FileFormats: get_file_readers() replaces readers dict

Orange/data/io.py

@@ -29,7 +29,7 @@
     _io, is_discrete_values, MISSING_VALUES, Table, Domain, Variable,
     DiscreteVariable, StringVariable, ContinuousVariable, TimeVariable,
 )
-from Orange.util import Registry, flatten, namegen
+from Orange.util import Registry, flatten, namegen, deprecated
 
 
 # Support values longer than 128K (i.e. text contents features)
@@ -262,7 +262,6 @@ def split(s):
     '|'.join(flatten(filter(None, i) for i in Flags.ALL.items()))
     ))
 
-
 class FileFormatMeta(Registry):
 
     def __new__(cls, name, bases, attrs):
@@ -287,7 +286,8 @@ def __new__(cls, name, bases, attrs):
 
     @property
     def formats(cls):
-        return cls.registry.values()
+        # also sort by description to make order predictable
+        return iter(sorted(cls.registry.values(), key=lambda x: (x.PRIORITY, x.DESCRIPTION)))
 
     @lru_cache(5)
     def _ext_to_attr_if_attr2(cls, attr, attr2):
@@ -299,7 +299,7 @@ def _ext_to_attr_if_attr2(cls, attr, attr2):
         with the lowest priority.
         """
         formats = OrderedDict()
-        for format in sorted(cls.registry.values(), key=lambda x: x.PRIORITY):
+        for format in cls.formats:
             if not hasattr(format, attr2):
                 continue
             for ext in getattr(format, 'EXTENSIONS', []):
@@ -316,6 +316,7 @@ def writers(cls):
         return cls._ext_to_attr_if_attr2('', 'write_file')
 
     @property
+    @deprecated
     def readers(cls):
         return cls._ext_to_attr_if_attr2('', 'read')
 
@@ -327,6 +328,10 @@ def img_writers(cls):
     def graph_writers(cls):
         return cls._ext_to_attr_if_attr2('', 'write_graph')
 
+    def get_file_readers(cls):
+        return (f for f in cls.formats
+                if getattr(f, 'read', None) and f.EXTENSIONS)
+
 
 class FileFormat(metaclass=FileFormatMeta):
     """
@@ -354,6 +359,8 @@ def write_file(cls, filename, data):
     # Priority when multiple formats support the same extension. Also
     # the sort order in file open/save combo boxes. Lower is better.
     PRIORITY = 10000
+    DESCRIPTION = ""
+    EXTENSIONS = ()
 
     def __init__(self, filename):
         """
@@ -398,12 +405,13 @@ def get_reader(cls, filename):
         -------
         FileFormat
         """
-        for ext, reader in cls.readers.items():
-            # Skip ambiguous, invalid compression-only extensions added on OSX
-            if ext in Compression.all:
-                continue
-            if fnmatch(path.basename(filename), '*' + ext):
-                return reader(filename)
+        for reader in cls.get_file_readers():
+            for ext in reader.EXTENSIONS:
+                # Skip ambiguous, invalid compression-only extensions added on OSX
+                if ext in Compression.all:
+                    continue
+                if fnmatch(path.basename(filename), '*' + ext):
+                    return reader(filename)
 
         raise IOError('No readers for file "{}"'.format(filename))
 
@@ -464,11 +472,13 @@ def locate(cls, filename, search_dirs=('.',)):
         if path.exists(filename):
             return filename
 
+        supported_extensions = set(chain.from_iterable(
+            r.EXTENSIONS for r in cls.get_file_readers()))
         for directory in search_dirs:
             absolute_filename = path.join(directory, filename)
             if path.exists(absolute_filename):
                 break
-            for ext in cls.readers:
+            for ext in supported_extensions:
                 if fnmatch(path.basename(filename), '*' + ext):
                     break
                 # glob uses fnmatch internally

Orange/tests/test_table.py

@@ -1334,9 +1334,11 @@ def test_read_data_calls_reader(self):
         table_mock = Mock(data.Table)
         reader_instance = Mock(read=Mock(return_value=table_mock))
         reader_mock = Mock(return_value=reader_instance)
+        reader_mock.EXTENSIONS = (".xlsx",)
 
-        with patch.dict(data.io.FileFormat.readers,
-                        {'.xlsx': reader_mock}):
+        old_readers = list(data.io.FileFormat.get_file_readers())
+        with patch.object(data.io.FileFormat, "get_file_readers",
+                          lambda: [reader_mock] + old_readers):
             table = data.Table.from_file("test.xlsx")
 
         reader_mock.assert_called_with("test.xlsx")

Orange/widgets/data/owfile.py

@@ -268,8 +268,7 @@ def browse_file(self, in_demos=False):
         else:
             start_file = self.last_path() or os.path.expanduser("~/")
 
-        readers = [f for f in FileFormat.formats
-                   if getattr(f, 'read', None) and getattr(f, "EXTENSIONS", None)]
+        readers = FileFormat.get_file_readers()
         filename, reader, _ = get_file_name_open(start_file, None, readers)
         if not filename:
             return

Orange/widgets/utils/filedialogs.py

@@ -1,3 +1,4 @@
+from itertools import chain
 import os
 
 from AnyQt.QtCore import QFileInfo, Qt
@@ -45,11 +46,12 @@ def dialog_formats():
     """
     Return readable file types for QFileDialogs.
     """
+    formats = list(FileFormat.get_file_readers())
+    all_extensions = sorted(set(chain.from_iterable(
+        r.EXTENSIONS for r in formats)))
     return ("All readable files ({});;".format(
-        '*' + ' *'.join(FileFormat.readers.keys())) +
-            ";;".join("{} (*{})".format(f.DESCRIPTION, ' *'.join(f.EXTENSIONS))
-                      for f in sorted(set(FileFormat.readers.values()),
-                                      key=list(FileFormat.readers.values()).index)))
+        '*' + ' *'.join(all_extensions)) +
+            ";;".join(format_filter(f) for f in formats))
 
 
 def get_file_name(start_dir, start_filter, file_formats):
@@ -118,7 +120,7 @@ def get_file_name_open(start_dir, start_filter, file_formats, title="Open...",
     Returns:
         (filename, filter, file_format), or `(None, None, None)` on cancel
     """
-    file_formats = sorted(set(file_formats), key=lambda w: (w.PRIORITY, w.DESCRIPTION))
+    file_formats = list(file_formats)
     filters = [format_filter(f) for f in file_formats]
 
     # add all readable files option