Revisit reference finding, scan a lot of folders

jedi/file_io.py

@@ -43,9 +43,11 @@ class FolderIO(AbstractFolderIO):
                 modified_folder_ios,
                 [FileIO(os.path.join(root, f)) for f in files],
             )
-            modified_iterator = iter(modified_folder_ios)
+            modified_iterator = iter(reversed(modified_folder_ios))
             current = next(modified_iterator, None)
-            for i, folder_io in enumerate(original_folder_ios):
+            i = len(original_folder_ios)
+            for folder_io in reversed(original_folder_ios):
+                i -= 1   # Basically enumerate but reversed
                 if current is folder_io:
                     current = next(modified_iterator, None)
                 else:

jedi/inference/dynamic_params.py

@@ -116,8 +116,14 @@ def _search_function_arguments(module_context, funcdef, string_name):
     found_arguments = False
     i = 0
     inference_state = module_context.inference_state
-    for for_mod_context in get_module_contexts_containing_name(
+
-            inference_state, [module_context], string_name):
+    if settings.dynamic_params_for_other_modules:
+        module_contexts = get_module_contexts_containing_name(
+            inference_state, [module_context], string_name)
+    else:
+        module_contexts = [module_context]
+
+    for for_mod_context in module_contexts:
         for name, trailer in _get_potential_nodes(for_mod_context, string_name):
             i += 1
 

jedi/inference/references.py

@@ -1,4 +1,4 @@
-import os
+import re
 
 from parso import python_bytes_to_unicode
 
@@ -9,6 +9,8 @@ from jedi.inference.compiled import CompiledObject
 from jedi.inference.filters import ParserTreeFilter
 from jedi.inference.gradual.conversion import convert_names
 
+_IGNORE_FOLDERS = ('.tox', 'venv', '__pycache__')
+
 
 def _resolve_names(definition_names, avoid_names=()):
     for name in definition_names:
@@ -113,11 +115,12 @@ def find_references(module_context, tree_name):
     found_names_dct = _dictionarize(found_names)
 
     module_contexts = set(d.get_root_context() for d in found_names)
-    module_contexts = set(m for m in module_contexts if not m.is_compiled())
 
     non_matching_reference_maps = {}
     potential_modules = get_module_contexts_containing_name(
-        inf, module_contexts, search_name
+        inf,
+        [module_context] + [m for m in module_contexts if m != module_context],
+        search_name
     )
     for module_context in potential_modules:
         for name_leaf in module_context.tree_node.get_used_names().get(search_name, []):
@@ -139,59 +142,80 @@ def find_references(module_context, tree_name):
     return found_names_dct.values()
 
 
+def _check_fs(inference_state, file_io, regex):
+    try:
+        code = file_io.read()
+    except FileNotFoundError:
+        return None
+    code = python_bytes_to_unicode(code, errors='replace')
+    if not regex.search(code):
+        return None
+    new_file_io = KnownContentFileIO(file_io.path, code)
+    m = load_module_from_path(inference_state, new_file_io)
+    if isinstance(m, CompiledObject):
+        return None
+    return m.as_context()
+
+
+def _recurse_find_python_files(folder_io, except_paths):
+    for root_folder_io, folder_ios, file_ios in folder_io.walk():
+        # Delete folders that we don't want to iterate over.
+        folder_ios[:] = [
+            folder_io
+            for folder_io in folder_ios
+            if folder_io.path not in except_paths
+            and folder_io.get_base_name() not in _IGNORE_FOLDERS
+        ]
+
+        for file_io in file_ios:
+            path = file_io.path
+            if path.endswith('.py') or path.endswith('.pyi'):
+                if path not in except_paths:
+                    yield file_io
+
+
+def _find_python_files_in_sys_path(inference_state, folder_io):
+    sys_path = inference_state.get_sys_path()
+    except_paths = set()
+    while True:
+        path = folder_io.path
+        if not any(path.startswith(p) for p in sys_path):
+            break
+        for file_io in _recurse_find_python_files(folder_io, except_paths):
+            yield file_io
+        except_paths.add(path)
+        folder_io = folder_io.get_parent_folder()
+
+
 def get_module_contexts_containing_name(inference_state, module_contexts, name):
     """
     Search a name in the directories of modules.
     """
-    def check_directory(folder_io):
+    def iter_file_ios():
-        for file_name in folder_io.list():
+        yielded_paths = [m.py__file__() for m in module_contexts]
-            if file_name.endswith('.py'):
+        for module_context in module_contexts:
-                yield folder_io.get_file_io(file_name)
-
-    def check_fs(file_io, base_names):
-        try:
-            code = file_io.read()
-        except FileNotFoundError:
-            return None
-        code = python_bytes_to_unicode(code, errors='replace')
-        if name not in code:
-            return None
-        new_file_io = KnownContentFileIO(file_io.path, code)
-        m = load_module_from_path(inference_state, new_file_io, base_names)
-        if isinstance(m, CompiledObject):
-            return None
-        return m.as_context()
-
-    # skip non python modules
-    used_mod_paths = set()
-    folders_with_names_to_be_checked = []
-    for module_context in module_contexts:
-        path = module_context.py__file__()
-        if path not in used_mod_paths:
             file_io = module_context.get_value().file_io
-            if file_io is not None:
+            if file_io is None:
-                used_mod_paths.add(path)
+                continue
-                folders_with_names_to_be_checked.append((
+
-                    file_io.get_parent_folder(),
+            folder_io = file_io.get_parent_folder()
-                    module_context.get_value().py__package__()
+            for file_io in _find_python_files_in_sys_path(inference_state, folder_io):
-                ))
+                if file_io.path not in yielded_paths:
+                    yield file_io
+
+    # Skip non python modules
+    for module_context in module_contexts:
+        if module_context.is_compiled():
+            continue
         yield module_context
 
-    if not settings.dynamic_params_for_other_modules:
+    if len(name) <= 2 or name == 'self':
         return
 
-    def get_file_ios_to_check():
+    file_io_count = 0
-        for folder_io, base_names in folders_with_names_to_be_checked:
+    regex = re.compile(r'\b' + re.escape(name) + r'\b')
-            for file_io in check_directory(folder_io):
+    for file_io in iter_file_ios():
-                if file_io.path not in used_mod_paths:
+        file_io_count += 1
-                    yield file_io, base_names
+        m = _check_fs(inference_state, file_io, regex)
-
-        for p in settings.additional_dynamic_modules:
-            p = os.path.abspath(p)
-            if p not in used_mod_paths:
-                yield FileIO(p), None
-
-    for file_io, base_names in get_file_ios_to_check():
-        m = check_fs(file_io, base_names)
         if m is not None:
             yield m

test/test_file_io.py

@@ -17,3 +17,11 @@ def test_folder_io_walk():
     assert root.path == join(root_dir, 'ns2')
     folder_ios.clear()
     assert next(iterator, None) is None
+
+
+def test_folder_io_walk2():
+    root_dir = get_example_dir('namespace_package')
+    iterator = FolderIO(root_dir).walk()
+    root, folder_ios, file_ios = next(iterator)
+    folder_ios.clear()
+    assert next(iterator, None) is None