Log epicyon instances seen

daemon.py

@@ -94,6 +94,7 @@ from httpcodes import http_304
 from httpcodes import http_400
 from httpcodes import write2
 from httpheaders import set_headers
+from daemon_utils import load_known_epicyon_instances
 from daemon_utils import has_accept
 from daemon_utils import is_authorized
 from poison import load_dictionary
@@ -482,6 +483,8 @@ class EpicyonServer(ThreadingHTTPServer):
     headers_catalog = {}
     dictionary = []
     twograms = {}
+    searchable_by_default = {}
+    known_epicyon_instances = []
 
     def handle_error(self, request, client_address):
         # surpress connection reset errors
@@ -706,6 +709,9 @@ def run_daemon(accounts_data_dir: str,
     # default "searchable by" for new posts for each account
     httpd.searchable_by_default = load_searchable_by_default(base_dir)
 
+    # load the list of known Epicyon instances
+    httpd.known_epicyon_instances = load_known_epicyon_instances(base_dir)
+
     # if a custom robots.txt exists then read it
     robots_txt_filename = data_dir(base_dir) + '/robots.txt'
     httpd.robots_txt = None

daemon_get.py

@@ -67,6 +67,7 @@ from daemon_utils import has_accept
 from daemon_utils import show_person_options
 from daemon_utils import is_authorized
 from daemon_utils import get_user_agent
+from daemon_utils import log_epicyon_instances
 from httpheaders import update_headers_catalog
 from httpheaders import set_headers_etag
 from httpheaders import login_headers
@@ -338,6 +339,10 @@ def daemon_http_get(self) -> None:
 
     ua_str = get_user_agent(self)
 
+    if 'Epicyon/' in ua_str:
+        log_epicyon_instances(self.server.base_dir, calling_domain,
+                              self.server.known_epicyon_instances)
+
     if not _permitted_crawler_path(self.path):
         block, self.server.blocked_cache_last_updated, llm = \
             blocked_user_agent(calling_domain, ua_str,

daemon_head.py

@@ -21,6 +21,8 @@ from httpcodes import http_404
 from httpheaders import set_headers_head
 from media import path_is_video
 from media import path_is_audio
+from daemon_utils import get_user_agent
+from daemon_utils import log_epicyon_instances
 
 
 def daemon_http_head(self) -> None:
@@ -33,6 +35,13 @@ def daemon_http_head(self) -> None:
         return
 
     calling_domain = self.server.domain_full
+
+    ua_str = get_user_agent(self)
+
+    if 'Epicyon/' in ua_str:
+        log_epicyon_instances(self.server.base_dir, calling_domain,
+                              self.server.known_epicyon_instances)
+
     if self.headers.get('Host'):
         calling_domain = decoded_host(self.headers['Host'])
         if self.server.onion_domain:

daemon_post.py

@@ -39,6 +39,7 @@ from httpcodes import http_503
 from httpheaders import contains_suspicious_headers
 from httpheaders import update_headers_catalog
 from httpheaders import redirect_headers
+from daemon_utils import log_epicyon_instances
 from daemon_utils import get_user_agent
 from daemon_utils import post_to_outbox
 from daemon_utils import update_inbox_queue
@@ -163,6 +164,10 @@ def daemon_http_post(self) -> None:
 
     ua_str = get_user_agent(self)
 
+    if 'Epicyon/' in ua_str:
+        log_epicyon_instances(self.server.base_dir, calling_domain,
+                              self.server.known_epicyon_instances)
+
     block, self.server.blocked_cache_last_updated, _ = \
         blocked_user_agent(calling_domain, ua_str,
                            self.server.news_instance,

daemon_utils.py

@@ -26,6 +26,9 @@ from blocking import is_blocked_nickname
 from blocking import is_blocked_domain
 from content import valid_url_lengths
 from posts import add_to_field
+from utils import data_dir
+from utils import load_json
+from utils import save_json
 from utils import get_instance_url
 from utils import remove_html
 from utils import get_locked_account
@@ -863,3 +866,29 @@ def etag_exists(self, media_filename: str) -> bool:
                 # The file has not changed
                 return True
     return False
+
+
+def log_epicyon_instances(base_dir: str, calling_domain: str,
+                          known_epicyon_instances: []) -> None:
+    """Saves a log of known epicyon instances
+    """
+    if calling_domain in known_epicyon_instances:
+        return
+    known_epicyon_instances.append(calling_domain)
+    known_epicyon_instances.sort()
+    epicyon_instances_filename = \
+        data_dir(base_dir) + '/known_epicyon_instances.txt'
+    save_json(known_epicyon_instances, epicyon_instances_filename)
+
+
+def load_known_epicyon_instances(base_dir: str) -> []:
+    """Loads a list of known epicyon instances
+    """
+    epicyon_instances_filename = \
+        data_dir(base_dir) + '/known_epicyon_instances.txt'
+    if not os.path.isfile(epicyon_instances_filename):
+        return []
+    known_epicyon_instances = load_json(epicyon_instances_filename)
+    if not known_epicyon_instances:
+        return []
+    return known_epicyon_instances