X-Git-Url: http://git.scottworley.com/paperdoorknob/blobdiff_plain/f75c16294c01dc2bdbe867e4477c2073c10680a8..228bb7bb52ef9df820fb41312144c87ee987434d:/glowfic.py diff --git a/glowfic.py b/glowfic.py index d1966ca..065c715 100644 --- a/glowfic.py +++ b/glowfic.py @@ -8,6 +8,7 @@ from abc import ABC, abstractmethod from dataclasses import dataclass import itertools +from urllib.parse import parse_qsl, urlencode, urlparse, urlunparse from typing import Iterable @@ -18,12 +19,29 @@ 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: str | None - screen_name: str | None - author: 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 @@ -56,34 +74,36 @@ def chunkDOMs(html: BeautifulSoup) -> Iterable[Tag]: def makeChunk(chunk_dom: Tag, image_store: ImageStore) -> Chunk: def getIcon() -> str | None: - icon_div = chunk_dom.find_next('div', class_='post-icon') + icon_div = chunk_dom.findChild('div', class_='post-icon') if icon_div is None: return None - icon_img = icon_div.find_next('img') + 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 getTextByClass(css_class: str) -> str | None: - div = chunk_dom.find_next('div', class_=css_class) - if div is None: - return None - return div.text.strip() + 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 - content = chunk_dom.find_next('div', class_='post-content') + content = chunk_dom.findChild('div', class_='post-content') assert isinstance(content, Tag) return Chunk(getIcon(), - getTextByClass('post-character'), - getTextByClass('post-screenname'), - getTextByClass('post-author'), + getByClass('post-character'), + getByClass('post-screenname'), + getByClass('post-author'), content) -def renderIcon(icon_path: str | None, image_size: float) -> str | None: - params = f'width={image_size}mm,height={image_size}mm,keepaspectratio' - return f'\\includegraphics[{params}]{{{icon_path}}}' if icon_path else None +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): @@ -109,12 +129,11 @@ class BelowIconLayout(Layout): self._image_size = image_size def renderChunk(self, chunk: Chunk) -> bytes: - meta = [ - renderIcon(chunk.icon, self._image_size), - chunk.character, - chunk.screen_name, - chunk.author, - ] + 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] @@ -138,7 +157,7 @@ class BelowIconLayout(Layout): \\noindent %s ''' % ( - b'\\\\*'.join(x.encode('UTF-8') for x in meta if x is not None), + b'\\\\*'.join(meta), self._texifier.texify(chunk.content)) @@ -172,6 +191,6 @@ class BesideIconLayout(Layout): \\strut ''' % ( - icon.encode('UTF-8') if icon else b'', - b'\\\\*'.join(x.encode('UTF-8') for x in meta if x is not None), + 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))