update bi paragraph/subheadline selector

flairNLP · Apr 19, 2024 · 9e6868e · 9e6868e
1 parent bab5b3d
commit 9e6868e
Showing 1 changed file with 3 additions and 1 deletion.
diff --git a/src/fundus/publishers/us/business_insider.py b/src/fundus/publishers/us/business_insider.py
@@ -16,7 +16,7 @@
 class BusinessInsiderParser(ParserProxy):
     class V1(BaseParser):
         _summary_selector = CSSSelector("article ul[class^='summary-list'] > li")
-        _subheadline_selector = CSSSelector("article h2")
+        _subheadline_selector = CSSSelector("article h2, div.slideshow-slide-container h2")
         _paragraph_selector = XPath(
             """
             //article 
@@ -25,6 +25,8 @@ class V1(BaseParser):
             //article 
             //div[contains(@class, 'content-lock-content')]
             /div[contains(@class, 'premium-content')] 
+            /p[not(contains(@class, 'disclaimer'))] | 
+            //div[@class='slide-layout clearfix']
             /p[not(contains(@class, 'disclaimer'))]
             """
         )