]> git.scottworley.com Git - paperdoorknob/blame_incremental - glowfic.py
Learning TeX: Keep icon image size in TeX
[paperdoorknob] / glowfic.py
... / ...
CommitLineData
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
8from abc import ABC, abstractmethod
9from dataclasses import dataclass
10import itertools
11from urllib.parse import parse_qsl, urlencode, urlparse, urlunparse
12
13from typing import Iterable
14
15from bs4 import BeautifulSoup
16from bs4.element import Tag
17
18from images import ImageStore
19from texify import Texifier
20
21
22def _removeViewFromURL(url: str) -> str:
23 u = urlparse(url)
24 old_qs = parse_qsl(u.query)
25 new_qs = [(k, v) for k, v in old_qs if k != 'view']
26 return urlunparse(u._replace(query=urlencode(new_qs)))
27
28
29def nonFlatURL(url: str) -> str:
30 return _removeViewFromURL(url)
31
32
33def flatURL(url: str) -> str:
34 u = urlparse(_removeViewFromURL(url))
35 qs = parse_qsl(u.query) + [('view', 'flat')]
36 return urlunparse(u._replace(query=urlencode(qs)))
37
38
39@dataclass(frozen=True)
40class Chunk:
41 icon: str | None
42 character: Tag | None
43 screen_name: Tag | None
44 author: Tag | None
45 content: Tag
46
47# We avoid the name "post" because the Glowfic community uses the term
48# inconsistently:
49# * The Glowfic software sometimes uses "post" to refer to a whole thread
50# (in the URL), sometimes uses "post" to refer to chunks (in the CSS),
51# but mostly uses "post" to refer to just the first chunk in a thread
52# (in the HTML and UI). The non-first chunks are "replies".
53# * Readers and this software don't need to distinguish first-chunks and
54# non-first-chunks.
55# * Humans in the community tend to use "posts" to mean chunks.
56
57
58def chunkDOMs(html: BeautifulSoup) -> Iterable[Tag]:
59 def text() -> Tag:
60 body = html.body
61 assert body
62 text = body.find_next("div", class_="post-post")
63 assert isinstance(text, Tag)
64 return text
65
66 def the_replies() -> Iterable[Tag]:
67 rs = html.find_all("div", class_="post-reply")
68 assert all(isinstance(r, Tag) for r in rs)
69 return rs
70
71 return itertools.chain([text()], the_replies())
72
73
74def makeChunk(chunk_dom: Tag, image_store: ImageStore) -> Chunk:
75
76 def getIcon() -> str | None:
77 icon_div = chunk_dom.findChild('div', class_='post-icon')
78 if icon_div is None:
79 return None
80 assert isinstance(icon_div, Tag)
81 icon_img = icon_div.findChild('img')
82 if icon_img is None:
83 return None
84 assert isinstance(icon_img, Tag)
85 return image_store.get_image(icon_img.attrs['src'])
86
87 def getByClass(css_class: str) -> Tag | None:
88 tag = chunk_dom.findChild('div', class_=css_class)
89 assert tag is None or isinstance(tag, Tag)
90 return tag
91
92 def stripHREF(tag: Tag) -> None:
93 for c in tag.findChildren("a"):
94 if "href" in c.attrs:
95 del c.attrs["href"]
96
97 def getMeta(css_class: str) -> Tag | None:
98 tag = getByClass(css_class)
99 if tag is None:
100 return None
101 stripHREF(tag)
102 return tag
103
104 content = chunk_dom.findChild('div', class_='post-content')
105 assert isinstance(content, Tag)
106
107 return Chunk(getIcon(),
108 getMeta('post-character'),
109 getMeta('post-screenname'),
110 getMeta('post-author'),
111 content)
112
113
114def renderIcon(icon_path: str | None) -> bytes | None:
115 if icon_path is None:
116 return None
117 return (
118 br'\includegraphics[width=\glowiconsize,height=\glowiconsize,keepaspectratio]{%s}' %
119 icon_path.encode('UTF-8'))
120
121
122class Layout(ABC):
123
124 @abstractmethod
125 def renderChunk(self, chunk: Chunk) -> bytes:
126 raise NotImplementedError()
127
128
129class ContentOnlyLayout(Layout):
130
131 def __init__(self, texifier: Texifier) -> None:
132 self._texifier = texifier
133
134 def renderChunk(self, chunk: Chunk) -> bytes:
135 return self._texifier.texify(chunk.content) + b'\n'
136
137
138class BelowIconLayout(Layout):
139
140 def __init__(self, texifier: Texifier) -> None:
141 self._texifier = texifier
142
143 def renderChunk(self, chunk: Chunk) -> bytes:
144 icon = renderIcon(chunk.icon)
145 meta = [icon] if icon else []
146 meta += [self._texifier.texify(x)
147 for x in [chunk.character, chunk.screen_name, chunk.author]
148 if x is not None]
149
150 return br'''\wrapstuffclear
151\begin{wrapstuff}[l]
152\fbox{
153\begin{varwidth}{0.5\textwidth}
154 \smash{\parbox[t][0pt]{0pt}{
155 \setlength{\fboxrule}{0.2pt}
156 \setlength{\fboxsep}{0pt}
157 \vspace{-3.4pt}
158 \fbox{\hspace{107mm}}
159 }\\*}
160 \vspace{-1em}
161\begin{center}
162%s
163\end{center}
164\end{varwidth}
165}
166\end{wrapstuff}
167
168\strut
169
170\noindent %s
171''' % (
172 br'\\*'.join(meta),
173 self._texifier.texify(chunk.content))
174
175
176class BesideIconLayout(Layout):
177
178 def __init__(self, texifier: Texifier) -> None:
179 self._texifier = texifier
180
181 def renderChunk(self, chunk: Chunk) -> bytes:
182 icon = renderIcon(chunk.icon)
183 meta = [
184 chunk.character,
185 chunk.screen_name,
186 chunk.author,
187 ]
188
189 # Why is \textwidth not the width of the text?
190 # Why is the width of the text .765\textwidth?
191 return br'''\noindent\fbox{
192%s
193\parbox[b]{.765\textwidth}{
194\begin{center}
195%s
196\end{center}
197}
198}\\*
199\vspace{-0.75em}\\*
200\noindent %s
201
202\strut
203
204''' % (
205 icon if icon else b'',
206 br'\\*'.join(self._texifier.texify(x) for x in meta if x is not None),
207 self._texifier.texify(chunk.content))