Upgrade bleach stubs for 6.0 (#9583)

Co-authored-by: Alex Waygood <[email protected]>

Author: srittau

stubs/bleach/METADATA.toml

@@ -1 +1 @@
-version = "5.0.*"
+version = "6.0.*"

stubs/bleach/bleach/__init__.pyi

@@ -1,7 +1,9 @@
 from collections.abc import Container, Iterable
+from typing_extensions import TypeAlias
 
+from .callbacks import _Callback
 from .css_sanitizer import CSSSanitizer
-from .linkifier import DEFAULT_CALLBACKS as DEFAULT_CALLBACKS, Linker as Linker, _Callback
+from .linkifier import DEFAULT_CALLBACKS as DEFAULT_CALLBACKS, Linker as Linker
 from .sanitizer import (
     ALLOWED_ATTRIBUTES as ALLOWED_ATTRIBUTES,
     ALLOWED_PROTOCOLS as ALLOWED_PROTOCOLS,
@@ -15,11 +17,13 @@ __all__ = ["clean", "linkify"]
 __releasedate__: str
 __version__: str
 
+_HTMLAttrKey: TypeAlias = tuple[str | None, str]  # noqa: Y047
+
 def clean(
     text: str,
-    tags: Container[str] = ...,
+    tags: Iterable[str] = ...,
     attributes: _Attributes = ...,
-    protocols: Container[str] = ...,
+    protocols: Iterable[str] = ...,
     strip: bool = ...,
     strip_comments: bool = ...,
     css_sanitizer: CSSSanitizer | None = ...,

stubs/bleach/bleach/callbacks.pyi

@@ -1,8 +1,13 @@
 from collections.abc import MutableMapping
-from typing import Any
+from typing import Protocol
 from typing_extensions import TypeAlias
 
-_Attrs: TypeAlias = MutableMapping[Any, str]
+from bleach import _HTMLAttrKey
 
-def nofollow(attrs: _Attrs, new: bool = ...) -> _Attrs: ...
-def target_blank(attrs: _Attrs, new: bool = ...) -> _Attrs: ...
+_HTMLAttrs: TypeAlias = MutableMapping[_HTMLAttrKey, str]
+
+class _Callback(Protocol):  # noqa: Y046
+    def __call__(self, attrs: _HTMLAttrs, new: bool = ...) -> _HTMLAttrs: ...
+
+def nofollow(attrs: _HTMLAttrs, new: bool = ...) -> _HTMLAttrs: ...
+def target_blank(attrs: _HTMLAttrs, new: bool = ...) -> _HTMLAttrs: ...

stubs/bleach/bleach/linkifier.pyi

@@ -1,16 +1,10 @@
 from _typeshed import Incomplete
-from collections.abc import Container, Iterable, MutableMapping
+from collections.abc import Container, Iterable, Iterator
 from re import Pattern
-from typing import Any, Protocol
-from typing_extensions import TypeAlias
 
+from .callbacks import _Callback
 from .html5lib_shim import Filter
 
-_Attrs: TypeAlias = MutableMapping[Any, str]
-
-class _Callback(Protocol):
-    def __call__(self, attrs: _Attrs, new: bool = ...) -> _Attrs: ...
-
 DEFAULT_CALLBACKS: list[_Callback]
 
 TLDS: list[str]
@@ -28,21 +22,34 @@ class Linker:
     def __init__(
         self,
         callbacks: Iterable[_Callback] = ...,
-        skip_tags: Container[str] | None = ...,
-        parse_email: bool = ...,
+        skip_tags: Container[str] | None = None,
+        parse_email: bool = False,
         url_re: Pattern[str] = ...,
         email_re: Pattern[str] = ...,
         recognized_tags: Container[str] | None = ...,
     ) -> None: ...
     def linkify(self, text: str) -> str: ...
 
 class LinkifyFilter(Filter):
-    callbacks: Any
+    callbacks: Iterable[_Callback]
     skip_tags: Container[str]
     parse_email: bool
-    url_re: Any
-    email_re: Any
+    url_re: Pattern[str]
+    email_re: Pattern[str]
     def __init__(
-        self, source, callbacks=..., skip_tags: Container[str] | None = ..., parse_email: bool = ..., url_re=..., email_re=...
+        self,
+        source,
+        callbacks: Iterable[_Callback] | None = ...,
+        skip_tags: Container[str] | None = None,
+        parse_email: bool = False,
+        url_re: Pattern[str] = ...,
+        email_re: Pattern[str] = ...,
     ) -> None: ...
-    def __getattr__(self, item: str) -> Incomplete: ...
+    def apply_callbacks(self, attrs, is_new): ...
+    def extract_character_data(self, token_list): ...
+    def handle_email_addresses(self, src_iter): ...
+    def strip_non_url_bits(self, fragment): ...
+    def handle_links(self, src_iter): ...
+    def handle_a_tag(self, token_buffer): ...
+    def extract_entities(self, token): ...
+    def __iter__(self) -> Iterator[Incomplete]: ...

