Upgrade PyYAML to version 6 (#6183)

* Fix load_all() argument
* Remove Python 2 remnants
  + Replace Text with str
  + Replace alias "_Str" with str
  + Import from collections.abc and re where applicable
  + Remove Python 2 branches
* Fix PyYaml allowlist entries
  + Add yaml._yaml and move CParser and CEmitter there.
  + Add missing functions, classes, and arguments.
* Use relative imports in some modules.
* Add __all__ to yaml.cyaml.
* Remove unnecessary noqa markers.

Author: srittau

stubs/PyYAML/@tests/stubtest_allowlist.txt

@@ -1,2 +1 @@
-version = "5.4.*"
-python2 = true
+version = "6.0.*"

stubs/PyYAML/METADATA.toml

@@ -1,24 +1,19 @@
-import sys
-from typing import IO, Any, Callable, Iterable, Iterator, Pattern, Sequence, Text, Type, TypeVar, Union, overload
-
-from yaml.constructor import BaseConstructor
-from yaml.dumper import *  # noqa: F403
-from yaml.error import *  # noqa: F403
-from yaml.events import *  # noqa: F403
-from yaml.loader import *  # noqa: F403
-from yaml.nodes import *  # noqa: F403
-from yaml.representer import BaseRepresenter
-from yaml.resolver import BaseResolver
-from yaml.tokens import *  # noqa: F403
+from collections.abc import Callable, Iterable, Iterator, Sequence
+from typing import IO, Any, Pattern, Type, TypeVar, overload
 
 from . import resolver as resolver  # Help mypy a bit; this is implied by loader and dumper
+from .constructor import BaseConstructor
 from .cyaml import *
+from .dumper import *
+from .error import *
+from .events import *
+from .loader import *
+from .nodes import *
+from .representer import BaseRepresenter
+from .resolver import BaseResolver
+from .tokens import *
 
-if sys.version_info >= (3, 0):
-    _Str = str
-else:
-    _Str = Union[Text, str]
-# FIXME: the functions really return py2:unicode/py3:str if encoding is None, otherwise py2:str/py3:bytes. Waiting for python/mypy#5621
+# FIXME: the functions really return str if encoding is None, otherwise bytes. Waiting for python/mypy#5621
 _Yaml = Any
 
 __with_libyaml__: Any
@@ -28,18 +23,19 @@ _T = TypeVar("_T")
 _Constructor = TypeVar("_Constructor", bound=BaseConstructor)
 _Representer = TypeVar("_Representer", bound=BaseRepresenter)
 
