Implement StubParserTreeFilter.values

jedi/api/classes.py

@@ -204,7 +204,6 @@ class BaseDefinition(object):
 
     def in_builtin_module(self):
         """Whether this is a builtin module."""
-        print(self._module)
         if isinstance(self._module, StubOnlyModuleContext):
             return any(isinstance(context, compiled.CompiledObject)
                        for context in self._module.non_stub_context_set)

jedi/plugins/typeshed.py

@@ -10,9 +10,9 @@ from jedi.parser_utils import get_call_signature_for_any
 from jedi.evaluate.base_context import ContextSet, iterator_to_context_set, \
     ContextWrapper
 from jedi.evaluate.filters import AbstractTreeName, ParserTreeFilter, \
-    TreeNameDefinition, NameWrapper
+    TreeNameDefinition, NameWrapper, MergedFilter
 from jedi.evaluate.context import ModuleContext, FunctionContext, \
-    ClassContext, BoundMethod
+    ClassContext
 from jedi.evaluate.context.typing import TypingModuleFilterWrapper, \
     TypingModuleName
 from jedi.evaluate.compiled import CompiledObject
@@ -251,8 +251,24 @@ class StubParserTreeFilter(ParserTreeFilter):
 
         return self._convert_names(self._filter(names))
 
-    # TODO maybe implement values, because currently the names that don't exist
-    # in the stub file are not part of values.
+    def values(self):
+        used_stub_names = set()
+        result_names = []
+        for key_name, names in self._used_names.items():
+            result_names += self._convert_names(self._filter(names))
+            used_stub_names.add(key_name)
+
+        non_stub_filters = []
+        for f in self._non_stub_filters:
+            if isinstance(f, MergedFilter):
+                non_stub_filters += f._filters
+            else:
+                non_stub_filters.append(f)
+        for non_stub_filter in non_stub_filters:
+            for key_name in non_stub_filter._used_names:
+                if key_name not in used_stub_names:
+                    result_names += non_stub_filter.get(key_name)
+        return result_names
 
     def _check_flows(self, names):
         return names
@@ -361,6 +377,7 @@ class StubOnlyModuleContext(ModuleContext):
 
     def _get_first_non_stub_filters(self):
         for context in self.non_stub_context_set:
+            if not isinstance(context, CompiledObject):
                 yield next(context.get_filters(search_global=False))
 
     def get_filters(self, search_global, until_position=None,
@@ -368,8 +385,10 @@ class StubOnlyModuleContext(ModuleContext):
         filters = super(StubOnlyModuleContext, self).get_filters(
             search_global, until_position, origin_scope, **kwargs
         )
-        next(filters)  # Ignore the first filter and replace it with our own
+        first_non_stub_filters = list(self._get_first_non_stub_filters())
 
+        if first_non_stub_filters:
+            next(filters)  # Ignore the first filter and replace it with our own
             # Here we remap the names from stubs to the actual module. This is
             # important if type inferences is needed in that module.
             yield StubParserTreeFilter(

test/test_api/test_classes.py

@@ -116,8 +116,9 @@ def test_class_call_signature(Script):
 
 def test_position_none_if_builtin(Script):
     gotos = Script('import sys; sys.path').goto_assignments()
-    assert gotos[0].line is None
-    assert gotos[0].column is None
+    assert gotos[0].in_builtin_module()
+    assert gotos[0].line is not None
+    assert gotos[0].column is not None
 
 
 def test_completion_docstring(Script, jedi_path):