beautifulsoup4: small fixes and updates (#7816)

stubs/beautifulsoup4/@tests/stubtest_allowlist.txt

@@ -1,56 +0,0 @@
-bs4.BeautifulSoup.handle_starttag
-bs4.PageElement.fetchNextSiblings
-bs4.PageElement.fetchPrevious
-bs4.PageElement.fetchPreviousSiblings
-bs4.PageElement.findAllNext
-bs4.PageElement.findAllPrevious
-bs4.PageElement.findNext
-bs4.PageElement.findNextSibling
-bs4.PageElement.findNextSiblings
-bs4.PageElement.findPrevious
-bs4.PageElement.findPreviousSibling
-bs4.PageElement.findPreviousSiblings
-bs4.PageElement.find_all_next
-bs4.PageElement.find_all_previous
-bs4.PageElement.find_next
-bs4.PageElement.find_next_sibling
-bs4.PageElement.find_next_siblings
-bs4.PageElement.find_previous
-bs4.PageElement.find_previous_sibling
-bs4.PageElement.find_previous_siblings
-bs4.SoupStrainer.__init__
-bs4.Tag.__init__
-bs4.Tag.find
-bs4.Tag.findAll
-bs4.Tag.findChild
-bs4.Tag.findChildren
-bs4.Tag.find_all
-bs4.dammit.chardet_type
-bs4.element.PY3K
-bs4.element.PageElement.fetchNextSiblings
-bs4.element.PageElement.fetchPrevious
-bs4.element.PageElement.fetchPreviousSiblings
-bs4.element.PageElement.findAllNext
-bs4.element.PageElement.findAllPrevious
-bs4.element.PageElement.findNext
-bs4.element.PageElement.findNextSibling
-bs4.element.PageElement.findNextSiblings
-bs4.element.PageElement.findPrevious
-bs4.element.PageElement.findPreviousSibling
-bs4.element.PageElement.findPreviousSiblings
-bs4.element.PageElement.find_all_next
-bs4.element.PageElement.find_all_previous
-bs4.element.PageElement.find_next
-bs4.element.PageElement.find_next_sibling
-bs4.element.PageElement.find_next_siblings
-bs4.element.PageElement.find_previous
-bs4.element.PageElement.find_previous_sibling
-bs4.element.PageElement.find_previous_siblings
-bs4.element.SoupStrainer.__init__
-bs4.element.Tag.__init__
-bs4.element.Tag.find
-bs4.element.Tag.findAll
-bs4.element.Tag.findChild
-bs4.element.Tag.findChildren
-bs4.element.Tag.find_all
-bs4.formatter.Formatter.__init__

stubs/beautifulsoup4/bs4/__init__.pyi

@@ -72,7 +72,16 @@ class BeautifulSoup(Tag):
     def pushTag(self, tag) -> None: ...
     def endData(self, containerClass: Any | None = ...) -> None: ...
     def object_was_parsed(self, o, parent: Any | None = ..., most_recent_element: Any | None = ...) -> None: ...
-    def handle_starttag(self, name, namespace, nsprefix, attrs, sourceline: Any | None = ..., sourcepos: Any | None = ...): ...
+    def handle_starttag(
+        self,
+        name,
+        namespace,
+        nsprefix,
+        attrs,
+        sourceline: Any | None = ...,
+        sourcepos: Any | None = ...,
+        namespaces: dict[str, str] | None = ...,
+    ): ...
     def handle_endtag(self, name, nsprefix: Any | None = ...) -> None: ...
     def handle_data(self, data) -> None: ...
     def decode(  # type: ignore[override]

stubs/beautifulsoup4/bs4/dammit.pyi

@@ -3,8 +3,6 @@ from logging import Logger
 from typing import Any
 from typing_extensions import Literal
 
-chardet_type: Any
-
 def chardet_dammit(s): ...
 
 xml_encoding: str

stubs/beautifulsoup4/bs4/element.pyi

@@ -8,7 +8,6 @@ from .builder import TreeBuilder
 from .formatter import Formatter, _EntitySubstitution
 