+def warnings(settings=...): ...
 def scan(stream, Loader=...): ...
 def parse(stream, Loader=...): ...
 def compose(stream, Loader=...): ...
 def compose_all(stream, Loader=...): ...
-def load(stream: bytes | IO[bytes] | Text | IO[Text], Loader=...) -> Any: ...
-def load_all(stream: bytes | IO[bytes] | Text | IO[Text], Loader=...) -> Iterator[Any]: ...
-def full_load(stream: bytes | IO[bytes] | Text | IO[Text]) -> Any: ...
-def full_load_all(stream: bytes | IO[bytes] | Text | IO[Text]) -> Iterator[Any]: ...
-def safe_load(stream: bytes | IO[bytes] | Text | IO[Text]) -> Any: ...
-def safe_load_all(stream: bytes | IO[bytes] | Text | IO[Text]) -> Iterator[Any]: ...
-def unsafe_load(stream: bytes | IO[bytes] | Text | IO[Text]) -> Any: ...
-def unsafe_load_all(stream: bytes | IO[bytes] | Text | IO[Text]) -> Iterator[Any]: ...
+def load(stream: bytes | IO[bytes] | str | IO[str], Loader) -> Any: ...
+def load_all(stream: bytes | IO[bytes] | str | IO[str], Loader) -> Iterator[Any]: ...
+def full_load(stream: bytes | IO[bytes] | str | IO[str]) -> Any: ...
+def full_load_all(stream: bytes | IO[bytes] | str | IO[str]) -> Iterator[Any]: ...
+def safe_load(stream: bytes | IO[bytes] | str | IO[str]) -> Any: ...
+def safe_load_all(stream: bytes | IO[bytes] | str | IO[str]) -> Iterator[Any]: ...
+def unsafe_load(stream: bytes | IO[bytes] | str | IO[str]) -> Any: ...
+def unsafe_load_all(stream: bytes | IO[bytes] | str | IO[str]) -> Iterator[Any]: ...
 def emit(events, stream=..., Dumper=..., canonical=..., indent=..., width=..., allow_unicode=..., line_break=...): ...
 @overload
 def serialize_all(
@@ -67,7 +63,7 @@ def serialize_all(
     width=...,
     allow_unicode=...,
     line_break=...,
-    encoding: _Str | None = ...,
+    encoding: str | None = ...,
     explicit_start=...,
     explicit_end=...,
     version=...,
@@ -101,7 +97,7 @@ def serialize(
     width=...,
     allow_unicode=...,
     line_break=...,
-    encoding: _Str | None = ...,
+    encoding: str | None = ...,
     explicit_start=...,
     explicit_end=...,
     version=...,
@@ -138,7 +134,7 @@ def dump_all(
     width=...,
     allow_unicode=...,
     line_break=...,
-    encoding: _Str | None = ...,
+    encoding: str | None = ...,
     explicit_start=...,
     explicit_end=...,
     version=...,
@@ -178,7 +174,7 @@ def dump(
     width=...,
     allow_unicode=...,
     line_break=...,
-    encoding: _Str | None = ...,
+    encoding: str | None = ...,
     explicit_start=...,
     explicit_end=...,
     version=...,
@@ -216,7 +212,7 @@ def safe_dump_all(
     width=...,
     allow_unicode=...,
     line_break=...,
-    encoding: _Str | None = ...,
+    encoding: str | None = ...,
     explicit_start=...,
     explicit_end=...,
     version=...,
@@ -254,40 +250,40 @@ def safe_dump(
     width=...,
     allow_unicode=...,
     line_break=...,
-    encoding: _Str | None = ...,
+    encoding: str | None = ...,
     explicit_start=...,
     explicit_end=...,
     version=...,
     tags=...,
     sort_keys: bool = ...,
 ) -> _Yaml: ...
 def add_implicit_resolver(
-    tag: _Str,
+    tag: str,
     regexp: Pattern[str],
     first: Iterable[Any] | None = ...,
     Loader: Type[BaseResolver] | None = ...,
     Dumper: Type[BaseResolver] = ...,
 ) -> None: ...
 def add_path_resolver(
-    tag: _Str,
+    tag: str,
     path: Iterable[Any],
     kind: Type[Any] | None = ...,
     Loader: Type[BaseResolver] | None = ...,
     Dumper: Type[BaseResolver] = ...,
 ) -> None: ...
 @overload
 def add_constructor(
-    tag: _Str, constructor: Callable[[Loader | FullLoader | UnsafeLoader, Node], Any], Loader: None = ...
+    tag: str, constructor: Callable[[Loader | FullLoader | UnsafeLoader, Node], Any], Loader: None = ...
 ) -> None: ...
 @overload
-def add_constructor(tag: _Str, constructor: Callable[[_Constructor, Node], Any], Loader: Type[_Constructor]) -> None: ...
+def add_constructor(tag: str, constructor: Callable[[_Constructor, Node], Any], Loader: Type[_Constructor]) -> None: ...
 @overload
 def add_multi_constructor(
-    tag_prefix: _Str, multi_constructor: Callable[[Loader | FullLoader | UnsafeLoader, _Str, Node], Any], Loader: None = ...
+    tag_prefix: str, multi_constructor: Callable[[Loader | FullLoader | UnsafeLoader, str, Node], Any], Loader: None = ...
 ) -> None: ...
 @overload
 def add_multi_constructor(
-    tag_prefix: _Str, multi_constructor: Callable[[_Constructor, _Str, Node], Any], Loader: Type[_Constructor]
+    tag_prefix: str, multi_constructor: Callable[[_Constructor, str, Node], Any], Loader: Type[_Constructor]
 ) -> None: ...
 @overload
 def add_representer(data_type: Type[_T], representer: Callable[[Dumper, _T], Node]) -> None: ...
@@ -301,7 +297,7 @@ def add_multi_representer(
 ) -> None: ...
 
 class YAMLObjectMetaclass(type):
-    def __init__(self, name, bases, kwds) -> None: ...
+    def __init__(cls, name, bases, kwds) -> None: ...
 
