X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/67612898f2efcbb985da8534e0cce124152601db..53f396b459889b46e6feea48e8fa8e89b2452e9f:/glowfic.py diff --git a/glowfic.py b/glowfic.py index bda6d19..81a5342 100644 --- a/glowfic.py +++ b/glowfic.py @@ -5,7 +5,6 @@ # 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 @@ -16,6 +15,7 @@ from bs4 import BeautifulSoup from bs4.element import Tag from images import ImageStore +from spec import Spec from texify import Texifier @@ -55,20 +55,48 @@ class Chunk: # * 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 +class Thread: - def the_replies() -> Iterable[Tag]: - rs = html.find_all("div", class_="post-reply") - assert all(isinstance(r, Tag) for r in rs) - return rs + def __init__(self, spec: Spec) -> None: + def find_next_thread(dom: BeautifulSoup) -> str | None: + for c in dom.findChildren('div', class_='post-navheader'): + for a in c.findChildren('a'): + if 'Next Post' in a.text and 'href' in a.attrs and isinstance( + a.attrs['href'], str): + return a.attrs['href'] + return None + + spec.log('Fetching HTML...\r') + html = spec.fetcher.fetch(spec.url) + flat_html = spec.fetcher.fetch(flatURL(spec.url)) + spec.log('Parsing HTML...\r') + self._next_thread = find_next_thread( + BeautifulSoup(spec.htmlfilter(html), 'html.parser')) + self._dom = BeautifulSoup(spec.htmlfilter(flat_html), 'html.parser') + + def title(self) -> str | None: + span = self._dom.findChild("span", id="post-title") + if not isinstance(span, Tag): + return None + return span.text.strip() + + def next_thread(self) -> str | None: + return self._next_thread + + def chunkDOMs(self) -> Iterable[Tag]: + def text() -> Tag: + body = self._dom.body + assert body + text = body.find_next("div", class_="post-post") + assert isinstance(text, Tag) + return text - return itertools.chain([text()], the_replies()) + def the_replies() -> Iterable[Tag]: + rs = self._dom.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: @@ -111,43 +139,27 @@ def makeChunk(chunk_dom: Tag, image_store: ImageStore) -> Chunk: content) -def renderIcon(icon_path: str | None, image_size: float) -> bytes | None: - if icon_path is None: - return None - return br'\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() - +def renderChunk(texifier: Texifier, chunk: Chunk) -> bytes: + return b''.join([ + br'\glowhead{', + br'\glowicon{%s}' % chunk.icon.encode('UTF-8') if chunk.icon else b'', + b'}{', + texifier.texify(chunk.character) if chunk.character else b'', + b'}{', + texifier.texify(chunk.screen_name) if chunk.screen_name else b'', + b'}{', + texifier.texify(chunk.author) if chunk.author else b'', + b'}', + texifier.texify(chunk.content)]) -class ContentOnlyLayout(Layout): - def __init__(self, texifier: Texifier) -> None: - self._texifier = texifier +ContentOnlyLayout = br''' +\newcommand{\glowhead}[4]{} +''' - 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 br'''\wrapstuffclear +BelowIconLayout = br''' +\newcommand{\glowhead}[4]{\wrapstuffclear \begin{wrapstuff}[l] \fbox{ \begin{varwidth}{0.5\textwidth} @@ -159,7 +171,10 @@ class BelowIconLayout(Layout): }\\*} \vspace{-1em} \begin{center} -%s +#1\ifnotempty +{#1}{\\*}#2\ifnotempty +{#2}{\\*}#3\ifnotempty +{#3}{\\*}#4 \end{center} \end{varwidth} } @@ -167,42 +182,25 @@ class BelowIconLayout(Layout): \strut -\noindent %s -''' % ( - br'\\*'.join(meta), - self._texifier.texify(chunk.content)) - +\noindent}''' -class BesideIconLayout(Layout): - def __init__(self, texifier: Texifier, image_size: float) -> None: - self._texifier = texifier - self._image_size = image_size +# Why is \textwidth not the width of the text? +# Why is the width of the text .765\textwidth? +BesideIconLayout = br''' +\newcommand{\glowhead}[4]{ - def renderChunk(self, chunk: Chunk) -> bytes: - icon = renderIcon(chunk.icon, self._image_size) - meta = [ - chunk.character, - chunk.screen_name, - chunk.author, - ] +\strut - # Why is \textwidth not the width of the text? - # Why is the width of the text .765\textwidth? - return br'''\noindent\fbox{ -%s +\noindent\fbox{ +#1 \parbox[b]{.765\textwidth}{ \begin{center} -%s +#2\ifnotempty +{#2}{\\*}#3\ifnotempty +{#3}{\\*}#4 \end{center} } }\\* \vspace{-0.75em}\\* -\noindent %s - -\strut - -''' % ( - icon if icon else b'', - br'\\*'.join(self._texifier.texify(x) for x in meta if x is not None), - self._texifier.texify(chunk.content)) +\noindent}'''