[WIP] make unicode vs bytes more distinct in py2

stdlib/2/_threading_local.pyi

@@ -1,14 +1,14 @@
 # Source: https://hg.python.org/cpython/file/2.7/Lib/_threading_local.py
-from typing import Any, List
+from typing import Any, List, Text
 
-__all__: List[str]
+__all__: List[Text]
 
 class _localbase(object): ...
 
 class local(_localbase):
-    def __getattribute__(self, name: str) -> Any: ...
-    def __setattr__(self, name: str, value: Any) -> None: ...
-    def __delattr__(self, name: str) -> None: ...
+    def __getattribute__(self, name: Text) -> Any: ...
+    def __setattr__(self, name: Text, value: Any) -> None: ...
+    def __delattr__(self, name: Text) -> None: ...
     def __del__(self) -> None: ...
 
 def _patch(self: local) -> None: ...

stdlib/2/encodings/utf_8.pyi

@@ -1,11 +1,12 @@
 import codecs
+from six import text_type as unicode
 from typing import Text, Tuple
 
 class IncrementalEncoder(codecs.IncrementalEncoder):
     def encode(self, input: Text, final: bool = ...) -> bytes: ...
 
 class IncrementalDecoder(codecs.BufferedIncrementalDecoder):
-    def _buffer_decode(self, input: bytes, errors: str, final: bool) -> Tuple[Text, int]: ...
+    def _buffer_decode(self, input: bytes, errors: str, final: bool) -> Tuple[unicode, int]: ...
 
 class StreamWriter(codecs.StreamWriter): ...
 class StreamReader(codecs.StreamReader): ...

stdlib/2/json.pyi

@@ -20,7 +20,7 @@ def dumps(obj: Any,
           **kwds: Any) -> str: ...
 
 def dump(obj: Any,
-         fp: Union[IO[str], IO[Text]],
+         fp: Union[IO[str], IO[unicode]],
          skipkeys: bool = ...,
          ensure_ascii: bool = ...,
          check_circular: bool = ...,

stdlib/2/string.pyi

@@ -2,7 +2,7 @@
 
 # Based on http://docs.python.org/3.2/library/string.html
 
-from typing import Mapping, Sequence, Any, Optional, Union, List, Tuple, Iterable, AnyStr
+from typing import Mapping, Sequence, Any, Optional, Union, List, Text, Tuple, Iterable, AnyStr
 
 ascii_letters = ...  # type: str
 ascii_lowercase = ...  # type: str
@@ -70,5 +70,5 @@ class Formatter(object):
         raise KeyError()
     def check_unused_args(self, used_args: Sequence[Union[int, str]], args: Sequence[Any],
                           kwargs: Mapping[str, Any]) -> None: ...
-    def format_field(self, value: Any, format_spec: str) -> Any: ...
+    def format_field(self, value: Any, format_spec: Text) -> Any: ...
     def convert_field(self, value: Any, conversion: str) -> Any: ...

stdlib/2/tempfile.pyi

@@ -25,7 +25,7 @@ class _TemporaryFileWrapper(IO[str]):
     def __del__(self) -> None: ...
     def __enter__(self) -> _TemporaryFileWrapper: ...
     def __exit__(self, exc, value, tb) -> bool: ...
-    def __getattr__(self, name: unicode) -> Any: ...
+    def __getattr__(self, name: Text) -> Any: ...
     def close(self) -> None: ...
     def unlink(self, path: unicode) -> None: ...
     # These methods don't exist directly on this object, but

stdlib/2/toaiff.pyi

@@ -2,10 +2,10 @@
 
 # Source: https://hg.python.org/cpython/file/2.7/Lib/toaiff.py
 from pipes import Template
-from typing import Dict, List
+from typing import Dict, List, Text
 
 
-__all__: List[str]
+__all__: List[Text]
 table: Dict[str, Template]
 t: Template
 uncompress: Template

stdlib/2/types.pyi

@@ -3,7 +3,7 @@
 
 from typing import (
     Any, Callable, Dict, Iterable, Iterator, List, Optional,
-    Tuple, Type, TypeVar, Union, overload,
+    Text, Tuple, Type, TypeVar, Union, overload,
 )
 