 class YAMLObject(metaclass=YAMLObjectMetaclass):
     yaml_loader: Any

stubs/PyYAML/yaml/__init__.pyi

@@ -0,0 +1,56 @@
+from _typeshed import SupportsRead
+from collections.abc import Mapping, Sequence
+from typing import IO, Any
+
+from .events import Event
+from .nodes import Node
+from .tokens import Token
+
+def get_version_string() -> str: ...
+def get_version() -> tuple[int, int, int]: ...
+
+class Mark:
+    name: Any
+    index: int
+    line: int
+    column: int
+    buffer: Any
+    pointer: Any
+    def __init__(self, name, index: int, line: int, column: int, buffer, pointer) -> None: ...
+    def get_snippet(self): ...
+
+class CParser:
+    def __init__(self, stream: str | bytes | SupportsRead[str | bytes]) -> None: ...
+    def dispose(self) -> None: ...
+    def get_token(self) -> Token | None: ...
+    def peek_token(self) -> Token | None: ...
+    def check_token(self, *choices) -> bool: ...
+    def get_event(self) -> Event | None: ...
+    def peek_event(self) -> Event | None: ...
+    def check_event(self, *choices) -> bool: ...
+    def check_node(self) -> bool: ...
+    def get_node(self) -> Node | None: ...
+    def get_single_node(self) -> Node | None: ...
+    def raw_parse(self) -> int: ...
+    def raw_scan(self) -> int: ...
+
+class CEmitter:
+    def __init__(
+        self,
+        stream: IO[Any],
+        canonical: Any | None = ...,
+        indent: int | None = ...,
+        width: int | None = ...,
+        allow_unicode: Any | None = ...,
+        line_break: str | None = ...,
+        encoding: str | None = ...,
+        explicit_start: Any | None = ...,
+        explicit_end: Any | None = ...,
+        version: Sequence[int] | None = ...,
+        tags: Mapping[str, str] | None = ...,
+    ) -> None: ...
+    def dispose(self) -> None: ...
+    def emit(self, event_object) -> None: ...
+    def open(self) -> None: ...
+    def close(self) -> None: ...
+    def serialize(self, node) -> None: ...

stubs/PyYAML/yaml/_yaml.pyi

@@ -1,10 +1,9 @@
-import sys
-from typing import Any, Text, Union
+from typing import Any, Pattern, Union
 
 from yaml.error import MarkedYAMLError
 from yaml.nodes import ScalarNode
 
-_Scalar = Union[Text, int, float, bool, None]
+_Scalar = Union[str, int, float, bool, None]
 
 class ConstructorError(MarkedYAMLError): ...
 
@@ -17,6 +16,7 @@ class BaseConstructor:
     deep_construct: Any
     def __init__(self) -> None: ...
     def check_data(self): ...
+    def check_state_key(self, key: str) -> None: ...
     def get_data(self): ...
     def get_single_data(self) -> Any: ...
     def construct_document(self, node): ...
@@ -54,6 +54,8 @@ class SafeConstructor(BaseConstructor):
     def construct_undefined(self, node): ...
 
 class FullConstructor(SafeConstructor):
+    def get_state_keys_blacklist(self) -> list[str]: ...
+    def get_state_keys_blacklist_regexp(self) -> Pattern[str]: ...
     def construct_python_str(self, node): ...
     def construct_python_unicode(self, node): ...
     def construct_python_bytes(self, node): ...
@@ -65,7 +67,7 @@ class FullConstructor(SafeConstructor):
     def construct_python_name(self, suffix, node): ...
     def construct_python_module(self, suffix, node): ...
     def make_python_instance(self, suffix, node, args=..., kwds=..., newobj=..., unsafe=...): ...
-    def set_python_instance_state(self, instance, state): ...
+    def set_python_instance_state(self, instance, state, unsafe: bool = ...) -> None: ...
     def construct_python_object(self, suffix, node): ...
     def construct_python_object_apply(self, suffix, node, newobj=...): ...
     def construct_python_object_new(self, suffix, node): ...
@@ -86,8 +88,6 @@ class Constructor(SafeConstructor):
     def find_python_name(self, name, mark): ...
     def construct_python_name(self, suffix, node): ...
     def construct_python_module(self, suffix, node): ...
-    if sys.version_info < (3, 0):
-        class classobj: ...
     def make_python_instance(self, suffix, node, args=..., kwds=..., newobj=...): ...
     def set_python_instance_state(self, instance, state): ...
     def construct_python_object(self, suffix, node): ...

