Reify Thread

[paperdoorknob] / texify.py

# paperdoorknob: Print glowfic
#
# This program is free software: you can redistribute it and/or modify it
# under the terms of the GNU General Public License as published by the
# Free Software Foundation, version 3.


from abc import ABC, abstractmethod
import re
import subprocess

from bs4.element import Tag


class Texifier(ABC):
    @abstractmethod
    def texify(self, html: Tag) -> bytes:
        raise NotImplementedError()


class PandocTexifier(Texifier):

    def __init__(self, pandoc_path: str) -> None:
        self._pandoc_path = pandoc_path

    def texify(self, html: Tag) -> bytes:
        return subprocess.run([self._pandoc_path, '--from=html', '--to=latex'],
                              input=html.encode(),
                              stdout=subprocess.PIPE,
                              check=True).stdout


class TexifierError(Exception):
    pass


class DirectTexifier(Texifier):
    def _texify_children(self, html: Tag) -> bytes:
        out = b''
        for c in html.children:
            if isinstance(c, str):
                out += c.encode('UTF-8')
            elif isinstance(c, Tag):
                out += self.texify(c).strip()
            else:
                raise TexifierError(f"Unsupported PageElement: {type(c)}")
        return re.sub(b'[ \n]+', b' ', out).strip() + b'\n'

    def texify(self, html: Tag) -> bytes:
        if html.name == 'em':
            return b'\\emph{' + self._texify_children(html).strip() + b'}\n'
        if html.name == 'a' and 'href' in html.attrs:
            return b'\\href{' + html.attrs['href'].encode(
                'UTF-8') + b'}{' + self._texify_children(html).strip() + b'}\n'
        return self._texify_children(html)


class TexifierVerificationError(Exception):
    pass


class VerifyingTexifier(Texifier):
    def __init__(self, a: Texifier, b: Texifier) -> None:
        self._a = a
        self._b = b

    def texify(self, html: Tag) -> bytes:
        aout = self._a.texify(html)
        bout = self._b.texify(html)
        if aout != bout:
            raise TexifierVerificationError(aout, bout)
        return aout