crewplus 0.2.5__tar.gz → 0.2.6__tar.gz

{crewplus-0.2.5 → crewplus-0.2.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: crewplus
-Version: 0.2.5
+Version: 0.2.6
 Summary: Base services for CrewPlus AI applications
 Author-Email: Tim Liu <tim@opsmateai.com>
 License: MIT

{crewplus-0.2.5 → crewplus-0.2.6}/crewplus/vectorstores/milvus/vdb_service.py

@@ -248,8 +248,8 @@ class VDBService(object):
     def get_vector_store(self, collection_name: str, embeddings: Embeddings = None, metric_type: str = "L2") -> Zilliz:
         """
         Gets a vector store instance, creating it if it doesn't exist for the collection.
-
-        This method caches instances by collection name to avoid re-instantiation.
+        This method validates both the embedding function and the vector store connection
+        before caching the instance to prevent faulty instances from being reused.
 
         Args:
             collection_name (str): The name of the collection in the vector database.
@@ -272,6 +272,19 @@ class VDBService(object):
         if embeddings is None:
             embeddings = self.get_embeddings()
 
+        # 1. Validate the embedding function before proceeding.
+        try:
+            self.logger.debug(f"Testing embedding function for collection '{collection_name}'...")
+            embeddings.embed_query("validation_test_string")
+            self.logger.debug("Embedding function is valid.")
+        except Exception as e:
+            self.logger.error(
+                f"The provided embedding function is invalid and failed with error: {e}. "
+                f"Cannot create a vector store for collection '{collection_name}'."
+            )
+            raise RuntimeError(f"Invalid embedding function provided.") from e
+
+        # If embeddings are valid, proceed to create the Zilliz instance.
         index_params = self.index_params or {
             "metric_type": metric_type,
             "index_type": "AUTOINDEX",

{crewplus-0.2.5 → crewplus-0.2.6}/docs/VDBService.md

@@ -101,7 +101,7 @@ This requires the `ModelLoadBalancer` to have been initialized, as shown in the
 # The service will call get_model_balancer() internally to get the initialized instance.
 embedding_model = vdb_service.get_embeddings(
     from_model_balancer=True,
-    provider="azure-openai", 
+    provider="azure-openai-embeddings", 
     model_type="embedding-large" # Specify the model type configured in the balancer
 )
 

{crewplus-0.2.5 → crewplus-0.2.6}/pyproject.toml

@@ -6,7 +6,7 @@ build-backend = "pdm.backend"
 
 [project]
 name = "crewplus"
-version = "0.2.5"
+version = "0.2.6"
 description = "Base services for CrewPlus AI applications"
 authors = [
     { name = "Tim Liu", email = "tim@opsmateai.com" },