Addresses #145, based off of @fmassa

torchvision/datasets/folder.py

@@ -32,29 +32,6 @@ def is_image_file(filename):
     return has_file_allowed_extension(filename, IMG_EXTENSIONS)
 
 
-def find_classes(dir):
-    classes = [d for d in os.listdir(dir) if os.path.isdir(os.path.join(dir, d))]
-    classes.sort()
-    class_to_idx = {classes[i]: i for i in range(len(classes))}
-    return classes, class_to_idx
-
-
-def make_dataset(dir, class_to_idx, extensions):
-    images = []
-    dir = os.path.expanduser(dir)
-    for target in sorted(os.listdir(dir)):
-        d = os.path.join(dir, target)
-        if not os.path.isdir(d):
-            continue
-
-        for root, _, fnames in sorted(os.walk(d)):
-            for fname in sorted(fnames):
-                if has_file_allowed_extension(fname, extensions):
-                    path = os.path.join(root, fname)
-                    item = (path, class_to_idx[target])
-                    images.append(item)
-
-    return images
 
 
 class DatasetFolder(data.Dataset):
@@ -86,8 +63,8 @@ class DatasetFolder(data.Dataset):
     """
 
     def __init__(self, root, loader, extensions, transform=None, target_transform=None):
-        classes, class_to_idx = find_classes(root)
-        samples = make_dataset(root, class_to_idx, extensions)
+        classes, class_to_idx = self._find_classes(root)
+        samples = self._make_dataset(root, class_to_idx, extensions)
         if len(samples) == 0:
             raise(RuntimeError("Found 0 files in subfolders of: " + root + "\n"
                                "Supported extensions are: " + ",".join(extensions)))
@@ -104,6 +81,52 @@ def __init__(self, root, loader, extensions, transform=None, target_transform=No
         self.transform = transform
         self.target_transform = target_transform
 
+    def _find_classes(dir):
+        """
+        Finds the classes in a dataset directory.
+
+        Args:
+            dir (string): Root directory path.
+
+        Returns:
+            tuple: (classes, class_to_idx) where class_to_idx is a dictionary
+        """
+        classes = [d for d in os.listdir(dir) if os.path.isdir(os.path.join(dir, d))]
+        classes.sort()
+        class_to_idx = {classes[i]: i for i in range(len(classes))}
+        return classes, class_to_idx
+
+
+    def _make_dataset(dir, class_to_idx, extensions):
+        """
+        A generic method for obtaining paths to all data files.
+
+        Args:
+            dir (string): Root directory path.
+            class_to_idx (dictionary): A mapping of class names to id's.
+            extensions (list): A list of permitted data file extensions.
+
+        Returns:
+            images: A list of (path, target) per data file.
+
+        """
+        images = []
+        dir = os.path.expanduser(dir)
+        for target in sorted(os.listdir(dir)):
+            d = os.path.join(dir, target)
+            if not os.path.isdir(d):
+                continue
+
+            for root, _, fnames in sorted(os.walk(d)):
+                for fname in sorted(fnames):
+                    if has_file_allowed_extension(fname, extensions):
+                        path = os.path.join(root, fname)
+                        item = (path, class_to_idx[target])
+                        images.append(item)
+
+        return images
+
+
     def __getitem__(self, index):
         """
         Args: