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 def chunkDOMs(html: BeautifulSoup) -> Iterable[Tag]:
  58     def text() -> Tag:
  59         body = html.body
  60         assert body
  61         text = body.find_next("div", class_="post-post")
  62         assert isinstance(text, Tag)
  63         return text
  64
  65     def the_replies() -> Iterable[Tag]:
  66         rs = html.find_all("div", class_="post-reply")
  67         assert all(isinstance(r, Tag) for r in rs)
  68         return rs
  69
  70     return itertools.chain([text()], the_replies())
  71
  72
  73 def makeChunk(chunk_dom: Tag, image_store: ImageStore) -> Chunk:
  74
  75     def getIcon() -> str | None:
  76         icon_div = chunk_dom.findChild('div', class_='post-icon')
  77         if icon_div is None:
  78             return None
  79         assert isinstance(icon_div, Tag)
  80         icon_img = icon_div.findChild('img')
  81         if icon_img is None:
  82             return None
  83         assert isinstance(icon_img, Tag)
  84         return image_store.get_image(icon_img.attrs['src'])
  85
  86     def getByClass(css_class: str) -> Tag | None:
  87         tag = chunk_dom.findChild('div', class_=css_class)
  88         assert tag is None or isinstance(tag, Tag)
  89         return tag
  90
  91     def stripHREF(tag: Tag) -> None:
  92         for c in tag.findChildren("a"):
  93             if "href" in c.attrs:
  94                 del c.attrs["href"]
  95
  96     def getMeta(css_class: str) -> Tag | None:
  97         tag = getByClass(css_class)
  98         if tag is None:
  99             return None
 100         stripHREF(tag)
 101         return tag
 102
 103     content = chunk_dom.findChild('div', class_='post-content')
 104     assert isinstance(content, Tag)
 105
 106     return Chunk(getIcon(),
 107                  getMeta('post-character'),
 108                  getMeta('post-screenname'),
 109                  getMeta('post-author'),
 110                  content)
 111
 112
 113 def renderChunk(texifier: Texifier, chunk: Chunk) -> bytes:
 114     return b''.join([
 115         br'\glowhead{',
 116         br'\glowicon{%s}' % chunk.icon.encode('UTF-8') if chunk.icon else b'',
 117         b'}{',
 118         texifier.texify(chunk.character) if chunk.character else b'',
 119         b'}{',
 120         texifier.texify(chunk.screen_name) if chunk.screen_name else b'',
 121         b'}{',
 122         texifier.texify(chunk.author) if chunk.author else b'',
 123         b'}',
 124         texifier.texify(chunk.content)])
 125
 126
 127 ContentOnlyLayout = br'''
 128 \newcommand{\glowhead}[4]{}
 129 '''
 130
 131
 132 BelowIconLayout = br'''
 133 \newcommand{\glowhead}[4]{\wrapstuffclear
 134 \begin{wrapstuff}[l]
 135 \fbox{
 136 \begin{varwidth}{0.5\textwidth}
 137   \smash{\parbox[t][0pt]{0pt}{
 138     \setlength{\fboxrule}{0.2pt}
 139     \setlength{\fboxsep}{0pt}
 140     \vspace{-3.4pt}
 141     \fbox{\hspace{107mm}}
 142   }\\*}
 143   \vspace{-1em}
 144 \begin{center}
 145 #1\ifnotempty
 146 {#1}{\\*}#2\ifnotempty
 147 {#2}{\\*}#3\ifnotempty
 148 {#3}{\\*}#4
 149 \end{center}
 150 \end{varwidth}
 151 }
 152 \end{wrapstuff}
 153
 154 \strut
 155
 156 \noindent}'''
 157
 158
 159 # Why is \textwidth not the width of the text?
 160 # Why is the width of the text .765\textwidth?
 161 BesideIconLayout = br'''
 162 \newcommand{\glowhead}[4]{
 163
 164 \strut
 165
 166 \noindent\fbox{
 167 #1
 168 \parbox[b]{.765\textwidth}{
 169 \begin{center}
 170 #2\ifnotempty
 171 {#2}{\\*}#3\ifnotempty
 172 {#3}{\\*}#4
 173 \end{center}
 174 }
 175 }\\*
 176 \vspace{-0.75em}\\*
 177 \noindent}'''