 DEFAULT_OUTPUT_ENCODING: str
-PY3K: bool
 nonwhitespace_re: Pattern[str]
 whitespace_re: Pattern[str]
 PYTHON_SPECIFIC_ENCODINGS: set[str]
@@ -73,7 +72,7 @@ class PageElement:
         self,
         name: _Strainable | SoupStrainer | None = ...,
         attrs: dict[str, _Strainable] | _Strainable = ...,
-        text: _Strainable | None = ...,
+        string: _Strainable | None = ...,
         **kwargs: _Strainable,
     ) -> Tag | NavigableString | None: ...
     findNext = find_next
@@ -81,7 +80,7 @@ class PageElement:
         self,
         name: _Strainable | SoupStrainer | None = ...,
         attrs: dict[str, _Strainable] | _Strainable = ...,
-        text: _Strainable | None = ...,
+        string: _Strainable | None = ...,
         limit: int | None = ...,
         **kwargs: _Strainable,
     ) -> ResultSet[PageElement]: ...
@@ -90,7 +89,7 @@ class PageElement:
         self,
         name: _Strainable | SoupStrainer | None = ...,
         attrs: dict[str, _Strainable] | _Strainable = ...,
-        text: _Strainable | None = ...,
+        string: _Strainable | None = ...,
         **kwargs: _Strainable,
     ) -> Tag | NavigableString | None: ...
     findNextSibling = find_next_sibling
@@ -98,7 +97,7 @@ class PageElement:
         self,
         name: _Strainable | SoupStrainer | None = ...,
         attrs: dict[str, _Strainable] | _Strainable = ...,
-        text: _Strainable | None = ...,
+        string: _Strainable | None = ...,
         limit: int | None = ...,
         **kwargs: _Strainable,
     ) -> ResultSet[PageElement]: ...
@@ -108,7 +107,7 @@ class PageElement:
         self,
         name: _Strainable | SoupStrainer | None = ...,
         attrs: dict[str, _Strainable] | _Strainable = ...,
-        text: _Strainable | None = ...,
+        string: _Strainable | None = ...,
         **kwargs: _Strainable,
     ) -> Tag | NavigableString | None: ...
     findPrevious = find_previous
@@ -116,7 +115,7 @@ class PageElement:
         self,
         name: _Strainable | SoupStrainer | None = ...,
         attrs: dict[str, _Strainable] | _Strainable = ...,
-        text: _Strainable | None = ...,
+        string: _Strainable | None = ...,
         limit: int | None = ...,
         **kwargs: _Strainable,
     ) -> ResultSet[PageElement]: ...
@@ -126,7 +125,7 @@ class PageElement:
         self,
         name: _Strainable | SoupStrainer | None = ...,
         attrs: dict[str, _Strainable] | _Strainable = ...,
-        text: _Strainable | None = ...,
+        string: _Strainable | None = ...,
         **kwargs: _Strainable,
     ) -> Tag | NavigableString | None: ...
     findPreviousSibling = find_previous_sibling
@@ -134,7 +133,7 @@ class PageElement:
         self,
         name: _Strainable | SoupStrainer | None = ...,
         attrs: dict[str, _Strainable] | _Strainable = ...,
-        text: _Strainable | None = ...,
+        string: _Strainable | None = ...,
         limit: int | None = ...,
         **kwargs: _Strainable,
     ) -> ResultSet[PageElement]: ...
@@ -257,6 +256,7 @@ class Tag(PageElement):
         cdata_list_attributes: list[str] | None = ...,
         preserve_whitespace_tags: list[str] | None = ...,
         interesting_string_types: type[NavigableString] | tuple[type[NavigableString], ...] | None = ...,
+        namespaces: dict[str, str] | None = ...,
     ) -> None: ...
     parserClass: type[BeautifulSoup] | None
     def __copy__(self: Self) -> Self: ...
@@ -308,7 +308,7 @@ class Tag(PageElement):
         name: _Strainable | None = ...,
         attrs: dict[str, _Strainable] | _Strainable = ...,
         recursive: bool = ...,
-        text: _Strainable | None = ...,
+        string: _Strainable | None = ...,
         **kwargs: _Strainable,
     ) -> Tag | NavigableString | None: ...
     findChild = find
@@ -317,7 +317,7 @@ class Tag(PageElement):
         name: _Strainable | None = ...,
         attrs: dict[str, _Strainable] | _Strainable = ...,
         recursive: bool = ...,
-        text: _Strainable | None = ...,
+        string: _Strainable | None = ...,
         limit: int | None = ...,
         **kwargs: _Strainable,
     ) -> ResultSet[Any]: ...
@@ -337,12 +337,12 @@ class Tag(PageElement):
 class SoupStrainer:
     name: _NormalizedStrainable
     attrs: dict[str, _NormalizedStrainable]
-    text: _NormalizedStrainable
+    string: _NormalizedStrainable
     def __init__(
         self,
         name: _Strainable | None = ...,
         attrs: dict[str, _Strainable] | _Strainable = ...,
-        text: _Strainable | None = ...,
+        string: _Strainable | None = ...,
         **kwargs: _Strainable,
     ) -> None: ...
     def search_tag(self, markup_name: Tag | str | None = ..., markup_attrs=...): ...

stubs/beautifulsoup4/bs4/formatter.pyi

@@ -21,6 +21,7 @@ class Formatter(EntitySubstitution):
         void_element_close_prefix: str = ...,
         cdata_containing_tags: list[str] | None = ...,
         empty_attributes_are_booleans: bool = ...,
+        indent: int = ...,
     ) -> None: ...
     def substitute(self, ns: str) -> str: ...
     def attribute_value(self, value: str) -> str: ...