Android fixes

bohdanbobrowski · Nov 14, 2024 · cf294e3 · cf294e3
1 parent 68a2c0e
commit cf294e3
Showing 1 changed file with 7 additions and 0 deletions.
diff --git a/blog2epub/crawlers/article_factory/default.py b/blog2epub/crawlers/article_factory/default.py
@@ -3,7 +3,9 @@
 from datetime import datetime
 from typing import Optional
 
+import bs4
 import dateutil
+import html5lib
 from lxml.etree import tostring
 from lxml.html.soupparser import fromstring
 from strip_tags import strip_tags  # type: ignore
@@ -125,6 +127,11 @@ def get_content(self) -> str:
                         content_html = re.sub(r'<a name=["\']more["\']/>', "", content_html)
                         content_html = re.sub(r"<div[^>]*>", "<p>", content_html)
                         content_html = content_html.replace("</div>", "")
+                        self.interface.print(f"html5lib {html5lib.__version__}")
+                        self.interface.print(f"BeautifulSoup {bs4.__version__}")
+                        soup = bs4.BeautifulSoup(content_html, "html5lib", multi_valued_attributes=False)
+                        if soup:
+                            pass
                         content = strip_tags(
                             content_html,
                             minify=True,