Ensure that number of hashtags stored per post is limited

src/daemon_utils.py

@@ -168,7 +168,8 @@ def post_to_outbox(self, message_json: {}, version: str,
                                   self.server.auto_cw_cache,
                                   self.server.block_federated,
                                   self.server.mitm_servers,
-                                  self.server.instance_software)
+                                  self.server.instance_software,
+                                  self.server.max_hashtags)
 
 
 def _get_outbox_thread_index(self, nickname: str,

src/inbox.py

@@ -2347,7 +2347,8 @@ def _inbox_after_initial(server, inbox_start_time,
                         server.block_military,
                         server.block_government,
                         server.block_bluesky,
-                        server.block_nostr):
+                        server.block_nostr,
+                        max_hashtags):
         if debug:
             print('DEBUG: Announce accepted from ' + actor)
         fitness_performance(inbox_start_time, server.fitness,
@@ -2832,7 +2833,8 @@ def _inbox_after_initial(server, inbox_start_time,
 
             store_hash_tags(base_dir, handle_name, domain,
                             http_prefix, domain_full,
-                            post_json_object, translate, session)
+                            post_json_object, translate, session,
+                            max_hashtags)
             fitness_performance(inbox_start_time,
                                 server.fitness,
                                 'INBOX', 'store_hash_tags',

src/inbox_receive.py

@@ -2109,7 +2109,8 @@ def receive_announce(recent_posts_cache: {},
                      block_military: {},
                      block_government: {},
                      block_bluesky: {},
-                     block_nostr: {}) -> bool:
+                     block_nostr: {},
+                     max_hashtags: int) -> bool:
     """Receives an announce activity within the POST section of HTTPServer
     """
     if message_json['type'] != 'Announce':
@@ -2392,7 +2393,8 @@ def receive_announce(recent_posts_cache: {},
 
         store_hash_tags(base_dir, nickname, domain,
                         http_prefix, domain_full,
-                        post_json_object, translate, session)
+                        post_json_object, translate, session,
+                        max_hashtags)
         # Try to obtain the actor for this person
         # so that their avatar can be shown
         lookup_actor = None

src/newsdaemon.py

@@ -293,7 +293,7 @@ def hashtag_rule_tree(operators: [],
 def _hashtag_add(base_dir: str, http_prefix: str, domain_full: str,
                  post_json_object: {},
                  action_str: str, hashtags: [], system_language: str,
-                 translate: {}, session) -> None:
+                 translate: {}, session, max_hashtags: int) -> None:
     """Adds a hashtag via a hashtag rule
     """
     add_hashtag = action_str.split('add ', 1)[1].strip()
@@ -341,9 +341,8 @@ def _hashtag_add(base_dir: str, http_prefix: str, domain_full: str,
     domain = domain_full
     if ':' in domain:
         domain = remove_domain_port(domain)
-    store_hash_tags(base_dir, 'news', domain,
-                    http_prefix, domain_full,
-                    post_json_object, translate, session)
+    store_hash_tags(base_dir, 'news', domain, http_prefix, domain_full,
+                    post_json_object, translate, session, max_hashtags)
 
 
 def _hashtag_remove(http_prefix: str, domain_full: str, post_json_object: {},
@@ -384,7 +383,8 @@ def _newswire_hashtag_processing(base_dir: str, post_json_object: {},
                                  domain: str, port: int,
                                  moderated: bool, url: str,
                                  system_language: str,
-                                 translate: {}, session) -> bool:
+                                 translate: {}, session,
+                                 max_hashtags: int) -> bool:
     """Applies hashtag rules to a news post.
     Returns true if the post should be saved to the news timeline
     of this instance
@@ -433,7 +433,8 @@ def _newswire_hashtag_processing(base_dir: str, post_json_object: {},
             # add a hashtag
             _hashtag_add(base_dir, http_prefix, domain_full,
                          post_json_object, action_str, hashtags,
-                         system_language, translate, session)
+                         system_language, translate, session,
+                         max_hashtags)
         elif action_str.startswith('remove '):
             # remove a hashtag
             _hashtag_remove(http_prefix, domain_full, post_json_object,
@@ -554,7 +555,8 @@ def _convert_rss_to_activitypub(base_dir: str, http_prefix: str,
                                 content_license_url: str,
                                 media_license_url: str,
                                 media_creator: str,
-                                session, debug: bool) -> None:
+                                session, debug: bool,
+                                max_hashtags: int) -> None:
     """Converts rss items in a newswire into posts
     """
     if not newswire:
@@ -705,7 +707,7 @@ def _convert_rss_to_activitypub(base_dir: str, http_prefix: str,
             _newswire_hashtag_processing(base_dir, blog, hashtags,
                                          http_prefix, domain, port,
                                          moderated, url, system_language,
-                                         translate, session)
+                                         translate, session, max_hashtags)
 
