def transform_char(self, element: Element, /):
char_placeholder_pattern = re.compile(
r"""
(
(?P<u_plus> U\+ )? # optional U+ prefix
(
(?P<code_point_placeholder> X{4} ) # XXXX as code point placeholder
| (?P<code_point> 1?[0-9A-F]?[0-9A-F]{4} ) # or actual code point
)
)?
(
\s
(?P<glyph_placeholder> \[X\] ) # [X] as glyph placeholder
)?
(
\s
(?P<name> [A-Z0-9 -]+ ) # actual name
)?
""",
flags=re.VERBOSE,
)
transformed: Element = element.makeelement("span", {"class": "character"}, None)
transformed.tail = element.tail # type: ignore
text: str = element.text
element.getparent().replace(element, transformed)
if match := char_placeholder_pattern.fullmatch(text):
cps = set[int]()
if name := match.group("name"):
cps.add(ord(unicodedata2.lookup(name)))
def transform_element(self, element: Element, /):
match element.tag:
case "char":
self.transform_char(element)
case ("h1" | "figcaption" | "a") as tag if (mode := "numbering") in element.keys():
self.expand_placeholder_in_element(element, mode)
if tag == "a":
wrap = element.makeelement("cite", {}, None)
wrap.tail, element.tail = element.tail, None # type: ignore
element.getparent().replace(element, wrap)
wrap.append(element)
