Improve deduplication

A stanza-id is only unique within a specific archive
This adds the archive jid to the query

gajim/common/connection_handlers_events.py

@@ -1062,7 +1062,7 @@ class MamMessageReceivedEvent(nec.NetworkIncomingEvent, HelperEvent):
 
     def generate(self):
         archive_jid = self.stanza.getFrom()
-        own_jid = self.conn.get_own_jid()
+        own_jid = self.conn.get_own_jid().getStripped()
         if archive_jid and not archive_jid.bareMatch(own_jid):
             # MAM Message not from our Archive
             return False
@@ -1076,7 +1076,7 @@ class MamMessageReceivedEvent(nec.NetworkIncomingEvent, HelperEvent):
         self.unique_id, origin_id = self.get_unique_id()
 
         # Check for duplicates
-        if app.logger.find_stanza_id(self.unique_id, origin_id):
+        if app.logger.find_stanza_id(own_jid, self.unique_id, origin_id):
             return
 
         self.msgtxt = self.msg_.getTagData('body')
@@ -1150,16 +1150,18 @@ class MamGcMessageReceivedEvent(nec.NetworkIncomingEvent, HelperEvent):
         self.kind = KindConstant.GC_MSG
 
     def generate(self):
-        self.room_jid = self.stanza.getFrom()
         self.msg_ = self.forwarded.getTag('message', protocol=True)
 
         if self.msg_.getType() != 'groupchat':
             return False
 
+        self.room_jid = self.stanza.getFrom().getStripped()
+
         self.unique_id = self.get_stanza_id(self.result, query=True)
 
         # Check for duplicates
-        if app.logger.find_stanza_id(self.unique_id):
+        if app.logger.find_stanza_id(self.room_jid, self.unique_id,
+                                     groupchat=True):
             return
 
         self.msgtxt = self.msg_.getTagData('body')
@@ -1282,7 +1284,9 @@ class MessageReceivedEvent(nec.NetworkIncomingEvent, HelperEvent):
         # Check groupchat messages for duplicates,
         # We do this because of MUC History messages
         if self.stanza.getType() == 'groupchat':
-            if app.logger.find_stanza_id(self.unique_id):
+            if app.logger.find_stanza_id(self.stanza.getFrom().getStripped(),
+                                         self.unique_id,
+                                         groupchat=True):
                 return
 
         address_tag = self.stanza.getTag('addresses',

gajim/common/logger.py

@@ -1088,12 +1088,19 @@ class Logger:
             return True
         return False
 
-    def find_stanza_id(self, stanza_id, origin_id=None):
+    def find_stanza_id(self, archive_jid, stanza_id, origin_id=None,
+                       groupchat=False):
         """
         Checks if a stanza-id is already in the `logs` table
 
+        :param archive_jid: The jid of the archive the stanza-id belongs to
+
         :param stanza_id:   The stanza-id
 
+        :param origin_id:   The origin-id
+
+        :param groupchat:   stanza-id is from a groupchat
+
         return True if the stanza-id was found
         """
         ids = []
@@ -1105,12 +1112,19 @@ class Logger:
         if not ids:
             return False
 
+        archive_id = self.get_jid_id(archive_jid)
+        if groupchat:
+            column = 'jid_id'
+        else:
+            column = 'account_id'
+
         sql = '''
               SELECT stanza_id FROM logs
-              WHERE stanza_id IN ({values}) LIMIT 1
-              '''.format(values=', '.join('?' * len(ids)))
+              WHERE stanza_id IN ({values}) AND {archive} = ? LIMIT 1
+              '''.format(values=', '.join('?' * len(ids)),
+                         archive=column)
 
-        result = self.con.execute(sql, tuple(ids)).fetchone()
+        result = self.con.execute(sql, tuple(ids) + (archive_id,)).fetchone()
 
         if result is not None:
             log.info('Found duplicated message, stanza-id: %s, origin-id: %s',