git.scottworley.com Git - paperdoorknob/blob - glowfic.py

   1 # paperdoorknob: Print glowfic
   2 #
   3 # This program is free software: you can redistribute it and/or modify it
   4 # under the terms of the GNU General Public License as published by the
   5 # Free Software Foundation, version 3.
   6
   7
   8 from dataclasses import dataclass
   9 import itertools
  10 from urllib.parse import parse_qsl, urlencode, urlparse, urlunparse
  11
  12 from typing import Iterable
  13
  14 from bs4 import BeautifulSoup
  15 from bs4.element import Tag
  16
  17 from images import ImageStore
  18 from texify import Texifier
  19
  20
  21 def _removeViewFromURL(url: str) -> str:
  22     u = urlparse(url)
  23     old_qs = parse_qsl(u.query)
  24     new_qs = [(k, v) for k, v in old_qs if k != 'view']
  25     return urlunparse(u._replace(query=urlencode(new_qs)))
  26
  27
  28 def nonFlatURL(url: str) -> str:
  29     return _removeViewFromURL(url)
  30
  31
  32 def flatURL(url: str) -> str:
  33     u = urlparse(_removeViewFromURL(url))
  34     qs = parse_qsl(u.query) + [('view', 'flat')]
  35     return urlunparse(u._replace(query=urlencode(qs)))
  36
  37
  38 @dataclass(frozen=True)
  39 class Chunk:
  40     icon: str | None
  41     character: Tag | None
  42     screen_name: Tag | None
  43     author: Tag | None
  44     content: Tag
  45
  46 # We avoid the name "post" because the Glowfic community uses the term
  47 # inconsistently:
  48 #  * The Glowfic software sometimes uses "post" to refer to a whole thread
  49 #    (in the URL), sometimes uses "post" to refer to chunks (in the CSS),
  50 #    but mostly uses "post" to refer to just the first chunk in a thread
  51 #    (in the HTML and UI).  The non-first chunks are "replies".
  52 #  * Readers and this software don't need to distinguish first-chunks and
  53 #    non-first-chunks.
  54 #  * Humans in the community tend to use "posts" to mean chunks.
  55
  56
  57 class Thread:
  58
  59     def __init__(self, html: BeautifulSoup) -> None:
  60         self._html = html
  61
  62     def chunkDOMs(self) -> Iterable[Tag]:
  63         def text() -> Tag:
  64             body = self._html.body
  65             assert body
  66             text = body.find_next("div", class_="post-post")
  67             assert isinstance(text, Tag)
  68             return text
  69
  70         def the_replies() -> Iterable[Tag]:
  71             rs = self._html.find_all("div", class_="post-reply")
  72             assert all(isinstance(r, Tag) for r in rs)
  73             return rs
  74
  75         return itertools.chain([text()], the_replies())
  76
  77
  78 def makeChunk(chunk_dom: Tag, image_store: ImageStore) -> Chunk:
  79
  80     def getIcon() -> str | None:
  81         icon_div = chunk_dom.findChild('div', class_='post-icon')
  82         if icon_div is None:
  83             return None
  84         assert isinstance(icon_div, Tag)
  85         icon_img = icon_div.findChild('img')
  86         if icon_img is None:
  87             return None
  88         assert isinstance(icon_img, Tag)
  89         return image_store.get_image(icon_img.attrs['src'])
  90
  91     def getByClass(css_class: str) -> Tag | None:
  92         tag = chunk_dom.findChild('div', class_=css_class)
  93         assert tag is None or isinstance(tag, Tag)
  94         return tag
  95
  96     def stripHREF(tag: Tag) -> None:
  97         for c in tag.findChildren("a"):
  98             if "href" in c.attrs:
  99                 del c.attrs["href"]
 100
 101     def getMeta(css_class: str) -> Tag | None:
 102         tag = getByClass(css_class)
 103         if tag is None:
 104             return None
 105         stripHREF(tag)
 106         return tag
 107
 108     content = chunk_dom.findChild('div', class_='post-content')
 109     assert isinstance(content, Tag)
 110
 111     return Chunk(getIcon(),
 112                  getMeta('post-character'),
 113                  getMeta('post-screenname'),
 114                  getMeta('post-author'),
 115                  content)
 116
 117
 118 def renderChunk(texifier: Texifier, chunk: Chunk) -> bytes:
 119     return b''.join([
 120         br'\glowhead{',
 121         br'\glowicon{%s}' % chunk.icon.encode('UTF-8') if chunk.icon else b'',
 122         b'}{',
 123         texifier.texify(chunk.character) if chunk.character else b'',
 124         b'}{',
 125         texifier.texify(chunk.screen_name) if chunk.screen_name else b'',
 126         b'}{',
 127         texifier.texify(chunk.author) if chunk.author else b'',
 128         b'}',
 129         texifier.texify(chunk.content)])
 130
 131
 132 ContentOnlyLayout = br'''
 133 \newcommand{\glowhead}[4]{}
 134 '''
 135
 136
 137 BelowIconLayout = br'''
 138 \newcommand{\glowhead}[4]{\wrapstuffclear
 139 \begin{wrapstuff}[l]
 140 \fbox{
 141 \begin{varwidth}{0.5\textwidth}
 142   \smash{\parbox[t][0pt]{0pt}{
 143     \setlength{\fboxrule}{0.2pt}
 144     \setlength{\fboxsep}{0pt}
 145     \vspace{-3.4pt}
 146     \fbox{\hspace{107mm}}
 147   }\\*}
 148   \vspace{-1em}
 149 \begin{center}
 150 #1\ifnotempty
 151 {#1}{\\*}#2\ifnotempty
 152 {#2}{\\*}#3\ifnotempty
 153 {#3}{\\*}#4
 154 \end{center}
 155 \end{varwidth}
 156 }
 157 \end{wrapstuff}
 158
 159 \strut
 160
 161 \noindent}'''
 162
 163
 164 # Why is \textwidth not the width of the text?
 165 # Why is the width of the text .765\textwidth?
 166 BesideIconLayout = br'''
 167 \newcommand{\glowhead}[4]{
 168
 169 \strut
 170
 171 \noindent\fbox{
 172 #1
 173 \parbox[b]{.765\textwidth}{
 174 \begin{center}
 175 #2\ifnotempty
 176 {#2}{\\*}#3\ifnotempty
 177 {#3}{\\*}#4
 178 \end{center}
 179 }
 180 }\\*
 181 \vspace{-0.75em}\\*
 182 \noindent}'''