ruby-directives.py

from docutils.core import publish_string
from docutils.nodes import TextElement, Inline, Text, unescape
from docutils.parsers import rst
from docutils.parsers.rst import directives
from docutils.writers.html4css1 import Writer, HTMLTranslator
import re

# ruby node
class ruby(Inline, TextElement): pass
class rt(Inline, TextElement): pass
class rp(Inline, TextElement): pass

def build_node(rawtext, node, delimiters='()'):
    dlm = re.escape(delimiters)
    rx = re.compile(f"([^{dlm}]+)(\s*[{dlm}]\s*)([^{dlm}]+)(\s*[{dlm}]\s*)")
    for m in re.finditer(rx, unescape(rawtext)):
        rbase, rpl, rant, rpr = m.groups()
        node += [
            Text(rbase),
            rp(rawtext, rpl),
            rt(rawtext, rant),
            rp(rawtext, rpr),
        ]
    return node


# ruby directive
class Ruby(rst.Directive):
    required_arguments = 1
    option_spec = {'delimiters': directives.unchanged,}

    def run(self):
        rawtext = self.arguments[0]
        node = ruby(rawtext, '', **self.options)
        return [build_node(
            rawtext,
            node,
            self.options.get('delimiters')
        )]

directives.register_directive('ruby', Ruby)


# rubi role
def rubi_role(role, rawtext, text, lineno, inliner, options=None, content=None):
    options = roles.normalized_role_options(options)
    node = ruby(rawtext, '', **options)
    return [build_node(
        rawtext,
        node,
        options.get('delimiters')
    )], []

ruby_role.options = {'delimiters': directives.unchanged}
roles.register_canonical_role('rubi', rubi_role)


# Transformer and Writer
class MyWriter(Writer):
    def __init__(self):
        self.parts = {}
        self.translator_class = MyHTMLTranslator

class MyHTMLTranslator(HTMLTranslator):
    def visit_ruby(self, node):
        self.body.append(self.starttag(node, 'ruby', suffix=''))

    def depart_ruby(self, node):
        self.body.append('</ruby>')
        
    def visit_rp(self, node):
        self.body.append(self.starttag(node, 'rp', suffix=''))

    def depart_rp(self, node):
        self.body.append('</rp>')
        
    def visit_rt(self, node):
        self.body.append(self.starttag(node, 'rt', suffix=''))

    def depart_rt(self, node):
        self.body.append('</rt>')


# Publisher
with open(source) as src, \
     open(destination, 'wb') as out:
    output = publish_string(src.read(), writer=MyWriter())
    out.write(output)