add cleanup.py

5 files changed, 203 insertions, 90 deletions

diff --git a/api.py b/api.py
index a83bae7..ab25f0a 100644
--- a/api.py
+++ b/api.py
@@ -29,25 +29,6 @@ class EjabberdApi:
             return f"{self._login['user']}@{self._login['server']}", self._login['password']
         return None
 
-    @property
-    def verstring(self):
-        if self._login is not None:
-            ver_str = re.compile('([1-9][0-9.]+(?![.a-z]))\\b')
-            status = self.cmd('status', {})
-
-            # matches
-            try:
-                tmp = ver_str.findall(status)[0]
-            # raise SystemExit code 17 if no status message is received
-            except TypeError:
-                raise SystemExit(17)
-
-            # return parsed version string
-            logging.debug(f"fetch version: {tmp}")
-            return version.parse(tmp)
-
-        return None
-
     def _rest(self, command: str, data) -> dict:
         # add authentication header to the session obj
         if self.session.auth is None:
diff --git a/cleanup.py b/cleanup.py
new file mode 100755
index 0000000..63947bd
--- /dev/null
+++ b/cleanup.py
@@ -0,0 +1,83 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+import logging
+import datetime
+from control import EjabberdCtl
+
+
+class EjabberdCleanup(EjabberdCtl):
+    def __init__(self, url, login, api):
+        super().__init__(url, login, api)
+        self.ignore_hosts = []
+        self.ignore_usernames = []
+        self.dry = False
+        self.skip_by_roster_roster = True
+        self.delete_not_login = True
+        self.offline_since_days = None
+
+    def delete_user(self, host, user, reason=""):
+        if self.dry:
+            logging.warning(f"{user}@{host}: dry delete : {reason}")
+        else:
+            self.cmd("unregister", {"host":  host, "user": user})
+            logging.warning(f"{user}@{host}: deleted - {reason}")
+
+    def run_user(self, host, user):
+        if self.skip_by_roster:
+            roster = self.cmd("get_roster",{"host":  host, "user": user})
+            if len(roster) > 0:
+                logging.debug(f"{user}@{host}: skipped it has a roster")
+                return
+        last = self.cmd("get_last",{"host":  host, "user": user})
+        if self.delete_not_login and last["status"] == "Registered but didn't login":
+            self.delete_user(host, user, "not login")
+            return
+        if self.offline_since_days is not None:
+            last_stamp = last["timestamp"]
+            lastdate = None
+            try:
+                lastdate = datetime.datetime.strptime(last_stamp,"%Y-%m-%dT%H:%M:%SZ")
+            except:
+                try:
+                    lastdate = datetime.datetime.strptime(last_stamp,"%Y-%m-%dT%H:%M:%S.%fZ")
+                except:
+                    logging.error(f"{user}@{host}: not able to parse '{last_stamp}'")
+                    return
+            if lastdate != None and  lastdate-datetime.datetime.now() > datetime.timedelta(days=self.offline_since_days):
+                self.delete_user(host, user, f"last seen @ {lastdate}")
+                return
+                
+
+    def run(self):
+        for host in self.fetch_vhosts():
+            if host in self.ignore_hosts:
+                continue
+            logging.info(f"run cleanup for host: {host}")
+            for user in self.cmd("registered_users",{"host": host}):
+                if user in self.ignore_usernames:
+                    continue
+                self.run_user(host, user)
+
+if __name__ == "__main__":
+    import json
+    from config import Config
+
+    # load config
+    config = Config()
+    if config.get('debug', default=False):
+        logging.getLogger().setLevel(logging.DEBUG)
+
+    # credentials and parameters
+    url = config.get('url', default='http://localhost:5280/api')
+    login = config.get('login', default=None)
+    api = config.get('api', default='rest')
+
+    # init handler
+    cleaner = EjabberdCleanup(url, login, api)
+    cleaner.dry = config.get('cleaner_dry',default=False)
+    cleaner.ignore_hosts = config.get('cleaner_ignore_hosts',default=[])
+    cleaner.ignore_usernames = config.get('cleaner_ignore_usernames',default=[])
+    cleaner.skip_by_roster = config.get('cleaner_skip_by_roster',default=True)
+    cleaner.delete_not_login = config.get('cleaner_delete_not_login',default=True)
+    cleaner.offline_since_days = config.get('cleaner_offline_since_days', default=None)
+    cleaner.run()
diff --git a/control.py b/control.py
new file mode 100644
index 0000000..3bbf3c7
--- /dev/null
+++ b/control.py
@@ -0,0 +1,84 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+import re
+import logging
+
+from packaging import version
+from api import EjabberdApi
+
+class EjabberdCtl(EjabberdApi):
+
+
+    @property
+    def verstring(self):
+        if self._login is not None:
+            ver_str = re.compile('([1-9][0-9.]+(?![.a-z]))\\b')
+            status = self.cmd('status', {})
+
+            # matches
+            try:
+                tmp = ver_str.findall(status)[0]
+            # raise SystemExit code 17 if no status message is received
+            except TypeError:
+                raise SystemExit(17)
+
+            # return parsed version string
+            logging.debug(f"fetch version: {tmp}")
+            return version.parse(tmp)
+
+        return None
+
+    def fetch_onlineuser(self):
+        tmp = self.cmd("connected_users_info", {})
+        if "connected_users_info" not in tmp:
+            return tmp
+        data = []
+        for c in tmp["connected_users_info"]:
+            if "session" not in c:
+                continue
+            user = {}
+            for attrs in c["session"]:
+                for k, v in attrs.items():
+                    user[k] = v
+            data.append(user)
+        return data
+
+    def fetch_nodes(self):
+        result = self.cmd("list_cluster", {})
+        if "nodes" not in result:
+            return result
+        data = []
+        for node in result["nodes"]:
+            data.append(node["node"])
+        return data
+
+    def fetch_vhosts(self):
+        result = self.cmd("registered_vhosts", {})
+        if "vhosts" not in result:
+            return result
+        data = []
+        for vhost in result["vhosts"]:
+            data.append(vhost["vhost"])
+        return data
+
+    def fetch_s2s_in(self):
+        result = self.cmd("incoming_s2s_number", {})
+        if "s2s_incoming" not in result:
+            return result
+        return result["s2s_incoming"]
+
+    def fetch_s2s_out(self):
+        result = self.cmd("outgoing_s2s_number", {})
+        if "s2s_outgoing" not in result:
+            return result
+        return result["s2s_outgoing"]
+
+    def fetch_registered(self, vhost=None):
+        if vhost is None:
+            result = self.cmd("stats", {"name":"registeredusers"})
+            if "stat" in result:
+                return result["stat"]
+        else:
+            result = self.cmd("stats_host", {"name":"registeredusers", "host": vhost})
+            if "stat" in result:
+                return result["stat"]
diff --git a/ejabberd-metrics.yml.default b/ejabberd-metrics.yml.default
index e89d0ef..db1c89a 100644
--- a/ejabberd-metrics.yml.default
+++ b/ejabberd-metrics.yml.default
@@ -15,6 +15,8 @@ muc_host: "chat"
 # api configuration
 # default : rpc
 api: "rest"
+# show debugging log messages
+debug: True
 
