Always have Thread.__init__ fetch the HTML

[paperdoorknob] / glowfic.py

diff --git a/glowfic.py b/glowfic.py
index 92ae4ab81a195fceeda74869fad340d9718c4095..5ffaac31921e360961967832ad090074e5cd2aaa 100644 (file)
--- a/glowfic.py
+++ b/glowfic.py
@@ -15,6 +15,7 @@ from bs4 import BeautifulSoup
 from bs4.element import Tag
 
 from images import ImageStore
+from spec import Spec
 from texify import Texifier
 
 
@@ -56,8 +57,11 @@ class Chunk:
 
 class Thread:
 
-    def __init__(self, dom: BeautifulSoup) -> None:
-        self._dom = dom
+    def __init__(self, spec: Spec) -> None:
+        spec.log('Fetching HTML...\r')
+        html = spec.fetcher.fetch(flatURL(spec.url))
+        spec.log('Parsing HTML...\r')
+        self._dom = BeautifulSoup(spec.htmlfilter(html), 'html.parser')
 
     def title(self) -> str | None:
         span = self._dom.findChild("span", id="post-title")