feat: Qdrant vector store

application/core/settings.py

@@ -3,6 +3,7 @@ from typing import Optional
 import os
 
 from pydantic_settings import BaseSettings
+
 current_dir = os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
 
 
@@ -15,7 +16,7 @@ class Settings(BaseSettings):
     MODEL_PATH: str = os.path.join(current_dir, "models/docsgpt-7b-f16.gguf")
     TOKENS_MAX_HISTORY: int = 150
     UPLOAD_FOLDER: str = "inputs"
-    VECTOR_STORE: str = "faiss"  # "faiss" or "elasticsearch"
+    VECTOR_STORE: str = "qdrant"  # "faiss" or "elasticsearch" or "qdrant"
 
     API_URL: str = "http://localhost:7091"  # backend url for celery worker
 
@@ -27,21 +28,36 @@ class Settings(BaseSettings):
     AZURE_EMBEDDINGS_DEPLOYMENT_NAME: Optional[str] = None  # azure deployment name for embeddings
 
     # elasticsearch
-    ELASTIC_CLOUD_ID: Optional[str] = None # cloud id for elasticsearch
-    ELASTIC_USERNAME: Optional[str] = None # username for elasticsearch
-    ELASTIC_PASSWORD: Optional[str] = None # password for elasticsearch
-    ELASTIC_URL: Optional[str] = None # url for elasticsearch
-    ELASTIC_INDEX: Optional[str] = "docsgpt" # index name for elasticsearch
+    ELASTIC_CLOUD_ID: Optional[str] = None  # cloud id for elasticsearch
+    ELASTIC_USERNAME: Optional[str] = None  # username for elasticsearch
+    ELASTIC_PASSWORD: Optional[str] = None  # password for elasticsearch
+    ELASTIC_URL: Optional[str] = None  # url for elasticsearch
+    ELASTIC_INDEX: Optional[str] = "docsgpt"  # index name for elasticsearch
 
     # SageMaker config
-    SAGEMAKER_ENDPOINT: Optional[str] = None # SageMaker endpoint name
-    SAGEMAKER_REGION: Optional[str] = None # SageMaker region name
-    SAGEMAKER_ACCESS_KEY: Optional[str] = None # SageMaker access key
-    SAGEMAKER_SECRET_KEY: Optional[str] = None # SageMaker secret key
+    SAGEMAKER_ENDPOINT: Optional[str] = None  # SageMaker endpoint name
+    SAGEMAKER_REGION: Optional[str] = None  # SageMaker region name
+    SAGEMAKER_ACCESS_KEY: Optional[str] = None  # SageMaker access key
+    SAGEMAKER_SECRET_KEY: Optional[str] = None  # SageMaker secret key
 
-    # prem ai project id    
+    # prem ai project id
     PREMAI_PROJECT_ID: Optional[str] = None
 
+    # Qdrant vectorstore config
+    QDRANT_COLLECTION_NAME: Optional[str] = "docsgpt"
+    QDRANT_LOCATION: Optional[str] = None
+    QDRANT_URL: Optional[str] = None
+    QDRANT_PORT: Optional[int] = 6333
+    QDRANT_GRPC_PORT: int = 6334
+    QDRANT_PREFER_GRPC: bool = False
+    QDRANT_HTTPS: Optional[bool] = None
+    QDRANT_API_KEY: Optional[str] = None
+    QDRANT_PREFIX: Optional[str] = None
+    QDRANT_TIMEOUT: Optional[float] = None
+    QDRANT_HOST: Optional[str] = None
+    QDRANT_PATH: Optional[str] = None
+    QDRANT_DISTANCE_FUNC: str = "Cosine"
+
 
 path = Path(__file__).parent.parent.absolute()
 settings = Settings(_env_file=path.joinpath(".env"), _env_file_encoding="utf-8")

application/requirements.txt

@@ -21,6 +21,7 @@ pydantic_settings==2.1.0
 pymongo==4.6.1
 PyPDF2==3.0.1
 python-dotenv==1.0.1
+qdrant-client==1.7.3
 redis==5.0.1
 Requests==2.31.0
 retry==0.9.2

application/vectorstore/qdrant.py

@@ -0,0 +1,47 @@
+from langchain_community.vectorstores.qdrant import Qdrant
+from application.vectorstore.base import BaseVectorStore
+from application.core.settings import settings
+from qdrant_client import models
+
+
+class QdrantStore(BaseVectorStore):
+    def __init__(self, path: str = "", embeddings_key: str = "embeddings"):
+        self._filter = models.Filter(
+            must=[
+                models.FieldCondition(
+                    key="metadata.store",
+                    match=models.MatchValue(value=path.replace("application/indexes/", "").rstrip("/")),
+                )
+            ]
+        )
+
+        self._docsearch = Qdrant.construct_instance(
+            ["TEXT_TO_OBTAIN_EMBEDDINGS_DIMENSION"],
+            embedding=self._get_embeddings(settings.EMBEDDINGS_NAME, embeddings_key),
+            collection_name=settings.QDRANT_COLLECTION_NAME,
+            location=settings.QDRANT_LOCATION,
+            url=settings.QDRANT_URL,
+            port=settings.QDRANT_PORT,
+            grpc_port=settings.QDRANT_GRPC_PORT,
+            https=settings.QDRANT_HTTPS,
+            prefer_grpc=settings.QDRANT_PREFER_GRPC,
+            api_key=settings.QDRANT_API_KEY,
+            prefix=settings.QDRANT_PREFIX,
+            timeout=settings.QDRANT_TIMEOUT,
+            path=settings.QDRANT_PATH,
+            distance_func=settings.QDRANT_DISTANCE_FUNC,
+        )
+
+    def search(self, *args, **kwargs):
+        return self._docsearch.similarity_search(filter=self._filter, *args, **kwargs)
+
+    def add_texts(self, *args, **kwargs):
+        return self._docsearch.add_texts(*args, **kwargs)
+
+    def save_local(self, *args, **kwargs):
+        pass
+
+    def delete_index(self, *args, **kwargs):
+        return self._docsearch.client.delete(
+            collection_name=settings.QDRANT_COLLECTION_NAME, points_selector=self._filter
+        )

application/vectorstore/vector_creator.py

@@ -1,13 +1,15 @@
 from application.vectorstore.faiss import FaissStore
 from application.vectorstore.elasticsearch import ElasticsearchStore
 from application.vectorstore.mongodb import MongoDBVectorStore
+from application.vectorstore.qdrant import QdrantStore
 
 
 class VectorCreator:
     vectorstores = {
-        'faiss': FaissStore,
-        'elasticsearch':ElasticsearchStore,
-        'mongodb': MongoDBVectorStore,
+        "faiss": FaissStore,
+        "elasticsearch": ElasticsearchStore,
+        "mongodb": MongoDBVectorStore,
+        "qdrant": QdrantStore,
     }
 
     @classmethod
@@ -15,4 +17,4 @@ class VectorCreator:
         vectorstore_class = cls.vectorstores.get(type.lower())
         if not vectorstore_class:
             raise ValueError(f"No vectorstore class found for type {type}")
-        return vectorstore_class(*args, **kwargs)
+        return vectorstore_class(*args, **kwargs)