allow get_file_names to glob

Author: mmyrte

climada/util/files_handler.py

@@ -11,6 +11,7 @@
 import math
 import requests
 import tqdm
+import glob
 
 LOGGER = logging.getLogger(__name__)
 
@@ -74,11 +75,14 @@ def to_list(num_exp, values, val_name):
     return val_list
 
 def get_file_names(file_name):
-    """Return list of files contained.
+    """ Return list of files contained. Supports globbing.
 
     Parameters:
-        file_name (str or list(str)): file name, or list of file names or name
-            of the folder containing the files
+        file_name (str or list(str)): Either a single string or a list of
+            strings that are either 
+                - a file path
+                - or the path of the folder containing the files
+                - or a globbing pattern.
 
     Returns:
         list
@@ -92,12 +96,18 @@ def get_file_names(file_name):
     return file_list
 
 def _process_one_file_name(name, file_list):
-    """Apend to input list the file contained in name"""
-    if os.path.splitext(name)[1] == '':
-        tmp_files = os.listdir(name)
-        # append only files (absolute path), not folders
+    """ Apend to input list the file contained in name
+        Tries globbing if name is neither dir nor file.
+    """
+    if os.path.isdir(name):
+        tmp_files = glob.glob(os.path.join(name, '*'))
         for file in tmp_files:
-            if os.path.splitext(file)[1] != '':
-                file_list.append(os.path.join(name, file))
-    else:
+            if os.path.isfile(file):
+                file_list.append(file)
+    if os.path.isfile(name):
         file_list.append(name)
+    else:
+        tmp_files = sorted(glob.glob(name))
+        for file in tmp_files:
+            if os.path.isfile(file):
+                file_list.append(file)

climada/util/test/test_files.py

@@ -6,7 +6,7 @@
 import unittest
 
 from climada.util.files_handler import to_list, get_file_names, download_file
-from climada.util.constants import DATA_DIR
+from climada.util.constants import DATA_DIR, GLB_CENTROIDS_MAT, ENT_TEMPLATE_XLS
 
 class TestDownloadUrl(unittest.TestCase):
     """Test download_file function """
@@ -53,16 +53,17 @@ def test_list_wrong_length_fail(self):
         self.assertIn("Provide one or 3 values.", cm.output[0])
 
 class TestGetFileNames(unittest.TestCase):
-    """Test get_file_names function"""
+    """ Test get_file_names function. Only works with actually existing
+        files and directories. """
     def test_one_file_copy(self):
         """If input is one file name, return a list with this file name"""
-        file_name = "test.mat"
+        file_name = GLB_CENTROIDS_MAT
         out = get_file_names(file_name)
         self.assertEqual([file_name], out)
 
     def test_several_file_copy(self):
         """If input is a list with several file names, return the same list"""
-        file_name = ["test1.mat", "test2.mat", "test3.mat", "test4.mat"]
+        file_name = [GLB_CENTROIDS_MAT, ENT_TEMPLATE_XLS]
         out = get_file_names(file_name)
         self.assertEqual(file_name, out)
 
@@ -79,6 +80,19 @@ def test_folder_contents(self):
         for file in out:
             self.assertNotEqual('', os.path.splitext(file)[1])
 
+    def test_globbing(self):
+        """ If input is a glob pattern, return a list of matching visible
+            files; omit folders.
+        """
+        file_name = os.path.join(DATA_DIR, 'demo/')
+        out = get_file_names(file_name + '*')
+
+        tmp_files = os.listdir(file_name)
+        tmp_files = [file_name + f for f in tmp_files]
+        tmp_files = [f for f in tmp_files if not os.path.isdir(f) 
+                and not f.startswith('.')]
+        self.assertEqual(tmp_files, out)
+
 # Execute Tests
 TESTS = unittest.TestLoader().loadTestsFromTestCase(TestToStrList)
 TESTS.addTests(unittest.TestLoader().loadTestsFromTestCase(TestGetFileNames))