update

.github/workflows/actions.yml

@@ -3,9 +3,9 @@ name: Run Scripts
 on:
   schedule:
     # Run newday.py every day.
-    - cron: '3 13 * * *'
+    - cron: '52 13 * * *'
     # Run update_BoW.py every day
-    - cron: '45 13 * * *'
+    - cron: '20 14 * * *'
 
 jobs:
   build:
@@ -33,7 +33,7 @@ jobs:
         run: python project/newday.py
 
       - name: Execute update_BoW.py (every day)
-        if: ${{ github.event_name == 'schedule' && github.event.cron == '25 13 * * *' }}
+        if: ${{github.event.cron == '25 13 * * *' }}
         run: python update_BoW.py
 
       - name: Commit files

project/main2.py

@@ -33,7 +33,7 @@ def categorize_articles(all_links):
                         visited_links[link] = domain
                         print(link, " is ", domain) 
                     except Exception as e:
-                        print(f"Error downloading article from {link}: {e}")
+                        print(f"Error downloading article from {link}: {e}\n")
                         continue  # Continue with the next iteration of the loop
 
     # Separate links based on their domains

project/newday.py

@@ -3,7 +3,7 @@
 import nltk
 from pathlib import Path
 nltk.download('punkt')
-
+nltk.download('stopwords')
 csv_folder = Path(__file__).resolve().parent  # Get the directory of the script
 files = [
     csv_folder/'india.csv',

update_BoW.py

@@ -9,6 +9,7 @@
 from pathlib import Path
 
 nltk.download('punkt')
+nltk.download('stopwords')
 # Get the directory of the script
 csv_folder = Path(__file__).resolve().parent