stubs/PyYAML/yaml/constructor.pyi

@@ -1,27 +1,15 @@
 from _typeshed import SupportsRead
-from typing import IO, Any, Mapping, Sequence, Text, Union
+from collections.abc import Mapping, Sequence
+from typing import IO, Any, Union
 
-from yaml.constructor import BaseConstructor, FullConstructor, SafeConstructor, UnsafeConstructor
-from yaml.events import Event
-from yaml.nodes import Node
-from yaml.representer import BaseRepresenter, SafeRepresenter
-from yaml.resolver import BaseResolver, Resolver
-from yaml.tokens import Token
+from ._yaml import CEmitter, CParser
+from .constructor import BaseConstructor, FullConstructor, SafeConstructor, UnsafeConstructor
+from .representer import BaseRepresenter, SafeRepresenter
+from .resolver import BaseResolver, Resolver
 
-_Readable = SupportsRead[Union[Text, bytes]]
+__all__ = ["CBaseLoader", "CSafeLoader", "CFullLoader", "CUnsafeLoader", "CLoader", "CBaseDumper", "CSafeDumper", "CDumper"]
 
-class CParser:
-    def __init__(self, stream: str | bytes | _Readable) -> None: ...
-    def dispose(self) -> None: ...
-    def get_token(self) -> Token | None: ...
-    def peek_token(self) -> Token | None: ...
-    def check_token(self, *choices) -> bool: ...
-    def get_event(self) -> Event | None: ...
-    def peek_event(self) -> Event | None: ...
-    def check_event(self, *choices) -> bool: ...
-    def check_node(self) -> bool: ...
-    def get_node(self) -> Node | None: ...
-    def get_single_node(self) -> Node | None: ...
+_Readable = SupportsRead[Union[str, bytes]]
 
 class CBaseLoader(CParser, BaseConstructor, BaseResolver):
     def __init__(self, stream: str | bytes | _Readable) -> None: ...
@@ -38,40 +26,42 @@ class CFullLoader(CParser, FullConstructor, Resolver):
 class CUnsafeLoader(CParser, UnsafeConstructor, Resolver):
     def __init__(self, stream: str | bytes | _Readable) -> None: ...
 
-class CEmitter(object):
+class CBaseDumper(CEmitter, BaseRepresenter, BaseResolver):
     def __init__(
         self,
         stream: IO[Any],
+        default_style: str | None = ...,
+        default_flow_style: bool | None = ...,
         canonical: Any | None = ...,
         indent: int | None = ...,
         width: int | None = ...,
         allow_unicode: Any | None = ...,
         line_break: str | None = ...,
-        encoding: Text | None = ...,
+        encoding: str | None = ...,
         explicit_start: Any | None = ...,
         explicit_end: Any | None = ...,
         version: Sequence[int] | None = ...,
-        tags: Mapping[Text, Text] | None = ...,
+        tags: Mapping[str, str] | None = ...,
+        sort_keys: bool = ...,
     ) -> None: ...
 
-class CBaseDumper(CEmitter, BaseRepresenter, BaseResolver):
+class CDumper(CEmitter, SafeRepresenter, Resolver):
     def __init__(
         self,
         stream: IO[Any],
         default_style: str | None = ...,
-        default_flow_style: bool | None = ...,
+        default_flow_style: bool = ...,
         canonical: Any | None = ...,
         indent: int | None = ...,
         width: int | None = ...,
         allow_unicode: Any | None = ...,
         line_break: str | None = ...,
-        encoding: Text | None = ...,
+        encoding: str | None = ...,
         explicit_start: Any | None = ...,
         explicit_end: Any | None = ...,
         version: Sequence[int] | None = ...,
-        tags: Mapping[Text, Text] | None = ...,
+        tags: Mapping[str, str] | None = ...,
+        sort_keys: bool = ...,
     ) -> None: ...
 
-class CDumper(CEmitter, SafeRepresenter, Resolver): ...
-
 CSafeDumper = CDumper