owfile: support readers with repeated extensions

FileFormat uses PRIORITY for preference. When formats share extension,
use the format with the lowest priority.

Author: markotoplak

Orange/data/io.py

@@ -294,11 +294,18 @@ def _ext_to_attr_if_attr2(cls, attr, attr2):
         """
         Return ``{ext: `attr`, ...}`` dict if ``cls`` has `attr2`.
         If `attr` is '', return ``{ext: cls, ...}`` instead.
+
+        If there are multiple formats for an extension, return a format
+        with the lowest priority.
         """
-        return OrderedDict((ext, getattr(cls, attr, cls))
-                           for cls in cls.registry.values()
-                           if hasattr(cls, attr2)
-                           for ext in getattr(cls, 'EXTENSIONS', []))
+        formats = OrderedDict()
+        for format in sorted(cls.registry.values(), key=lambda x: x.PRIORITY):
+            if not hasattr(format, attr2):
+                continue
+            for ext in getattr(format, 'EXTENSIONS', []):
+                # Only adds if not yet registered
+                formats.setdefault(ext, getattr(format, attr, format))
+        return formats
 
     @property
     def names(cls):
@@ -344,7 +351,9 @@ def write_file(cls, filename, data):
     iterable (list (rows) of lists of values (cols)).
     """
 
-    PRIORITY = 10000  # Sort order in OWSave widget combo box, lower is better
+    # Priority when multiple formats support the same extension. Also
+    # the sort order in file open/save combo boxes. Lower is better.
+    PRIORITY = 10000
 
     def __init__(self, filename):
         """

Orange/tests/test_io.py

@@ -9,6 +9,7 @@
 from Orange.data.io import FileFormat, TabReader, CSVReader, PickleReader
 from Orange.data.table import get_sample_datasets_dir
 
+
 class WildcardReader(FileFormat):
     EXTENSIONS = ('.wild', '.wild[0-9]')
     DESCRIPTION = "Dummy reader for testing extensions"
@@ -35,6 +36,30 @@ def test_wildcard_extension(self):
             FileFormat.get_reader("t.wild2a")
 
 
+class SameExtension(FileFormat):
+    PRIORITY = 100
+    EXTENSIONS = ('.same_extension',)
+    DESCRIPTION = "Same extension, different priority"
+
+    def read(self):
+        pass
+
+
+class SameExtensionPreferred(SameExtension):
+    PRIORITY = 90
+
+
+class SameExtensionL(SameExtension):
+    PRIORITY = 110
+
+
+class TestMultipleSameExtension(unittest.TestCase):
+
+    def test_find_reader(self):
+        reader = FileFormat.get_reader("some.same_extension")
+        self.assertIsInstance(reader, SameExtensionPreferred)
+
+
 class TestLocate(unittest.TestCase):
 
     def test_locate_sample_datasets(self):
@@ -49,7 +74,6 @@ def test_locate_sample_datasets(self):
                                  search_dirs=[get_sample_datasets_dir()])
         self.assertEqual(os.path.basename(iris), "iris.tab")
 
-
     def test_locate_wildcard_extension(self):
         tempdir = tempfile.mkdtemp()
         with self.assertRaises(OSError):

Orange/widgets/data/owfile.py

@@ -269,7 +269,9 @@ def browse_file(self, in_demos=False):
         else:
             start_file = self.last_path() or os.path.expanduser("~/")
 
-        filename, reader, _ = open_filename_dialog(start_file, None, FileFormat.readers)
+        readers = [f for f in FileFormat.formats
+                   if getattr(f, 'read', None) and getattr(f, "EXTENSIONS", None)]
+        filename, reader, _ = open_filename_dialog(start_file, None, readers)
         if not filename:
             return
         self.add_path(filename)

Orange/widgets/data/tests/test_owfile.py

@@ -54,6 +54,15 @@ def read(self):
         return Orange.data.Table("iris")
 
 
+class MyCustomTabReader(FileFormat):
+    EXTENSIONS = ('.tab',)
+    DESCRIPTION = "Always return iris"
+    PRIORITY = 999999
+
+    def read(self):
+        return Orange.data.Table("iris")
+
+
 class TestOWFile(WidgetTest):
     # Attribute used to store event data so it does not get garbage
     # collected before event is processed.
@@ -230,6 +239,17 @@ def test_check_datetime_disabled(self):
                 vartype_delegate.setEditorData(combo, idx(i))
                 self.assertEqual(combo.count(), counts[i])
 
+    def test_reader_custom_tab(self):
+        with named_file("", suffix=".tab") as fn:
+            qname = MyCustomTabReader.qualified_name()
+            reader = RecentPath(fn, None, None, file_format=qname)
+            self.widget = self.create_widget(OWFile,
+                                             stored_settings={"recent_paths": [reader]})
+            self.widget.load_data()
+        self.assertFalse(self.widget.Error.missing_reader.is_shown())
+        outdata = self.get_output(self.widget.Outputs.data)
+        self.assertEqual(len(outdata), 150)  # loaded iris
+
     def test_no_reader_extension(self):
         with named_file("", suffix=".xyz_unknown") as fn:
             no_reader = RecentPath(fn, None, None)

Orange/widgets/utils/filedialogs.py

@@ -115,13 +115,13 @@ def open_filename_dialog(start_dir, start_filter, file_formats, title="Open...",
     Args:
         start_dir (str): initial directory, optionally including the filename
         start_filter (str): initial filter
-        file_formats (dict {extension: Orange.data.io.FileFormat}): file formats
+        file_formats (a list of Orange.data.io.FileFormat): file formats
         title (str): title of the dialog
         dialog: a function that creates a QT dialog
     Returns:
         (filename, file_format, filter), or `(None, None, None)` on cancel
     """
-    file_formats = sorted(set(file_formats.values()), key=lambda w: w.PRIORITY)
+    file_formats = sorted(set(file_formats), key=lambda w: (w.PRIORITY, w.DESCRIPTION))
     filters = [format_filter(f) for f in file_formats]
 
     # add all readable files option