 _T = TypeVar('_T')
@@ -33,9 +33,9 @@ class FunctionType:
     func_closure = ...  # type: Optional[Tuple[_Cell, ...]]
     func_code = ...  # type: CodeType
     func_defaults = ...  # type: Optional[Tuple[Any, ...]]
-    func_dict = ...  # type: Dict[str, Any]
-    func_doc = ...  # type: Optional[str]
-    func_globals = ...  # type: Dict[str, Any]
+    func_dict = ...  # type: Dict[Text, Any]
+    func_doc = ...  # type: Optional[Text]
+    func_globals = ...  # type: Dict[Text, Any]
     func_name = ...  # type: str
     __closure__ = func_closure
     __code__ = func_code
@@ -98,13 +98,13 @@ class BuiltinFunctionType:
 BuiltinMethodType = BuiltinFunctionType
 
 class ModuleType:
-    __doc__ = ...  # type: Optional[str]
-    __file__ = ...  # type: Optional[str]
-    __name__ = ...  # type: str
-    __package__ = ...  # type: Optional[str]
-    __path__ = ...  # type: Optional[Iterable[str]]
-    __dict__ = ...  # type: Dict[str, Any]
-    def __init__(self, name: str, doc: Optional[str] = ...) -> None: ...
+    __doc__ = ...  # type: Optional[Text]
+    __file__ = ...  # type: Optional[Text]
+    __name__ = ...  # type: Text
+    __package__ = ...  # type: Optional[Text]
+    __path__ = ...  # type: Optional[Iterable[Text]]
+    __dict__ = ...  # type: Dict[Text, Any]
+    def __init__(self, name: Text, doc: Optional[Text] = ...) -> None: ...
 FileType = file
 XRangeType = xrange
 
@@ -116,15 +116,15 @@ class TracebackType:
 
 class FrameType:
     f_back = ...  # type: FrameType
-    f_builtins = ...  # type: Dict[str, Any]
+    f_builtins = ...  # type: Dict[Text, Any]
     f_code = ...  # type: CodeType
     f_exc_type = ...  # type: None
     f_exc_value = ...  # type: None
     f_exc_traceback = ...  # type: None
-    f_globals = ...  # type: Dict[str, Any]
+    f_globals = ...  # type: Dict[Text, Any]
     f_lasti = ...  # type: int
     f_lineno = ...  # type: int
-    f_locals = ...  # type: Dict[str, Any]
+    f_locals = ...  # type: Dict[Text, Any]
     f_restricted = ...  # type: bool
     f_trace = ...  # type: Callable[[], None]
 

stdlib/2/typing.pyi

@@ -277,7 +277,7 @@ class MutableMapping(Mapping[_KT, _VT], Generic[_KT, _VT]):
     @overload
     def update(self, **kwargs: _VT) -> None: ...
 
-Text = unicode
+Text = Union[bytes, unicode]
 
 TYPE_CHECKING = True
 
@@ -397,7 +397,7 @@ class Match(Generic[AnyStr]):
 # Pattern is generic over AnyStr (determining the type of its .pattern
 # attribute), but at the same time its methods take either bytes or
 # Text and return the same type, regardless of the type of the pattern.
-_AnyStr2 = TypeVar('_AnyStr2', bytes, Text)
+_AnyStr2 = TypeVar('_AnyStr2', bytes, unicode, Text)
 
 class Pattern(Generic[AnyStr]):
     flags: int

stdlib/2/unittest.pyi

@@ -12,7 +12,7 @@ _T = TypeVar('_T')
 _FT = TypeVar('_FT')
 
 _ExceptionType = Union[Type[BaseException], Tuple[Type[BaseException], ...]]
-_Regexp = Union[Text, Pattern[Text]]
+_Regexp = Union[Text, Pattern[unicode], Pattern[str]]
 
 class Testable(metaclass=ABCMeta):
     @abstractmethod

stdlib/2and3/codecs.pyi

@@ -1,3 +1,4 @@
+from six import text_type as unicode
 import sys
 from typing import Any, BinaryIO, Callable, Generator, IO, Iterable, Iterator, List, Optional, Protocol, Text, TextIO, Tuple, Type, TypeVar, Union
 
@@ -10,7 +11,7 @@ import types
 # There *are* bytes->bytes and str->str encodings in the standard library.
 # They are much more common in Python 2 than in Python 3.
 
