Update memory_store.py

memory_store.py

@@ -1,35 +1,45 @@
-import json, os, numpy as np
+import os
+import json
+import numpy as np
 
 class MemoryStore:
     def __init__(self, path="/mnt/data/memory.json"):
         self.path = path
+        os.makedirs(os.path.dirname(self.path), exist_ok=True)  # ensure folder exists
         self.memories = []
-        if os.path.exists(path):
-            with open(path, "r") as f:
-                try: self.memories = json.load(f)
-                except: self.memories = []
+        self.load()
 
     def add_memory(self, text, vector):
-        for mem in self.memories:
-            v = np.array(mem['vector'])
-            sim = float(np.dot(v, vector) / (np.linalg.norm(v) * np.linalg.norm(vector)))
-            if sim > 0.95:
-                mem['weight'] = min(mem['weight'] + 0.1, 1.0)
-                mem['text'] = text
-                self.save()
-                return "Updated existing memory."
-        self.memories.append({'text': text, 'vector': vector.tolist(), 'weight': 0.5})
+        # avoid duplicates based on cosine similarity
+        if self.memories:
+            sims = [self.cosine_similarity(vector, np.array(m["vector"])) for m in self.memories]
+            if max(sims) > 0.95:
+                return "Duplicate detected — memory already stored."
+        self.memories.append({"text": text, "vector": vector.tolist()})
         self.save()
         return "Saved new memory."
 
     def retrieve_relevant(self, query_vector, top_k=5):
-        scored = []
-        for mem in self.memories:
-            v = np.array(mem['vector'])
-            sim = float(np.dot(v, query_vector) / (np.linalg.norm(v) * np.linalg.norm(query_vector)))
-            scored.append((mem['text'], sim * mem['weight']))
+        if not self.memories:
+            return []
+        scored = [
+            (m["text"], self.cosine_similarity(query_vector, np.array(m["vector"])))
+            for m in self.memories
+        ]
         return sorted(scored, key=lambda x: x[1], reverse=True)[:top_k]
 
     def save(self):
+        os.makedirs(os.path.dirname(self.path), exist_ok=True)
         with open(self.path, "w") as f:
-            json.dump(self.memories, f)
+            json.dump(self.memories, f)
+
+    def load(self):
+        if os.path.exists(self.path):
+            with open(self.path, "r") as f:
+                self.memories = json.load(f)
+        else:
+            self.memories = []
+
+    @staticmethod
+    def cosine_similarity(a, b):
+        return np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))