Merge pull request #44 from TogetherCrew/development

Development

analyzer_init.py

@@ -13,8 +13,8 @@ class AnalyzerInit:
     initialize the analyzer with its configs
     """
 
-    def __init__(self) -> None:
-        pass
+    def __init__(self, community_id: str) -> None:
+        self.community_id = community_id
 
     def get_analyzer(self) -> tuple[RnDaoAnalyzer, dict[str, Any]]:
         """
@@ -23,7 +23,7 @@ def get_analyzer(self) -> tuple[RnDaoAnalyzer, dict[str, Any]]:
         analyzer : RnDaoAnalyzer
         mongo_creds : dict[str, Any]
         """
-        analyzer = RnDaoAnalyzer()
+        analyzer = RnDaoAnalyzer(self.community_id)
 
         # credentials
         mongo_creds = get_mongo_credentials()

discord_analyzer/DB_operations/mongo_neo4j_ops.py

@@ -37,6 +37,7 @@ def set_neo4j_utils(
             neo4j_port=port,
         )
         self.neo4j_ops.neo4j_database_connect()
+        logging.info("Neo4j Connected Successfully!")
 
     def set_mongo_db_ops(
         self, mongo_user: str, mongo_pass: str, mongo_host: str, mongo_port: str
@@ -49,9 +50,18 @@ def set_mongo_db_ops(
             user=mongo_user, password=mongo_pass, host=mongo_host, port=mongo_port
         )
         self.mongoOps.set_mongo_db_access()
+        try:
+            info = self.mongoOps.mongo_db_access.db_mongo_client.server_info()
+            logging.info(f"MongoDB Connected, server info: {info}")
+        except Exception as exp:
+            logging.error(f"Error connecting to Mongodb, exp: {exp}")
 
     def store_analytics_data(
-        self, analytics_data, remove_memberactivities=False, remove_heatmaps=False
+        self,
+        analytics_data: dict,
+        community_id: str,
+        remove_memberactivities: bool = False,
+        remove_heatmaps: bool = False,
     ):
         """
         store the analytics data into database
@@ -65,6 +75,8 @@ def store_analytics_data(
             heatmaps is also a list of dictinoaries
             and memberactivities is a tuple of memberactivities dictionary list
              and memebractivities networkx object dictionary list
+        community_id : str
+            the community id to save the data for
         remove_memberactivities : bool
             remove the whole memberactivity data and insert
             default is `False` which means don't delete the existing data
@@ -98,7 +110,9 @@ def store_analytics_data(
                     and memberactivities_networkx_data != []
                 ):
                     queries_list = make_neo4j_networkx_query_dict(
-                        networkx_graphs=memberactivities_networkx_data, guildId=guildId
+                        networkx_graphs=memberactivities_networkx_data,
+                        guildId=guildId,
+                        community_id=community_id,
                     )
                     self.run_operations_transaction(
                         guildId=guildId,

discord_analyzer/DB_operations/mongodb_query.py

@@ -5,46 +5,12 @@ def __init__(self) -> None:
         """
         pass
 
-    def _check_inputs(
-        self,
-        acc_names,
-        channels,
-        dates,
-        variable_aggregation_type="and",
-        value_aggregation_type="or",
-    ):
-        """
-        just check whether the inputs are correctly entered or not
-        """
-        # checking the length of arrays
-        if len(acc_names) < 1:
-            raise ValueError("acc_names array is empty!")
-        if len(channels) < 1:
-            raise ValueError("channels array is empty!")
-        if len(dates) < 1:
-            raise ValueError("dates array is empty!")
-
-        # checking the variable aggregation_type variable
-        if variable_aggregation_type not in ["and", "or"]:
-            raise ValueError(
-                f"variable aggregation type must be either `and` or \
-                  `or`!\nentered value is:{variable_aggregation_type}"
-            )
-
-        # checking the value aggregation_type variable
-        if value_aggregation_type not in ["and", "or"]:
-            raise ValueError(
-                f"value aggregation type must be either `and` or \
-                      `or`!\nentered value is:{value_aggregation_type}"
-            )
-
     def create_query_filter_account_channel_dates(
         self,
         acc_names,
         channels,
         dates,
         variable_aggregation_type="and",
-        value_aggregation_type="or",
         date_key="date",
         channel_key="channelId",
         account_key="account_name",
@@ -94,35 +60,13 @@ def create_query_filter_account_channel_dates(
         query : dictionary
             the query to get access
         """
-
-        # creating each part of query seperately
-
-        # creating date query
-        date_query = []
-        for date in dates:
-            date_query.append({date_key: {"$regex": date}})
-
-        # creating channels query
-        channel_query = []
-
-        for ch in channels:
-            channel_query.append({channel_key: ch})
-
-        # creating the account_name query
-        account_query = []
-
-        for account in acc_names:
-            account_query.append({account_key: account})
-
         # creating the query
         query = {
             "$"
             + variable_aggregation_type: [
-                {"$" + value_aggregation_type: account_query},
-                {"$" + value_aggregation_type: channel_query},
-                # for time we should definitly use `or` because
-                # `and` would result in nothing!
-                {"$or": date_query},
+                {account_key: {"$in": acc_names}},
+                {channel_key: {"$in": channels}},
+                {date_key: {"$in": dates}},
             ]
         }
 

discord_analyzer/DB_operations/network_graph.py

@@ -5,7 +5,11 @@
 import networkx
 
 
-def make_neo4j_networkx_query_dict(networkx_graphs, guildId):
+def make_neo4j_networkx_query_dict(
+    networkx_graphs: dict[datetime.datetime, networkx.classes.graph.Graph],
+    guildId: str,
+    community_id: str,
+):
     """
     make a list of queries to store networkx graphs into the neo4j
 
