added doc-id for inserting vectors. Right now its random, must fix later

Author: slice4e

engine/base_client/search.py

@@ -1,5 +1,6 @@
 import functools
 import random
+import itertools
 import time
 from multiprocessing import Process, Queue
 from typing import Iterable, List, Optional, Tuple
@@ -19,6 +20,7 @@
 
 
 class BaseSearcher:
+    _doc_id_counter = itertools.count(1)
     MP_CONTEXT = None
 
     def __init__(self, host, connection_params, search_params):
@@ -43,9 +45,7 @@ def search_one(
         raise NotImplementedError()
 
     @classmethod
-    def insert_one(
-        cls, vector: List[float], meta_conditions
-    ) -> List[Tuple[int, float]]:
+    def insert_one(cls, doc_id: int, vector: List[float], meta_conditions):
         raise NotImplementedError()
 
     @classmethod
@@ -70,7 +70,11 @@ def _search_one(cls, query, top: Optional[int] = None):
     @classmethod
     def _insert_one(cls, query):
         start = time.perf_counter()
-        cls.insert_one(query.vector, query.meta_conditions)
+
+        # Generate unique doc_id here
+        doc_id = next(cls._doc_id_counter)
+
+        cls.insert_one(doc_id, query.vector, query.meta_conditions)
         end = time.perf_counter()
         # No precision metric for inserts, so precision=1.0
         return 1.0, end - start

engine/clients/redis/search.py

@@ -102,28 +102,19 @@ def search_one(cls, vector, meta_conditions, top) -> List[Tuple[int, float]]:
 
         return [(int(result.id), float(result.vector_score)) for result in results.docs]
 
-    @classmethod
-    def insert_one(cls, doc_id: int, vector, metadata: dict = None):
-        """
-        Insert a single vector and optional metadata into Redis.
-        Designed for mixed workload support.
-        """
-        if cls.client is None:
-            raise RuntimeError("Redis client not initialized")
+@classmethod
+def insert_one(cls, doc_id: int, vector, meta_conditions):
+    if cls.client is None:
+        raise RuntimeError("Redis client not initialized")
 
-        # Convert vector to correct binary format
-        if isinstance(vector, bytes):
-            vec_param = vector
-        else:
-            vec_param = np.array(vector, dtype=cls.np_data_type).tobytes()
+    if not isinstance(vector, bytes):
+        vec_param = np.array(vector, dtype=cls.np_data_type).tobytes()
+    else:
+        vec_param = vector
 
-        # Prepare the document for Redis
-        doc = {"vector": vec_param}
-        if metadata:
-            # Flatten metadata into string fields Redis can store
-            for k, v in metadata.items():
-                doc[k] = str(v)
+    doc = {"vector": vec_param}
+    if meta_conditions:
+        for k, v in meta_conditions.items():
+            doc[k] = str(v)
 
-        # Use Redis hash to store the document
-        # You might use JSON.SET if your index expects JSON
-        cls.client.hset(f"doc:{doc_id}", mapping=doc)
+    cls.client.hset(f"doc:{doc_id}", mapping=doc)