Updated to add pdf extraction

2024-04-02 11:30:27 +00:00 · 2024-04-02 11:30:27 +00:00 · cc91b10e04
commit cc91b10e04
parent 3649ea306f
1 changed files with 1 additions and 1 deletions
--- a/archiveteam_project_url_extractor.py
+++ b/archiveteam_project_url_extractor.py
@ -77,7 +77,7 @@ url_patterns = {
    r'(mediafire\.com|mfi\.re)\S+': 'filtered_mediafire_mfi.re.txt',
    r'\S*imgur\S*': 'filtered_imgur.txt',
    r'http(s)?://(www\.)?pastebin.com': 'filtered_pastebin.txt',
-    r'https://cdn.discordapp.com': 'filtered_cdn.discordapp.com.txt'
+    r'https?://\S+\.pdf\b'': 'filtered_cdn.discordapp.com.txt'
 }

 # Ask the user for the number of concurrent instances