-_Decoded = Text
+_Decoded = unicode
 _Encoded = bytes
 
 class _Encoder(Protocol):
@@ -171,7 +172,7 @@ class StreamReaderWriter(TextIO):
     if sys.version_info >= (3,):
         def __next__(self) -> Text: ...
     else:
-        def next(self) -> Text: ...
+        def next(self) -> unicode: ...
     def __iter__(self: _T) -> _T: ...
     # This actually returns None, but that's incompatible with the supertype
     def write(self, data: _Decoded) -> int: ...

stdlib/2and3/ctypes/__init__.pyi

@@ -234,11 +234,11 @@ class _StructUnionMeta(_CDataMeta):
     _fields_: Sequence[_UnionT[Tuple[str, Type[_CData]], Tuple[str, Type[_CData], int]]] = ...
     _pack_: int = ...
     _anonymous_: Sequence[str] = ...
-    def __getattr__(self, name: str) -> _CField: ...
+    def __getattr__(self, name: Text) -> _CField: ...
 class _StructUnionBase(_CData, metaclass=_StructUnionMeta):
     def __init__(self, *args: Any, **kw: Any) -> None: ...
-    def __getattr__(self, name: str) -> Any: ...
-    def __setattr__(self, name: str, value: Any) -> None: ...
+    def __getattr__(self, name: Text) -> Any: ...
+    def __setattr__(self, name: Text, value: Any) -> None: ...
 
 class Union(_StructUnionBase): ...
 class Structure(_StructUnionBase): ...

stdlib/2and3/decimal.pyi

@@ -2,7 +2,7 @@ import numbers
 import sys
 from types import TracebackType
 from typing import (
-    Any, Container, Dict, List, NamedTuple, Optional, Sequence, Text, Tuple, Type, TypeVar, Union,
+    Any, AnyStr, Container, Dict, List, NamedTuple, Optional, Sequence, Text, Tuple, Type, TypeVar, Union,
 )
 
 _Decimal = Union[Decimal, int]
@@ -193,7 +193,7 @@ class Decimal(object):
     def __reduce__(self) -> Tuple[Type[Decimal], Tuple[str]]: ...
     def __copy__(self) -> Decimal: ...
     def __deepcopy__(self, memo: Any) -> Decimal: ...
-    def __format__(self, specifier: str, context: Optional[Context] = ...) -> str: ...
+    def __format__(self, specifier: AnyStr, context: Optional[Context] = ...) -> AnyStr: ...
 
 class _ContextManager(object):
     new_context: Context

stdlib/2and3/imaplib.pyi

@@ -1,6 +1,7 @@
 # Stubs for imaplib (Python 2)
 
 import imaplib
+from six import text_type as unicode
 import subprocess
 import sys
 import time
@@ -15,7 +16,7 @@ class IMAP4:
     error: Type[Exception] = ...
     abort: Type[Exception] = ...
     readonly: Type[Exception] = ...
-    mustquote: Pattern[Text] = ...
+    mustquote: Union[Pattern[str],Pattern[unicode]] = ...
     debug: int = ...
     state: str = ...
     literal: Optional[Text] = ...
@@ -25,7 +26,7 @@ class IMAP4:
     is_readonly: bool = ...
     tagnum: int = ...
     tagpre: str = ...
-    tagre: Pattern[Text] = ...
+    tagre: Union[Pattern[str],Pattern[unicode]] = ...
     welcome: bytes = ...
     capabilities: Tuple[str] = ...
     PROTOCOL_VERSION: str = ...
@@ -34,7 +35,7 @@ class IMAP4:
     host: str = ...
     port: int = ...
     sock: _socket = ...
-    file: Union[IO[Text], IO[bytes]] = ...
+    file: Union[IO[unicode], IO[bytes]] = ...
     def open(self, host: str = ..., port: int = ...) -> None: ...
     def read(self, size: int) -> bytes: ...
     def readline(self) -> bytes: ...

stdlib/2and3/lib2to3/pgen2/driver.pyi

@@ -1,6 +1,7 @@
 # Stubs for lib2to3.pgen2.driver (Python 3.6)
 
 import os
+from six import text_type as unicode
 import sys
 from typing import Any, Callable, IO, Iterable, List, Optional, Text, Tuple, Union
 
@@ -16,8 +17,8 @@ class Driver:
     convert: _Convert
     def __init__(self, grammar: Grammar, convert: Optional[_Convert] = ..., logger: Optional[Logger] = ...) -> None: ...
     def parse_tokens(self, tokens: Iterable[Any], debug: bool = ...) -> _NL: ...
-    def parse_stream_raw(self, stream: IO[Text], debug: bool = ...) -> _NL: ...
-    def parse_stream(self, stream: IO[Text], debug: bool = ...) -> _NL: ...
+    def parse_stream_raw(self, stream: Union[IO[str],IO[unicode]], debug: bool = ...) -> _NL: ...
+    def parse_stream(self, stream: Union[IO[str],IO[unicode]], debug: bool = ...) -> _NL: ...
     def parse_file(self, filename: _Path, encoding: Optional[Text] = ..., debug: bool = ...) -> _NL: ...
     def parse_string(self, text: Text, debug: bool = ...) -> _NL: ...
 

