xml: improve bytes types (#9110)

Co-authored-by: Shantanu <[email protected]>

Author: JelleZijlstra

stdlib/xml/dom/expatbuilder.pyi

@@ -1,4 +1,4 @@
-from _typeshed import Incomplete
+from _typeshed import Incomplete, ReadableBuffer, SupportsRead
 from typing import Any, NoReturn
 from xml.dom.minidom import Document, DOMImplementation, Node, TypeInfo
 from xml.dom.xmlbuilder import DOMBuilderFilter, Options
@@ -30,8 +30,8 @@ class ExpatBuilder:
     def getParser(self): ...
     def reset(self) -> None: ...
     def install(self, parser) -> None: ...
-    def parseFile(self, file) -> Document: ...
-    def parseString(self, string: str) -> Document: ...
+    def parseFile(self, file: SupportsRead[ReadableBuffer | str]) -> Document: ...
+    def parseString(self, string: str | ReadableBuffer) -> Document: ...
     def start_doctype_decl_handler(self, doctypeName, systemId, publicId, has_internal_subset) -> None: ...
     def end_doctype_decl_handler(self) -> None: ...
     def pi_handler(self, target, data) -> None: ...
@@ -87,14 +87,14 @@ class ParseEscape(Exception): ...
 class InternalSubsetExtractor(ExpatBuilder):
     subset: Any | None
     def getSubset(self) -> Any | None: ...
-    def parseFile(self, file) -> None: ...  # type: ignore[override]
-    def parseString(self, string: str) -> None: ...  # type: ignore[override]
+    def parseFile(self, file: SupportsRead[ReadableBuffer | str]) -> None: ...  # type: ignore[override]
+    def parseString(self, string: str | ReadableBuffer) -> None: ...  # type: ignore[override]
     def start_doctype_decl_handler(self, name, publicId, systemId, has_internal_subset) -> None: ...  # type: ignore[override]
     def end_doctype_decl_handler(self) -> NoReturn: ...
     def start_element_handler(self, name, attrs) -> NoReturn: ...
 
-def parse(file, namespaces: bool = ...): ...
-def parseString(string: str, namespaces: bool = ...): ...
+def parse(file: str | SupportsRead[ReadableBuffer | str], namespaces: bool = ...): ...
+def parseString(string: str | ReadableBuffer, namespaces: bool = ...): ...
 def parseFragment(file, context, namespaces: bool = ...): ...
 def parseFragmentString(string: str, context, namespaces: bool = ...): ...
 def makeBuilder(options: Options) -> ExpatBuilderNS | ExpatBuilder: ...

stdlib/xml/dom/minidom.pyi

@@ -1,12 +1,12 @@
 import sys
 import xml.dom
-from _typeshed import Incomplete, Self, SupportsRead, SupportsWrite
+from _typeshed import Incomplete, ReadableBuffer, Self, SupportsRead, SupportsWrite
 from typing_extensions import Literal
 from xml.dom.xmlbuilder import DocumentLS, DOMImplementationLS
 from xml.sax.xmlreader import XMLReader
 
-def parse(file: str | SupportsRead[bytes] | SupportsRead[str], parser: XMLReader | None = ..., bufsize: int | None = ...): ...
-def parseString(string: str | bytes, parser: XMLReader | None = ...): ...
+def parse(file: str | SupportsRead[ReadableBuffer | str], parser: XMLReader | None = ..., bufsize: int | None = ...): ...
+def parseString(string: str | ReadableBuffer, parser: XMLReader | None = ...): ...
 def getDOMImplementation(features=...) -> DOMImplementation | None: ...
 
 class Node(xml.dom.Node):

stdlib/xml/etree/ElementInclude.pyi

@@ -1,4 +1,5 @@
 import sys
+from _typeshed import StrOrBytesPath
 from collections.abc import Callable
 from xml.etree.ElementTree import Element
 
@@ -11,7 +12,7 @@ if sys.version_info >= (3, 9):
 
 class FatalIncludeError(SyntaxError): ...
 
-def default_loader(href: str | bytes | int, parse: str, encoding: str | None = ...) -> str | Element: ...
+def default_loader(href: StrOrBytesPath | int, parse: str, encoding: str | None = ...) -> str | Element: ...
 