 # influx db configuration
 influxdb_host: "localhost"
@@ -25,3 +27,21 @@ influxdb_db: "example"
 prometheus_address: "127.0.0.1"
 prometheus_port: 8080
 prometheus_cache_ttl: 10
+
+# cleaner for users
+cleaner_dry: True # just test run (only loggging output)
+# skip remove of user if client has a roster
+cleaner_skip_by_roster: True
+# delete user if it was only registered but has never login
+cleaner_delete_not_login: True
+# delete user if the are offline for x days
+cleaner_offline_since_days: 7
+# ignore cleanup for host
+cleaner_ignore_hosts:
+ - "meckerspace.de"
+# ignore cleanup for username
+cleaner_ignore_usernames:
+ - "admin"
+ - "abuse"
+ - "bugs"
+ - "bot"
diff --git a/metrics.py b/metrics.py
index 2f6b82b..ad23033 100644..100755
--- a/metrics.py
+++ b/metrics.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 import ipaddress
 
-from api import EjabberdApi
+from control import EjabberdCtl
 
 # rfc6052: IPv6 Addressing of IPv4/IPv6 Translators
 nat64 = ipaddress.ip_network("64:ff9b::/96")
@@ -14,7 +14,7 @@ class EjabberdMetrics:
     """
     def __init__(self, url, login=None, api="rpc", muc_host: str = 'conference'):
         # init ejabberd api
-        self.api = EjabberdApi(url, login, api)
+        self.api = EjabberdCtl(url, login, api)
         self._cmd = self.api.cmd
 