stdlib/2and3/lib2to3/pgen2/pgen.pyi

@@ -1,7 +1,8 @@
 # Stubs for lib2to3.pgen2.pgen (Python 3.6)
 
+from six import text_type as unicode
 from typing import (
-    Any, Dict, IO, Iterable, Iterator, List, NoReturn, Optional, Text, Tuple
+    Any, Dict, IO, Iterable, Iterator, List, NoReturn, Optional, Text, Tuple, Union
 )
 
 from lib2to3.pgen2 import _Path, grammar
@@ -11,10 +12,10 @@ class PgenGrammar(grammar.Grammar): ...
 
 class ParserGenerator:
     filename: _Path
-    stream: IO[Text]
+    stream: Union[IO[str],IO[unicode]]
     generator: Iterator[_TokenInfo]
     first: Dict[Text, Dict[Text, int]]
-    def __init__(self, filename: _Path, stream: Optional[IO[Text]] = ...) -> None: ...
+    def __init__(self, filename: _Path, stream: Optional[Union[IO[str],IO[unicode]]] = ...) -> None: ...
     def make_grammar(self) -> PgenGrammar: ...
     def make_first(self, c: PgenGrammar, name: Text) -> Dict[int, int]: ...
     def make_label(self, c: PgenGrammar, label: Text) -> int: ...

stdlib/2and3/logging/__init__.pyi

@@ -51,7 +51,7 @@ class Logger(Filterer):
     def setLevel(self, lvl: Union[int, str]) -> None: ...
     def isEnabledFor(self, lvl: int) -> bool: ...
     def getEffectiveLevel(self) -> int: ...
-    def getChild(self, suffix: str) -> Logger: ...
+    def getChild(self, suffix: Text) -> Logger: ...
     if sys.version_info >= (3,):
         def debug(self, msg: Any, *args: Any, exc_info: _ExcInfoType = ...,
                   stack_info: bool = ..., extra: Optional[Dict[str, Any]] = ...,

stdlib/2and3/plistlib.pyi

@@ -2,7 +2,7 @@
 
 from typing import (
     Any, IO, Mapping, MutableMapping, Optional, Union,
-    Type, TypeVar,
+    Text, Type, TypeVar,
 )
 from typing import Dict as DictT
 import sys
@@ -48,9 +48,9 @@ if sys.version_info < (3,):
 
 if sys.version_info < (3, 7):
     class Dict(dict):
-        def __getattr__(self, attr: str) -> Any: ...
-        def __setattr__(self, attr: str, value: Any) -> None: ...
-        def __delattr__(self, attr: str) -> None: ...
+        def __getattr__(self, attr: Text) -> Any: ...
+        def __setattr__(self, attr: Text, value: Any) -> None: ...
+        def __delattr__(self, attr: Text) -> None: ...
 
 class Data:
     data = ...  # type: bytes

stdlib/2and3/poplib.pyi

@@ -1,11 +1,12 @@
 # Stubs for poplib (Python 2 and 3)
 
 import socket
+from six import text_type as unicode
 import ssl
 import sys
 from typing import (
     Any, BinaryIO, Dict, List, NoReturn, Optional, overload, Pattern, Text,
-    Tuple,
+    Tuple, Union,
 )
 
 _LongResp = Tuple[bytes, List[bytes], int]
@@ -44,7 +45,7 @@ class POP3:
     def close(self) -> None: ...
     def rpop(self, user: Text) -> bytes: ...
 
-    timestamp: Pattern[Text]
+    timestamp: Union[Pattern[str],Pattern[unicode]]
 
     if sys.version_info < (3, 0):
         def apop(self, user: Text, secret: Text) -> bytes: ...

stdlib/2and3/threading.pyi

@@ -1,7 +1,7 @@
 # Stubs for threading
 
 from typing import (
-    Any, Callable, Iterable, List, Mapping, Optional, Tuple, Type, Union,
+    Any, Callable, Iterable, List, Mapping, Optional, Tuple, Text, Type, Union,
     TypeVar,
 )
 from types import FrameType, TracebackType
@@ -41,9 +41,9 @@ class ThreadError(Exception): ...
 
 
 class local(object):
-    def __getattribute__(self, name: str) -> Any: ...
-    def __setattr__(self, name: str, value: Any) -> None: ...
-    def __delattr__(self, name: str) -> None: ...
+    def __getattribute__(self, name: Text) -> Any: ...
+    def __setattr__(self, name: Text, value: Any) -> None: ...
+    def __delattr__(self, name: Text) -> None: ...
 
 
 class Thread: