From aa060d9ba5891fd87fc62664fb30b75bf0463e3e Mon Sep 17 00:00:00 2001 From: Scott Worley Date: Wed, 20 Dec 2023 13:44:27 -0800 Subject: [PATCH 1/1] Reify Chunk --- glowfic.py | 48 +++++++++++++++++++++++++++++++++---- glowfic_test.py | 20 +++++++++++++++- testdata/empty-content.html | 32 +++++++++++++++++++++++++ 3 files changed, 95 insertions(+), 5 deletions(-) create mode 100644 testdata/empty-content.html diff --git a/glowfic.py b/glowfic.py index 901246b..4169786 100644 --- a/glowfic.py +++ b/glowfic.py @@ -5,6 +5,7 @@ # Free Software Foundation, version 3. +from dataclasses import dataclass import itertools from typing import Iterable @@ -12,15 +13,26 @@ from typing import Iterable from bs4 import BeautifulSoup from bs4.element import Tag +from images import ImageStore + + +@dataclass(frozen=True) +class Chunk: + icon: str | None + character: str | None + screen_name: str | None + author: str | None + content: Tag + # We avoid the name "post" because the Glowfic community uses the term # inconsistently: # * The Glowfic software sometimes uses "post" to refer to a whole thread -# (eg: in the URL), but more often uses "post" to refer to just the first -# chunk in a thread. The non-first chunks are "replies". +# (in the URL), sometimes uses "post" to refer to chunks (in the CSS), +# but mostly uses "post" to refer to just the first chunk in a thread +# (in the HTML and UI). The non-first chunks are "replies". # * Readers and this software don't need to distinguish first-chunks and # non-first-chunks. -# * Humans in the community tend to use "posts" to mean "chunks" ("replies" -# in the Glowfic software's lexicon). +# * Humans in the community tend to use "posts" to mean chunks. def chunkDOMs(html: BeautifulSoup) -> Iterable[Tag]: @@ -37,3 +49,31 @@ def chunkDOMs(html: BeautifulSoup) -> Iterable[Tag]: return rs return itertools.chain([text()], the_replies()) + + +def makeChunk(chunk_dom: Tag, image_store: ImageStore) -> Chunk: + + def getIcon() -> str | None: + icon_div = chunk_dom.find_next('div', class_='post-icon') + if icon_div is None: + return None + icon_img = icon_div.find_next('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() + + content = chunk_dom.find_next('div', class_='post-content') + assert isinstance(content, Tag) + + return Chunk(getIcon(), + getTextByClass('post-character'), + getTextByClass('post-screenname'), + getTextByClass('post-author'), + content) diff --git a/glowfic_test.py b/glowfic_test.py index d847333..d3a6d56 100644 --- a/glowfic_test.py +++ b/glowfic_test.py @@ -9,7 +9,8 @@ import unittest from bs4 import BeautifulSoup -from glowfic import chunkDOMs +from images import FakeImageStore +from glowfic import chunkDOMs, makeChunk class TestSplit(unittest.TestCase): @@ -46,5 +47,22 @@ class TestSplit(unittest.TestCase): [['The "post"'], ['1st reply'], ['2nd reply']]) +class TestMakeChunk(unittest.TestCase): + + def testEmptyContent(self) -> None: + with open('testdata/empty-content.html', 'rb') as f: + soup = BeautifulSoup(f, 'html.parser') + c = makeChunk(next(iter(chunkDOMs(soup))), FakeImageStore()) + self.assertEqual( + c.icon, + 'stored:https://d1anwqy6ci9o1i.cloudfront.net/' + + 'users%2F366%2Ficons%2Fxqmypqvflgdy28aorw9ml_shock.png') + self.assertEqual(c.character, 'Keltham') + self.assertEqual(c.screen_name, 'lawful chaotic') + self.assertEqual(c.author, 'Iarwain') + self.assertEqual(str(c.content), + '

') + + if __name__ == '__main__': unittest.main() diff --git a/testdata/empty-content.html b/testdata/empty-content.html new file mode 100644 index 0000000..e1dc3f5 --- /dev/null +++ b/testdata/empty-content.html @@ -0,0 +1,32 @@ + + +
+ +
+ +
+ + Permalink + +
+

+
+ +
+ + -- 2.44.1