Merge d92899f into 144deae

ark/utils/io_utils.py

@@ -77,8 +77,8 @@ def list_files(dir_name, substrs=None, exact_match=False):
         matches = [file
                    for file in files
                    if any([
-                        substr == os.path.splitext(file)[0]
-                        for substr in substrs
+                       substr == os.path.splitext(file)[0]
+                       for substr in substrs
                    ])]
     else:
         matches = [file
@@ -176,14 +176,17 @@ def extract_delimited_names(names, delimiter='_', delimiter_optional=True):
     return names
 
 
-def list_folders(dir_name, substrs=None):
+def list_folders(dir_name, substrs=None, exact_match=False):
     """ List all folders in a directory containing at least one given substring
 
     Args:
         dir_name (str):
             Parent directory for folders of interest
         substrs (str or list):
             Substring matching criteria, defaults to None (all folders)
+        exact_match (bool):
+            If True, will match exact folder names (so 'C' will match only 'C/').
+            If False, will match substr pattern in folder (so 'C' will match 'C/' & 'C_DIREC/').
 
     Returns:
         list:
@@ -204,12 +207,21 @@ def list_folders(dir_name, substrs=None):
     if type(substrs) is not list:
         substrs = [substrs]
 
-    matches = [folder
-               for folder in folders
-               if any([
-                   substr in folder
-                   for substr in substrs
-               ])]
+    # Exact match case
+    if exact_match:
+        matches = [folder
+                   for folder in folders
+                   if any([
+                       substr == os.path.splitext(folder)[0]
+                       for substr in substrs
+                   ])]
+    else:
+        matches = [folder
+                   for folder in folders
+                   if any([
+                       substr in folder
+                       for substr in substrs
+                   ])]
 
     return matches
 

ark/utils/io_utils_test.py

@@ -103,11 +103,7 @@ def test_list_files():
 
     # test file name exact matching
     with tempfile.TemporaryDirectory() as temp_dir:
-        filenames = [
-            'chan0.tif',
-            'chan.tif',
-            'c.tif'
-        ]
+        filenames = ['chan0.tif', 'chan.tif', 'c.tif']
         for filename in filenames:
             pathlib.Path(os.path.join(temp_dir, filename)).touch()
 
@@ -129,12 +125,7 @@ def test_list_files():
 
 def test_remove_file_extensions():
     # test a mixture of file paths and extensions
-    files = [
-        'fov1.tiff',
-        'fov2.tif',
-        'fov3.png',
-        'fov4.jpg'
-    ]
+    files = ['fov1.tiff', 'fov2.tif', 'fov3.png', 'fov4.jpg']
 
     assert iou.remove_file_extensions(None) is None
     assert iou.remove_file_extensions([]) == []
@@ -169,28 +160,60 @@ def test_extract_delimited_names():
 
 
 def test_list_folders():
+    # Tests "Fuzzy Substring Matching",`exact_match` = False
     with tempfile.TemporaryDirectory() as temp_dir:
         # set up temp_dir subdirs
         dirnames = [
             'tf_txt',
             'othertf_txt',
             'test_csv',
             'test_out',
+            'test_csv1',
+            'test_csv2',
+            'Ntest_csv',
         ]
+
+        dirnames.sort()
         for dirname in dirnames:
             os.mkdir(os.path.join(temp_dir, dirname))
 
         # add extra file
         pathlib.Path(os.path.join(temp_dir, 'test_badfile.txt')).touch()
 
         # test substrs is None (default)
-        get_all = iou.list_folders(temp_dir)
-        assert get_all.sort() == dirnames.sort()
+        get_all = iou.list_folders(temp_dir, exact_match=False)
+        assert sorted(get_all) == dirnames
 
         # test substrs is not list (single string)
-        get_txt = iou.list_folders(temp_dir, substrs='_txt')
-        assert get_txt.sort() == dirnames[0:2].sort()
+        get_txt = iou.list_folders(temp_dir, substrs='_txt', exact_match=False)
+        assert sorted(get_txt) == sorted(['othertf_txt', 'tf_txt'])
 
         # test substrs is list
-        get_test_and_other = iou.list_folders(temp_dir, substrs=['test_', 'other'])
-        assert get_test_and_other.sort() == dirnames[1:].sort()
+        get_test_and_other = iou.list_folders(
+            temp_dir, substrs=['test_', 'other'], exact_match=False
+        )
+        assert sorted(get_test_and_other) == sorted(
+            ['Ntest_csv', 'test_csv', 'test_csv1', 'test_csv2', 'test_out', 'othertf_txt']
+        )
+
+        # Tests "Exact Substring Matching", `exact_match` = True
+
+        # Test substrs is None (default)
+        get_all = iou.list_folders(temp_dir, exact_match=True)
+        assert sorted(get_all) == sorted(dirnames)
+
+        # Test exact substr is not list (single string)
+        get_othertf_txt = iou.list_folders(temp_dir, substrs='othertf_txt', exact_match=True)
+        assert get_othertf_txt == [dirnames[1]]
+
+        # Test substrs, querying two folders (exactly)
+        get_exact_n_substrs = iou.list_folders(
+            temp_dir, substrs=['tf_txt', 'othertf_txt'], exact_match=True
+        )
+        assert sorted(get_exact_n_substrs) == ['othertf_txt', 'tf_txt']
+
+        # Test the substr that the user specifies which is contained within multiple folders,
+        # and only the folder that exactly matches the substring, not the one that contains it,
+        # is returned when `exact_match=True`
+        get_test_o = iou.list_folders(temp_dir, substrs='test_csv', exact_match=True)
+        assert get_test_o == ["test_csv"]