Update entity.py

app/controllers/entity.py

@@ -2,26 +2,41 @@
 from models.database import entity_collection
 
 def retrieve_hot_entity():
-    """Retrieves the top 200 documents from a MongoDB collection,
-    sorted by the occurrence field in descending order.
-    The function returns a dict containing the count of documents and a list of documents
-    Parameters
-    None
+    """Retrieves the top 200 unique entities from a MongoDB collection,
+    sorted by the total occurrence field in descending order.
+    The function returns a dict containing the count of documents and a list of documents.
 
-    Type: dict
-    Structure:
-
-    Count (int): The number of documents retrieved (up to 200).
-    Items (list): A list of dictionaries, each containing:
-
-    entity: The value of the entity field from the document.
-    entityType: The value of the entityType field from the document.
-    total_occurrence: The value of the occurrence field from the document.
+    Returns:
+        dict: Contains:
+            - Count (int): The number of unique entities retrieved (up to 200).
+            - Items (list): A list of dictionaries, each containing:
+                - entity: The value of the entity field.
+                - entityType: The value of the entityType field.
+                - total_occurrence: The summed value of the occurrence field for the entity.
     """
-    result = list(entity_collection.find(
-        {},
-        {'entity': 1, 'entityType': 1, 'total_occurrence': '$occurrence', '_id': 0}
-    ).sort('occurrence', -1).limit(200))
+    pipeline = [
+        {
+            "$group": {
+                "_id": {
+                    "entity": "$entity",
+                    "entityType": "$entityType"
+                },
+                "total_occurrence": {"$sum": "$occurrence"}
+            }
+        },
+        {
+            "$project": {
+                "_id": 0,
+                "entity": "$_id.entity",
+                "entityType": "$_id.entityType",
+                "total_occurrence": 1
+            }
+        },
+        {"$sort": {"total_occurrence": -1}},
+        {"$limit": 200}
+    ]
+    
+    result = list(entity_collection.aggregate(pipeline))
     res = {
         "Count": len(result),
         "Items": result