Bit
+        )
+        assert out_img[0].startswith(self.test_dcm_file)
+
+    def test_level_patient(self):
+        flattened_to_level = "patient"
+        is16Bit = "False"
+        out_img = ImageExtractor.extract_images(
+            self.file_data,
+            self.index,
+            self.png_destination,
+            flattened_to_level,
+            self.failed,
+            is16Bit
+        )
+        assert out_img[0].startswith(self.test_dcm_file)
+
+    def test_level_study(self):
+        flattened_to_level = "study"
+        is16Bit = "False"
+        out_img = ImageExtractor.extract_images(
+            self.file_data,
+            self.index,
+            self.png_destination,
+            flattened_to_level,
+            self.failed,
+            is16Bit
+        )
+        assert out_img[0].startswith(self.test_dcm_file)
+
+    def test_level_other(self):
+        flattened_to_level = "other"
+        is16Bit = "False"
+        out_img = ImageExtractor.extract_images(
+            self.file_data,
+            self.index,
+            self.png_destination,
+            flattened_to_level,
+            self.failed,
+            is16Bit
+        )
+        assert out_img[0].startswith(self.test_dcm_file)
+
+    def test_level_other_no_study_uuid(self):
+        flattened_to_level = "other"
+        is16Bit = "False"
+        out_img = ImageExtractor.extract_images(
+            self.file_data.drop(['StudyInstanceUID'], axis=1),
+            self.index,
+            self.png_destination,
+            flattened_to_level,
+            self.failed,
+            is16Bit
+        )
+        assert out_img[0].startswith(self.test_dcm_file)
+
+    def test_level_study_no_study_uuid(self):
+        flattened_to_level = "study"
+        is16Bit = "False"
+        out_img = ImageExtractor.extract_images(
+            self.file_data.drop(['StudyInstanceUID'], axis=1),
+            self.index,
+            self.png_destination,
+            flattened_to_level,
+            self.failed,
+            is16Bit
+        )
+        assert out_img[0].startswith(self.test_dcm_file)
+
+    def test_failed_read_attrerr(self):
+        flattened_to_level = "study"
+        is16Bit = "False"
+        out_img = ImageExtractor.extract_images(
+            self.invalid_file_data,
+            self.index,
+            self.png_destination,
+            flattened_to_level,
+            self.failed,
+            is16Bit
+        )
+        assert out_img[1][0] == self.invalid_test_dcm_file
+        assert out_img[2] is not None