Fix #10214: html: invalid language tag was generated for zh_CN

So far, HTML builder output the `language` configuration as a language
tag for HTML.  But it takes locale string in ANSI C, not IETF language
code.

This converts locale string to language tag to generate valid language
tag for HTML.

CHANGES

@@ -67,6 +67,8 @@ Bugs fixed
 
 * #10279: autodoc: Default values for keyword only arguments in overloaded
   functions are rendered as a string literal
+* #10214: html: invalid language tag was generated if :confval:`language`
+  contains a country code (ex. zh_CN)
 * #10236: html search: objects are duplicated in search result
 * #9962: texinfo: Deprecation message for ``@definfoenclose`` command on
   bulding texinfo document

sphinx/builders/html/__init__.py

@@ -7,7 +7,7 @@ import re
 import sys
 from datetime import datetime
 from os import path
-from typing import IO, Any, Dict, Iterable, Iterator, List, Set, Tuple, Type
+from typing import IO, Any, Dict, Iterable, Iterator, List, Optional, Set, Tuple, Type
 from urllib.parse import quote
 
 from docutils import nodes
@@ -68,6 +68,17 @@ def get_stable_hash(obj: Any) -> str:
     return md5(str(obj).encode()).hexdigest()
 
 
+def convert_locale_to_language_tag(locale: Optional[str]) -> Optional[str]:
+    """Convert a locale string to a language tag (ex. en_US -> en-US).
+
+    refs: BCP 47 (:rfc:`5646`)
+    """
+    if locale:
+        return locale.replace('_', '-')
+    else:
+        return None
+
+
 class Stylesheet(str):
     """A metadata of stylesheet.
 
@@ -510,7 +521,7 @@ class StandaloneHTMLBuilder(Builder):
             'file_suffix': self.out_suffix,
             'link_suffix': self.link_suffix,
             'script_files': self.script_files,
-            'language': self.config.language,
+            'language': convert_locale_to_language_tag(self.config.language),
             'css_files': self.css_files,
             'sphinx_version': __display_version__,
             'sphinx_version_tuple': sphinx_version,