Update app.py

app.py

@@ -308,6 +308,7 @@ def inference(input_batch,isurl,use_archive,filt_companies_topic,limit_companies
             elif(EXTRACTOR_NET == 'trafilatura'):
               try:
                   extracted = trafilatura.extract(trafilatura.fetch_url(url), include_comments=False, config=trafilatura_config, include_tables=False)
+                  assert len(extracted)>100, "[W] Failed extracting "+url+" retrying with archived version"
               except:
                     archive = is_in_archive(url)
                     if archive['archived']: