generate_entities.py

import json
import sys

key_and_value = list(json.load(sys.stdin).items())
# Sort by descending length so we match the largest prefix first
key_and_value.sort(key=lambda x: (-len(x[0]), x[0]))

match_arms = {}

for key, value in key_and_value:
    assert key[0] == "&"
    key = key[1:]
    first_char = key[0]
    key = key[1:]

    match_arms.setdefault(first_char, []).append((key, value))

with open("src/entities.rs", "w") as f:
    f.write(
        """
// @generated
// this file is autogenerated by
// curl https://html.spec.whatwg.org/entities.json | python generate_entities.py

pub(crate) struct CharRef {
    /// Name as it appears escaped in HTML
    pub(crate) name: &'static str,
    /// Unescaped character codepoints
    pub(crate) characters: &'static str,
}

pub(crate) fn try_read_character_reference<E>(first_char: char, try_read: impl FnMut(&str) -> Result<bool, E>) -> Result<Option<CharRef>, E> {
    match first_char {
"""
    )

    for first_char, if_statements in sorted(match_arms.items()):
        # Write each branch of the match stmt as its own function such that
        # compilation is faster.
        f.write(
            """
        '%(first_char)s' => {
            #[allow(non_snake_case)]
            fn branch_%(first_char)s<E>(mut try_read: impl FnMut(&str) -> Result<bool, E>) -> Result<Option<CharRef>, E> {
                for (other_chars, characters) in &[
        """
            % {"first_char": first_char}
        )

        for other_chars, value in if_statements:
            characters = ""
            for c in value["codepoints"]:
                characters += r"\u{" + hex(c)[2:] + r"}"

            f.write(
                """
                    ("%(other_chars)s", "%(characters)s"),
            """
                % {"other_chars": other_chars, "characters": characters}
            )

        f.write(
            """
                ] {
                    if try_read(other_chars)? {
                        return Ok(Some(CharRef { name: other_chars, characters }));
                    }
                }

                Ok(None)
            }

            branch_%(first_char)s(try_read)

        }
        """
            % {"first_char": first_char}
        )

    f.write(
        """
        _ => Ok(None)
    }
}
    """
    )