         # save the post and update the index
         if save_post:
@@ -744,9 +746,8 @@ def _convert_rss_to_activitypub(base_dir: str, http_prefix: str,
                 if tag not in newswire[original_date_str][6]:
                     newswire[original_date_str][6].append(tag)
 
-            store_hash_tags(base_dir, 'news', domain,
-                            http_prefix, domain_full,
-                            blog, translate, session)
+            store_hash_tags(base_dir, 'news', domain, http_prefix, domain_full,
+                            blog, translate, session, max_hashtags)
 
             clear_from_post_caches(base_dir, recent_posts_cache, post_id)
             if save_json(blog, filename):
@@ -852,7 +853,8 @@ def run_newswire_daemon(base_dir: str, httpd,
                                     httpd.low_bandwidth,
                                     httpd.content_license_url,
                                     httpd.content_license_url, '',
-                                    httpd.session, httpd.debug)
+                                    httpd.session, httpd.debug,
+                                    httpd.max_hashtags)
         print('Newswire feed converted to ActivityPub')
 
         if httpd.max_news_posts > 0:

src/outbox.py

@@ -316,7 +316,8 @@ def post_message_to_outbox(session, translate: {},
                            auto_cw_cache: {},
                            block_federated: [],
                            mitm_servers: [],
-                           instance_software: {}) -> bool:
+                           instance_software: {},
+                           max_hashtags: int) -> bool:
     """post is received by the outbox
     Client to server message post
     https://www.w3.org/TR/activitypub/#client-to-server-outbox-delivery
@@ -564,7 +565,8 @@ def post_message_to_outbox(session, translate: {},
 
     store_hash_tags(base_dir, post_to_nickname, domain,
                     http_prefix, domain_full,
-                    message_json, translate, session)
+                    message_json, translate, session,
+                    max_hashtags)
 
     # if this is a blog post or an event then save to its own box
     if message_json['type'] == 'Create':

src/schedule.py

@@ -175,7 +175,8 @@ def _update_post_schedule(base_dir: str, handle: str, httpd,
                                           httpd.auto_cw_cache,
                                           httpd.block_federated,
                                           httpd.mitm_servers,
-                                          httpd.instance_software):
+                                          httpd.instance_software,
+                                          httpd.max_hashtags):
                 index_lines.remove(line)
                 ex_text = \
                     'EX: _update_post_schedule unable to delete ' + \

src/webapp_hashtagswarm.py

@@ -416,7 +416,7 @@ def _store_tag_name(base_dir: str, nickname: str,
 def store_hash_tags(base_dir: str, nickname: str, domain: str,
                     http_prefix: str, domain_full: str,
                     post_json_object: {}, translate: {},
-                    session) -> None:
+                    session, max_hashtags: int) -> None:
     """Extracts hashtags from an incoming post and updates the
     relevant tags files.
     """
@@ -484,12 +484,14 @@ def store_hash_tags(base_dir: str, nickname: str, domain: str,
                            map_links, published,
                            tag_maps_dir):
             hashtags_ctr += 1
+            if hashtags_ctr >= max_hashtags:
+                break
 
     # treat event categories as hashtags
     # See https://codeberg.org/fediverse/
     # fep/src/branch/main/fep/8a8e/fep-8a8e.md
     event_category_str = get_category_from_post(post_json_object, translate)
-    if event_category_str:
+    if event_category_str and hashtags_ctr < max_hashtags:
         if ', ' in event_category_str:
             event_category_list = event_category_str.split(', ')
         else:
@@ -506,6 +508,8 @@ def store_hash_tags(base_dir: str, nickname: str, domain: str,
                                    map_links, published,
                                    tag_maps_dir):
                     hashtags_ctr += 1
+                    if hashtags_ctr >= max_hashtags:
+                        break
 
     # if some hashtags were found then recalculate the swarm
     # ready for later display