Strip links from meta fields

[paperdoorknob] / glowfic.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
from dataclasses import dataclass
import itertools
from urllib.parse import parse_qsl, urlencode, urlparse, urlunparse

from typing import Iterable

from bs4 import BeautifulSoup
from bs4.element import Tag

from images import ImageStore
from texify import Texifier


def _removeViewFromURL(url: str) -> str:
    u = urlparse(url)
    old_qs = parse_qsl(u.query)
    new_qs = [(k, v) for k, v in old_qs if k != 'view']
    return urlunparse(u._replace(query=urlencode(new_qs)))


def nonFlatURL(url: str) -> str:
    return _removeViewFromURL(url)


def flatURL(url: str) -> str:
    u = urlparse(_removeViewFromURL(url))
    qs = parse_qsl(u.query) + [('view', 'flat')]
    return urlunparse(u._replace(query=urlencode(qs)))


@dataclass(frozen=True)
class Chunk:
    icon: str | None
    character: Tag | None
    screen_name: Tag | None
    author: Tag | None
    content: Tag

# We avoid the name "post" because the Glowfic community uses the term
# inconsistently:
#  * The Glowfic software sometimes uses "post" to refer to a whole thread
#    (in the URL), sometimes uses "post" to refer to chunks (in the CSS),
#    but mostly uses "post" to refer to just the first chunk in a thread
#    (in the HTML and UI).  The non-first chunks are "replies".
#  * Readers and this software don't need to distinguish first-chunks and
#    non-first-chunks.
#  * Humans in the community tend to use "posts" to mean chunks.


def chunkDOMs(html: BeautifulSoup) -> Iterable[Tag]:
    def text() -> Tag:
        body = html.body
        assert body
        text = body.find_next("div", class_="post-post")
        assert isinstance(text, Tag)
        return text

    def the_replies() -> Iterable[Tag]:
        rs = html.find_all("div", class_="post-reply")
        assert all(isinstance(r, Tag) for r in rs)
        return rs

    return itertools.chain([text()], the_replies())


def makeChunk(chunk_dom: Tag, image_store: ImageStore) -> Chunk:

    def getIcon() -> str | None:
        icon_div = chunk_dom.findChild('div', class_='post-icon')
        if icon_div is None:
            return None
        assert isinstance(icon_div, Tag)
        icon_img = icon_div.findChild('img')
        if icon_img is None:
            return None
        assert isinstance(icon_img, Tag)
        return image_store.get_image(icon_img.attrs['src'])

    def getByClass(css_class: str) -> Tag | None:
        tag = chunk_dom.findChild('div', class_=css_class)
        assert tag is None or isinstance(tag, Tag)
        return tag

    def stripHREF(tag: Tag) -> None:
        for c in tag.findChildren("a"):
            if "href" in c.attrs:
                del c.attrs["href"]

    def getMeta(css_class: str) -> Tag | None:
        tag = getByClass(css_class)
        if tag is None:
            return None
        stripHREF(tag)
        return tag

    content = chunk_dom.findChild('div', class_='post-content')
    assert isinstance(content, Tag)

    return Chunk(getIcon(),
                 getMeta('post-character'),
                 getMeta('post-screenname'),
                 getMeta('post-author'),
                 content)


def renderIcon(icon_path: str | None, image_size: float) -> bytes | None:
    if icon_path is None:
        return None
    return b'\\includegraphics[width=%fmm,height=%fmm,keepaspectratio]{%s}' % (
        image_size, image_size, icon_path.encode('UTF-8'))


class Layout(ABC):

    @abstractmethod
    def renderChunk(self, chunk: Chunk) -> bytes:
        raise NotImplementedError()


class ContentOnlyLayout(Layout):

    def __init__(self, texifier: Texifier) -> None:
        self._texifier = texifier

    def renderChunk(self, chunk: Chunk) -> bytes:
        return self._texifier.texify(chunk.content) + b'\n'


class BelowIconLayout(Layout):

    def __init__(self, texifier: Texifier, image_size: float) -> None:
        self._texifier = texifier
        self._image_size = image_size

    def renderChunk(self, chunk: Chunk) -> bytes:
        icon = renderIcon(chunk.icon, self._image_size)
        meta = [icon] if icon else []
        meta += [self._texifier.texify(x)
                 for x in [chunk.character, chunk.screen_name, chunk.author]
                 if x is not None]

        return b'''\\wrapstuffclear
\\begin{wrapstuff}[l]
\\fbox{
\\begin{varwidth}{0.5\\textwidth}
  \\smash{\\parbox[t][0pt]{0pt}{
    \\setlength{\\fboxrule}{0.2pt}
    \\setlength{\\fboxsep}{0pt}
    \\vspace{-3.4pt}
    \\fbox{\\hspace{107mm}}
  }\\\\*}
  \\vspace{-1em}
\\begin{center}
%s
\\end{center}
\\end{varwidth}
}
\\end{wrapstuff}

\\strut

\\noindent %s
''' % (
            b'\\\\*'.join(meta),
            self._texifier.texify(chunk.content))


class BesideIconLayout(Layout):

    def __init__(self, texifier: Texifier, image_size: float) -> None:
        self._texifier = texifier
        self._image_size = image_size

    def renderChunk(self, chunk: Chunk) -> bytes:
        icon = renderIcon(chunk.icon, self._image_size)
        meta = [
            chunk.character,
            chunk.screen_name,
            chunk.author,
        ]

        # Why is \textwidth not the width of the text?
        # Why is the width of the text .765\textwidth?
        return b'''\\noindent\\fbox{
%s
\\parbox[b]{.765\\textwidth}{
\\begin{center}
%s
\\end{center}
}
}\\\\*
\\vspace{-0.75em}\\\\*
\\noindent %s

\\strut

''' % (
            icon if icon else b'',
            b'\\\\*'.join(self._texifier.texify(x) for x in meta if x is not None),
            self._texifier.texify(chunk.content))