[html5lib] Improve serializer (#15036)

stubs/html5lib/@tests/stubtest_allowlist.txt

@@ -0,0 +1,6 @@
+# Side effects from module initialization:
+html5lib.serializer.k
+html5lib.serializer.v
+html5lib.treeadapters.sax.prefix
+html5lib.treeadapters.sax.localName
+html5lib.treeadapters.sax.namespace

stubs/html5lib/html5lib/serializer.pyi

@@ -1,20 +1,53 @@
 from _typeshed import Incomplete
 from collections.abc import Generator
-from typing import overload
+from typing import Literal, overload
 
-k: str
-v: str | int
-
-def htmlentityreplace_errors(exc: Exception) -> tuple[str | bytes, int]: ...
+def htmlentityreplace_errors(exc: UnicodeError) -> tuple[str | bytes, int]: ...
 @overload
-def serialize(input, tree: str = "etree", encoding: None = None, **serializer_opts) -> str: ...
+def serialize(
+    input,
+    tree: Literal["dom", "genshi", "lxml", "etree"] = "etree",
+    encoding: Literal[""] | None = None,
+    *,
+    quote_attr_values: Literal["legacy", "spec", "always"] = "legacy",
+    quote_char: str = '"',
+    use_best_quote_char: bool = ...,  # default value depends on whether quote_char was passed
+    omit_optional_tags: bool = True,
+    minimize_boolean_attributes: bool = True,
+    use_trailing_solidus: bool = False,
+    space_before_trailing_solidus: bool = True,
+    escape_lt_in_attrs: bool = False,
+    escape_rcdata: bool = False,
+    resolve_entities: bool = True,
+    alphabetical_attributes: bool = False,
+    inject_meta_charset: bool = True,
+    strip_whitespace: bool = False,
+    sanitize: bool = False,
+) -> str: ...
 @overload
-def serialize(input, tree: str, encoding: str, **serializer_opts) -> bytes: ...
-@overload
-def serialize(input, *, encoding: str, **serializer_opts) -> bytes: ...
+def serialize(
+    input,
+    tree: Literal["dom", "genshi", "lxml", "etree"] = "etree",
+    encoding: str = ...,
+    *,
+    quote_attr_values: Literal["legacy", "spec", "always"] = "legacy",
+    quote_char: str = '"',
+    use_best_quote_char: bool = ...,  # default value depends on whether quote_char was passed
+    omit_optional_tags: bool = True,
+    minimize_boolean_attributes: bool = True,
+    use_trailing_solidus: bool = False,
+    space_before_trailing_solidus: bool = True,
+    escape_lt_in_attrs: bool = False,
+    escape_rcdata: bool = False,
+    resolve_entities: bool = True,
+    alphabetical_attributes: bool = False,
+    inject_meta_charset: bool = True,
+    strip_whitespace: bool = False,
+    sanitize: bool = False,
+) -> bytes: ...
 
 class HTMLSerializer:
-    quote_attr_values: str
+    quote_attr_values: Literal["legacy", "spec", "always"]
     quote_char: str
     use_best_quote_char: bool
     omit_optional_tags: bool
@@ -28,15 +61,38 @@ class HTMLSerializer:
     inject_meta_charset: bool
     strip_whitespace: bool
     sanitize: bool
-    options: Incomplete
-    errors: Incomplete
+    options: tuple[str, ...]
+    errors: list[Incomplete]
     strict: bool
-    def __init__(self, **kwargs) -> None: ...
-    def encode(self, string): ...
-    def encodeStrict(self, string): ...
-    encoding: Incomplete
-    def serialize(self, treewalker, encoding=None) -> Generator[Incomplete]: ...
-    def render(self, treewalker, encoding=None): ...
-    def serializeError(self, data: str = "XXX ERROR MESSAGE NEEDED") -> None: ...
+    def __init__(
+        self,
+        *,
+        quote_attr_values: Literal["legacy", "spec", "always"] = "legacy",
+        quote_char: str = '"',
+        use_best_quote_char: bool = ...,  # default value depends on whether quote_char was passed
+        omit_optional_tags: bool = True,
+        minimize_boolean_attributes: bool = True,
+        use_trailing_solidus: bool = False,
+        space_before_trailing_solidus: bool = True,
+        escape_lt_in_attrs: bool = False,
+        escape_rcdata: bool = False,
+        resolve_entities: bool = True,
+        alphabetical_attributes: bool = False,
+        inject_meta_charset: bool = True,
+        strip_whitespace: bool = False,
+        sanitize: bool = False,
+    ) -> None: ...
+    def encode(self, string: str) -> str | bytes: ...  # result depends on self.encoding
+    def encodeStrict(self, string: str) -> str | bytes: ...  # result depends on self.encoding
+    encoding: str | None
+    @overload
+    def serialize(self, treewalker, encoding: Literal[""] | None = None) -> Generator[str]: ...
+    @overload
+    def serialize(self, treewalker, encoding: str = ...) -> Generator[bytes]: ...
+    @overload
+    def render(self, treewalker, encoding: Literal[""] | None = None) -> str: ...
+    @overload
+    def render(self, treewalker, encoding: str = ...) -> bytes: ...
+    def serializeError(self, data="XXX ERROR MESSAGE NEEDED") -> None: ...
 
 class SerializeError(Exception): ...

stubs/html5lib/html5lib/treeadapters/sax.pyi

@@ -1,6 +1,3 @@
-prefix: str | None
-localName: str
-namespace: str
 prefix_mapping: dict[str, str]
 
 def to_sax(walker, handler) -> None: ...