Add test for image extraction, closes #40

cleanX/cli/__init__.py

@@ -6,8 +6,8 @@
 """
 
 from .main import main
-from .dicom import dicom, extract_images
+from .dicom import dicom, extract, report
 from .images import images, run_pipeline, restore_pipeline
-from .dataset import dataset, generate_report
+from .dataset import dataset, report
 
 

cleanX/cli/dataset.py

@@ -112,7 +112,7 @@ def dataset(cfg):
     * txt
     ''',
 )
-def generate_report(
+def report(
         cfg,
         train_source,
         test_source,

cleanX/cli/dicom.py

@@ -68,7 +68,7 @@ def dicom(cfg):
     These will depend on the chosen reader.
     '''
 )
-def generate_report(cfg, input, output, config_reader):
+def report(cfg, input, output, config_reader):
     reader = create_reader(cfg, config_reader)
     df = reader.read(parse_sources(input, cfg))
     df.to_csv(os.path.join(output, 'report.csv'))
@@ -119,10 +119,9 @@ def generate_report(cfg, input, output, config_reader):
     These will depend on the chosen reader.
     '''
 )
-def extract_images(cfg, input, output, config_reader):
+def extract(cfg, input, output, config_reader):
     reader = create_reader(cfg, config_reader)
-    df = reader.read(parse_sources(input, cfg))
-    df.to_csv(os.path.join(output, 'report.csv'))
+    reader.rip_out_jpgs(parse_sources(input, cfg), output)
 
 
 def create_reader(cfg, config_reader):

cleanX/dicom_processing/pydicom_adapter.py

@@ -122,7 +122,7 @@ def rip_out_jpgs(self, source, destination):
         :type source: :class:`~cleanX.dicom_processing.Source`
         :param destination: The name of the directory where JPG files
                             should be stored.
-        :type destination: Compatible with :fn:`os.path.join`
+        :type destination: Compatible with :func:`os.path.join`
         """
         for key, parsed in source.items(dicom.dcmread):
             cv2.imwrite(

cleanX/dicom_processing/simpleitk_adapter.py

@@ -282,7 +282,7 @@ def rip_out_jpgs(self, source, destination):
         :type source: :class:`~cleanX.dicom_processing.Source`
         :param destination: The name of the directory where JPG files
                             should be stored.
-        :type destination: Compatible with :fn:`os.path.join`
+        :type destination: Compatible with :func:`os.path.join`
         """
         reader = sitk.ImageFileReader()
         m_reader = MetadataHelper(reader)

setup.py

@@ -128,13 +128,15 @@ def run(self):
         from sphinx.ext.apidoc import main
 
         src = os.path.join(project_dir, 'docs')
+        special = 'index.rst', 'cli.rst'
 
         for f in glob(os.path.join(src, '*.rst')):
-            if f.endswith('index.rst'):
-                continue
-            if f.endswith('cli.rst'):
-                continue
-            os.unlink(f)
+            for end in special:
+                if f.endswith(end):
+                    os.utime(f, None)
+                    break
+            else:
+                os.unlink(f)
 
         sys.exit(main([
             '-o', src,

test/test_cli.py

@@ -5,6 +5,7 @@
 import subprocess
 
 from tempfile import TemporaryDirectory
+from glob import glob
 
 import pandas as pd
 import pytest
@@ -18,7 +19,7 @@
 
 
 @skip_if_missing('no pydicom available', 'pydicom', 'click')
-def test_cli_pydicom():
+def test_cli_pydicom_report():
     dicomfile_directory1 = os.path.join(
         os.path.dirname(__file__),
         'dicom_example_folder',
@@ -27,7 +28,7 @@ def test_cli_pydicom():
         result = subprocess.call(
             [
                 sys.executable, '-m', 'cleanX',
-                'dicom', 'extract-images',
+                'dicom', 'report',
                 '-i', 'dir', dicomfile_directory1,
                 '-o', td,
             ]
@@ -38,14 +39,48 @@ def test_cli_pydicom():
         assert len(os.listdir(dicomfile_directory1)) == len(df)
 
 
+@skip_if_missing('no pydicom available', 'pydicom', 'click')
+def test_cli_pydicom_extract():
+    dicomfile_directory1 = os.path.join(
+        os.path.dirname(__file__),
+        'dicom_example_folder',
+    )
+    with TemporaryDirectory() as td:
+        result = subprocess.call(
+            [
+                sys.executable, '-m', 'cleanX',
+                'dicom', 'extract',
+                '-i', 'dir', dicomfile_directory1,
+                '-o', td,
+            ]
+        )
+        assert not result
+
+        jpegs = sorted(glob(os.path.join(td, '*.jpg')))
+        dicoms = sorted(os.listdir(dicomfile_directory1))
+        assert len(jpegs) == len(dicoms)
+
+        mismatches = []
+        for jpg, dcm in zip(jpegs, dicoms):
+            jpg = os.path.basename(jpg)
+            jpg = os.path.splitext(jpg)[0]
+
+            dcm = os.path.basename(dcm)
+            dcm = os.path.splitext(dcm)[0]
+
+            if jpg != dcm:
+                mismatches.append((jpg, dcm))
+        assert not mismatches
+
+
 @skip_if_missing('no pydicom available', 'pydicom', 'click')
 def test_cli_datasets():
     resources = os.path.dirname(__file__)
     with TemporaryDirectory() as td:
         result = subprocess.check_output(
             [
                 sys.executable, '-m', 'cleanX',
-                'dataset', 'generate-report',
+                'dataset', 'report',
                 '-r', os.path.join(resources, 'test_sample_df.csv'),
                 '-t', os.path.join(resources, 'train_sample_df.csv'),
                 # These two sets don't appear to have common columns