stubs/bleach/bleach/sanitizer.pyi

@@ -1,65 +1,78 @@
-from collections.abc import Callable, Container, Iterable
+from _typeshed import Incomplete
+from collections.abc import Callable, Iterable
 from re import Pattern
-from typing import Any
+from typing import Protocol
 from typing_extensions import TypeAlias
 
+from . import _HTMLAttrKey
 from .css_sanitizer import CSSSanitizer
 from .html5lib_shim import BleachHTMLParser, BleachHTMLSerializer, SanitizerFilter
 
-ALLOWED_TAGS: list[str]
+ALLOWED_TAGS: frozenset[str]
 ALLOWED_ATTRIBUTES: dict[str, list[str]]
-ALLOWED_PROTOCOLS: list[str]
+ALLOWED_PROTOCOLS: frozenset[str]
 
 INVISIBLE_CHARACTERS: str
 INVISIBLE_CHARACTERS_RE: Pattern[str]
 INVISIBLE_REPLACEMENT_CHAR: str
 
 # A html5lib Filter class
-_Filter: TypeAlias = Any
+class _Filter(Protocol):
+    def __call__(self, *, source: BleachSanitizerFilter) -> Incomplete: ...
+
+_AttributeFilter: TypeAlias = Callable[[str, str, str], bool]
+_AttributeDict: TypeAlias = dict[str, list[str] | _AttributeFilter] | dict[str, list[str]] | dict[str, _AttributeFilter]
+_Attributes: TypeAlias = _AttributeFilter | _AttributeDict | list[str]
+
+_TreeWalker: TypeAlias = Callable[[Incomplete], Incomplete]
 
 class Cleaner:
-    tags: Container[str]
+    tags: Iterable[str]
     attributes: _Attributes
-    protocols: Container[str]
+    protocols: Iterable[str]
     strip: bool
     strip_comments: bool
     filters: Iterable[_Filter]
     css_sanitizer: CSSSanitizer | None
     parser: BleachHTMLParser
-    walker: Any
+    walker: _TreeWalker
     serializer: BleachHTMLSerializer
     def __init__(
         self,
-        tags: Container[str] = ...,
+        tags: Iterable[str] = ...,
         attributes: _Attributes = ...,
-        protocols: Container[str] = ...,
+        protocols: Iterable[str] = ...,
         strip: bool = ...,
         strip_comments: bool = ...,
         filters: Iterable[_Filter] | None = ...,
         css_sanitizer: CSSSanitizer | None = ...,
     ) -> None: ...
     def clean(self, text: str) -> str: ...
 
-_AttributeFilter: TypeAlias = Callable[[str, str, str], bool]
-_AttributeDict: TypeAlias = dict[str, list[str] | _AttributeFilter] | dict[str, list[str]] | dict[str, _AttributeFilter]
-_Attributes: TypeAlias = _AttributeFilter | _AttributeDict | list[str]
-
 def attribute_filter_factory(attributes: _Attributes) -> _AttributeFilter: ...
 
 class BleachSanitizerFilter(SanitizerFilter):
+    allowed_tags: frozenset[str]
+    allowed_protocols: frozenset[str]
     attr_filter: _AttributeFilter
-    strip_disallowed_elements: bool
+    strip_disallowed_tags: bool
     strip_html_comments: bool
+    attr_val_is_uri: frozenset[_HTMLAttrKey]
+    svg_attr_val_allows_ref: frozenset[_HTMLAttrKey]
+    svg_allow_local_href: frozenset[_HTMLAttrKey]
+    css_sanitizer: CSSSanitizer | None
     def __init__(
         self,
         source,
-        allowed_elements: Container[str] = ...,
+        allowed_tags: Iterable[str] = ...,
         attributes: _Attributes = ...,
-        allowed_protocols: Container[str] = ...,
-        strip_disallowed_elements: bool = ...,
-        strip_html_comments: bool = ...,
-        css_sanitizer: CSSSanitizer | None = ...,
-        **kwargs,
+        allowed_protocols: Iterable[str] = ...,
+        attr_val_is_uri: frozenset[_HTMLAttrKey] = ...,
+        svg_attr_val_allows_ref: frozenset[_HTMLAttrKey] = ...,
+        svg_allow_local_href: frozenset[_HTMLAttrKey] = ...,
+        strip_disallowed_tags: bool = False,
+        strip_html_comments: bool = True,
+        css_sanitizer: CSSSanitizer | None = None,
     ) -> None: ...
     def sanitize_stream(self, token_iterator): ...
     def merge_characters(self, token_iterator): ...