 # TODO: loader is of type default_loader ie it takes a callable that has the
 # same signature as default_loader. But default_loader has a keyword argument

stdlib/xml/etree/ElementTree.pyi

@@ -1,6 +1,6 @@
 import sys
 from _collections_abc import dict_keys
-from _typeshed import FileDescriptor, StrOrBytesPath, SupportsRead, SupportsWrite
+from _typeshed import FileDescriptor, ReadableBuffer, StrOrBytesPath, SupportsRead, SupportsWrite
 from collections.abc import Callable, Generator, ItemsView, Iterable, Iterator, Mapping, Sequence
 from typing import Any, TypeVar, overload
 from typing_extensions import Literal, SupportsIndex, TypeAlias, TypeGuard
@@ -54,7 +54,7 @@ def iselement(element: object) -> TypeGuard[Element]: ...
 if sys.version_info >= (3, 8):
     @overload
     def canonicalize(
-        xml_data: str | bytes | None = ...,
+        xml_data: str | ReadableBuffer | None = ...,
         *,
         out: None = ...,
         from_file: _FileRead | None = ...,
@@ -68,7 +68,7 @@ if sys.version_info >= (3, 8):
     ) -> str: ...
     @overload
     def canonicalize(
-        xml_data: str | bytes | None = ...,
+        xml_data: str | ReadableBuffer | None = ...,
         *,
         out: SupportsWrite[str],
         from_file: _FileRead | None = ...,
@@ -270,19 +270,19 @@ def iterparse(
 
 class XMLPullParser:
     def __init__(self, events: Sequence[str] | None = ..., *, _parser: XMLParser | None = ...) -> None: ...
-    def feed(self, data: str | bytes) -> None: ...
+    def feed(self, data: str | ReadableBuffer) -> None: ...
     def close(self) -> None: ...
     # Second element in the tuple could be `Element`, `tuple[str, str]` or `None`.
     # Use `Any` to avoid false-positive errors.
     def read_events(self) -> Iterator[tuple[str, Any]]: ...
 
-def XML(text: str | bytes, parser: XMLParser | None = ...) -> Element: ...
-def XMLID(text: str | bytes, parser: XMLParser | None = ...) -> tuple[Element, dict[str, Element]]: ...
+def XML(text: str | ReadableBuffer, parser: XMLParser | None = ...) -> Element: ...
+def XMLID(text: str | ReadableBuffer, parser: XMLParser | None = ...) -> tuple[Element, dict[str, Element]]: ...
 
 # This is aliased to XML in the source.
 fromstring = XML
 
-def fromstringlist(sequence: Sequence[str | bytes], parser: XMLParser | None = ...) -> Element: ...
+def fromstringlist(sequence: Sequence[str | ReadableBuffer], parser: XMLParser | None = ...) -> Element: ...
 
 # This type is both not precise enough and too precise. The TreeBuilder
 # requires the elementfactory to accept tag and attrs in its args and produce
@@ -313,9 +313,11 @@ class TreeBuilder:
         def __init__(self, element_factory: _ElementFactory | None = ...) -> None: ...
 
     def close(self) -> Element: ...
-    def data(self, __data: str | bytes) -> None: ...
-    def start(self, __tag: str | bytes, __attrs: dict[str | bytes, str | bytes]) -> Element: ...
-    def end(self, __tag: str | bytes) -> Element: ...
+    def data(self, __data: str) -> None: ...
+    # tag and attrs are passed to the element_factory, so they could be anything
+    # depending on what the particular factory supports.
+    def start(self, __tag: Any, __attrs: dict[Any, Any]) -> Element: ...
+    def end(self, __tag: str) -> Element: ...
     if sys.version_info >= (3, 8):
         # These two methods have pos-only parameters in the C implementation
         def comment(self, __text: str | None) -> Element: ...
@@ -355,4 +357,4 @@ class XMLParser:
         def doctype(self, __name: str, __pubid: str, __system: str) -> None: ...
 
     def close(self) -> Any: ...
-    def feed(self, __data: str | bytes) -> None: ...
+    def feed(self, __data: str | ReadableBuffer) -> None: ...

stdlib/xml/sax/__init__.pyi

@@ -1,5 +1,5 @@
 import sys
-from _typeshed import SupportsRead, _T_co
+from _typeshed import ReadableBuffer, SupportsRead, _T_co
 from collections.abc import Iterable
 from typing import Any, NoReturn, Protocol
 from xml.sax.handler import ContentHandler as ContentHandler, ErrorHandler as ErrorHandler
@@ -36,5 +36,5 @@ else:
 def parse(
     source: str | _SupportsReadClose[bytes] | _SupportsReadClose[str], handler: ContentHandler, errorHandler: ErrorHandler = ...
 ) -> None: ...
-def parseString(string: bytes | str, handler: ContentHandler, errorHandler: ErrorHandler | None = ...) -> None: ...
+def parseString(string: ReadableBuffer | str, handler: ContentHandler, errorHandler: ErrorHandler | None = ...) -> None: ...
 def _create_parser(parser_name: str) -> XMLReader: ...