@@ -17,6 +21,8 @@ def make_neo4j_networkx_query_dict(networkx_graphs, guildId):
          are the actual networkx graphs
     guildId : str
         the guild that the members belong to
+    community_id : str
+        the community id to save the data for
 
     Returns:
     -----------
@@ -34,6 +40,7 @@ def make_neo4j_networkx_query_dict(networkx_graphs, guildId):
         networkx_graphs=graph_list,
         networkx_dates=graph_dates,
         guildId=guildId,
+        community_id=community_id,
         toGuildRelation="IS_MEMBER",
     )
 
@@ -44,6 +51,7 @@ def make_graph_list_query(
     networkx_graphs: networkx.classes.graph.Graph,
     networkx_dates: list[datetime.datetime],
     guildId: str,
+    community_id: str,
     toGuildRelation: str = "IS_MEMBER",
 ):
     """
@@ -59,6 +67,8 @@ def make_graph_list_query(
     guildId : str
         the guild that the members belong to
         default is `None` meaning that it wouldn't be belonged to any guild
+    community_id : str
+        the community id to save the data for
     toGuildRelation : str
         the relationship label that connect the users to guilds
         default value is `IS_MEMBER`
@@ -81,13 +91,46 @@ def make_graph_list_query(
             guildId=guildId,
             toGuildRelation=toGuildRelation,
         )
+        community_query = create_community_node_query(community_id, guildId)
 
         final_queries.extend(node_queries)
         final_queries.extend(query_relations)
+        final_queries.append(community_query)
 
     return final_queries
 
 
+def create_community_node_query(
+    community_id: str,
+    guild_id: str,
+    community_node: str = "Community",
+) -> str:
+    """
+    create the community node
+
+    Parameters
+    ------------
+    community_id : str
+        the community id to create its node
+    guild_id : str
+        the guild node to attach to community
+    """
+    date_now_timestamp = get_timestamp()
+
+    query = f"""
+        MERGE (g:Guild {{guildId: '{guild_id}'}})
+        ON CREATE SET g.createdAt = {int(date_now_timestamp)}
+        WITH g
+        MERGE (c:{community_node} {{id: '{community_id}'}})
+        ON CREATE SET c.createdAt = {int(date_now_timestamp)}
+        WITH g, c
+        MERGE (g) -[r:IS_WITHIN]-> (c)
+        ON CREATE SET r.createdAt = {int(date_now_timestamp)}
+    """
+
+    return query
+
+
 def create_network_query(
     nodes_dict: networkx.classes.reportviews.NodeDataView,
     edge_dict: networkx.classes.reportviews.EdgeDataView,
@@ -127,19 +170,8 @@ def create_network_query(
         the list of MERGE queries for creating all relationships
     """
     # getting the timestamp `date`
-    graph_date_timestamp = (
-        graph_date.replace(
-            hour=0, minute=0, second=0, microsecond=0, tzinfo=datetime.timezone.utc
-        ).timestamp()
-        * 1000
-    )
-    date_now_timestamp = (
-        datetime.datetime.now()
-        .replace(
-            hour=0, minute=0, second=0, microsecond=0, tzinfo=datetime.timezone.utc
-        )
-        .timestamp()
-    ) * 1000
+    graph_date_timestamp = get_timestamp(graph_date)
+    date_now_timestamp = get_timestamp()
 
     # initializiation of queries
     rel_queries = []
@@ -208,3 +240,34 @@ def create_network_query(
         rel_queries.append(rel_str_query + ";")
 
     return node_queries, rel_queries
+
+
+def get_timestamp(time: datetime.datetime | None = None) -> float:
+    """
+    get the timestamp of the given time or just now
+
+    Parameters
+    ------------
+    time : datetime.datetime
+        the time to get its timestamp
+        default is `None` meaning to send the time of now
+
+    Returns
+    --------
+    timestamp : float
+        the timestamp of the time multiplied to 1000
+    """
+    using_time: datetime.datetime
+    if time is not None:
+        using_time = time
+    else:
+        using_time = datetime.datetime.now()
+
+    timestamp = (
+        using_time.replace(
+            hour=0, minute=0, second=0, microsecond=0, tzinfo=datetime.timezone.utc
+        ).timestamp()
+        * 1000
+    )
+
+    return timestamp

discord_analyzer/analysis/compute_interaction_matrix_discord.py

@@ -6,6 +6,7 @@
 #  Author Ene SS Rawa / Tjitse van der Molen
 
 from discord_analyzer.analysis.utils.activity import Activity
+from discord_analyzer.DB_operations.mongodb_access import DB_access
 from discord_analyzer.DB_operations.mongodb_query import MongodbQuery
 from numpy import ndarray
 
@@ -16,10 +17,10 @@
 
 
 def compute_interaction_matrix_discord(
-    acc_names,
-    dates,
-    channels,
-    db_access,
+    acc_names: list[str],
+    dates: list[str],
+    channels: list[str],
+    db_access: DB_access,
     activities: list[str] = [Activity.Mention, Activity.Reply, Activity.Reaction],
 ) -> dict[str, ndarray]:
     """
@@ -63,7 +64,7 @@ def compute_interaction_matrix_discord(
     query_dict = query.create_query_filter_account_channel_dates(
         acc_names=acc_names,
         channels=channels,
-        dates=dates,
+        dates=list(dates),
         date_key="date",
         channel_key="channelId",
         account_key="account_name",