Don't use requests session with multithreading

src/oabot/main.py

@@ -263,7 +263,7 @@ def keep_existing_url(self, url):
             return True
 
         try:
-            r = SESSION.head(url, timeout=5, allow_redirects=True)
+            r = requests.head(url, timeout=(5, 1), allow_redirects=True, headers={'User-Agent': OABOT_USER_AGENT})
         except requests.exceptions.RequestException:
             r = None
         # Avoid changing an URL which already clearly points to an open PDF
@@ -418,7 +418,7 @@ def get_oa_link(paper, doi=None, only_unpaywall=True):
     for url in sort_links(candidate_urls):
         if url:
             try:
-                head = SESSION.head(url, timeout=10)
+                head = requests.head(url, timeout=(5, 1), allow_redirects=True, headers={'User-Agent': OABOT_USER_AGENT})
                 head.raise_for_status()
                 if head.status_code < 400 and 'Location' in head.headers and urllib.parse.urlparse(head.headers['Location']).path == '/':
                     # Redirects to main page: fake status code, should be not found

src/oabot/settings.py

@@ -9,7 +9,7 @@
     # Mount point is '/'
     OABOT_APP_MOUNT_POINT = ''
 
-OABOT_USER_AGENT = 'OAbot/1.0 (+http://enwp.org/WP:OABOT)'
+OABOT_USER_AGENT = 'OAbot/1.0 (+http://enwp.org/WP:OABOT) not Googlebot'
 
 # the bot will not make any changes to these templates
 excluded_templates = ['cite arxiv', 'cite web', 'cite news', 'cite book']