         # variables
@@ -55,61 +55,6 @@ class EjabberdMetrics:
                 return 4
         return addr.version
 
-    def fetch_onlineuser(self):
-        tmp = self._cmd("connected_users_info", {})
-        if "connected_users_info" not in tmp:
-            return tmp
-        data = []
-        for c in tmp["connected_users_info"]:
-            if "session" not in c:
-                continue
-            user = {}
-            for attrs in c["session"]:
-                for k, v in attrs.items():
-                    user[k] = v
-            data.append(user)
-        return data
-
-    def fetch_nodes(self):
-        result = self._cmd("list_cluster", {})
-        if "nodes" not in result:
-            return result
-        data = []
-        for node in result["nodes"]:
-            data.append(node["node"])
-        return data
-
-    def fetch_vhosts(self):
-        result = self._cmd("registered_vhosts", {})
-        if "vhosts" not in result:
-            return result
-        data = []
-        for vhost in result["vhosts"]:
-            data.append(vhost["vhost"])
-        return data
-
-    def fetch_s2s_in(self):
-        result = self._cmd("incoming_s2s_number", {})
-        if "s2s_incoming" not in result:
-            return result
-        return result["s2s_incoming"]
-
-    def fetch_s2s_out(self):
-        result = self._cmd("outgoing_s2s_number", {})
-        if "s2s_outgoing" not in result:
-            return result
-        return result["s2s_outgoing"]
-
-    def fetch_registered(self, vhost=None):
-        if vhost is None:
-            result = self._cmd("stats", {"name":"registeredusers"})
-            if "stat" in result:
-                return result["stat"]
-        else:
-            result = self._cmd("stats_host", {"name":"registeredusers", "host": vhost})
-            if "stat" in result:
-                return result["stat"]
-
     def fetch_muc(self, vhost=None):
         host = "global"
         if vhost is not None:
@@ -124,15 +69,15 @@ class EjabberdMetrics:
 
     def update(self):
         # nodes
-        self._nodes = self.fetch_nodes()
+        self._nodes = self.api.fetch_nodes()
 
         # vhosts
-        self._vhosts = self.fetch_vhosts()
+        self._vhosts = self.api.fetch_vhosts()
 
         # registered
         if not hasattr(self, "_registered"):
             self._registered = {}
-        self._registered[None] = self.fetch_registered()
+        self._registered[None] = self.api.fetch_registered()
 
         # muc
         if not hasattr(self, "_muc"):
@@ -141,20 +86,20 @@ class EjabberdMetrics:
 
         # registered + muc
         for vhost in self._vhosts:
-            self._registered[vhost] = self.fetch_registered(vhost)
+            self._registered[vhost] = self.api.fetch_registered(vhost)
             self._muc[vhost] = self.fetch_muc(vhost)
 
         # online user
-        self._onlineuser = self.fetch_onlineuser()
+        self._onlineuser = self.api.fetch_onlineuser()
 
         # s2s
-        self._s2s_in = self.fetch_s2s_in()
-        self._s2s_out = self.fetch_s2s_out()
+        self._s2s_in = self.api.fetch_s2s_in()
+        self._s2s_out = self.api.fetch_s2s_out()
 
     def get_online_by(self, by="node", parse=None, vhost=None, node=None):
         parser = parse or (lambda a: a)
         if not hasattr(self, "_onlineuser"):
-            self._onlineuser = self.fetch_onlineuser()
+            self._onlineuser = self.api.fetch_onlineuser()
         data = {}
 
         for conn in self._onlineuser:
@@ -192,7 +137,7 @@ class EjabberdMetrics:
     def get_online_client_by(self, by="ip", parse=None, vhost=None, node=None):
         parser = parse or self._ipversion
         if not hasattr(self, "_onlineuser"):
-            self._onlineuser = self.fetch_onlineuser()
+            self._onlineuser = self.api.fetch_onlineuser()
         data = {}
 
         for conn in self._onlineuser:
@@ -221,7 +166,7 @@ class EjabberdMetrics:
         if not hasattr(self, "_registered"):
             self._registered = {}
         if vhost not in self._registered:
-            self._registered[vhost] = self.fetch_registered(vhost)
+            self._registered[vhost] = self.api.fetch_registered(vhost)
         return self._registered[vhost]
 
     def get_muc(self, vhost=None):
@@ -233,17 +178,17 @@ class EjabberdMetrics:
 
     def get_vhosts(self):
         if not hasattr(self, "_vhosts"):
-            self._vhosts = self.fetch_vhosts()
+            self._vhosts = self.api.fetch_vhosts()
         return self._vhosts
 
     def get_s2s_in(self):
         if not hasattr(self, "_s2s_in"):
-            self._s2s_in = self.fetch_s2s_in()
+            self._s2s_in = self.api.fetch_s2s_in()
         return self._s2s_in
 
     def get_s2s_out(self):
         if not hasattr(self, "_s2s_out"):
-            self._s2s_out = self.fetch_s2s_out()
+            self._s2s_out = self.api.fetch_s2s_out()
         return self._s2s_out
 
     def get_vhost_metrics(self, vhost): 
@@ -261,7 +206,7 @@ class EjabberdMetrics:
 
     def get_nodes(self):
         if not hasattr(self, "_nodes"):
-            self._nodes = self.fetch_nodes()
+            self._nodes = self.api.fetch_nodes()
         return self._nodes
 
     def get_node_metrics(self, node):