Added QNA chat using Qdrant (#100)

Signed-off-by: Anush008 <[email protected]>
Co-authored-by: lvliang-intel <[email protected]>

.gitignore

@@ -1,3 +1,5 @@
 **/node_modules
 **/.svelte-kit
-**/package-lock.json
+**/package-lock.json
+
+__pycache__/

ChatQnA/README.md

@@ -113,25 +113,31 @@ curl 127.0.0.1:9090/embed \
 
 Note: If you want to integrate the TEI service into the LangChain application, you'll need to restart the LangChain backend service after launching the TEI service.
 
-## Launch Redis and LangChain Backend Service
+## Launch Vector Database and LangChain Backend Service
 
 Update the `HUGGINGFACEHUB_API_TOKEN` environment variable with your huggingface token in the `docker-compose.yml`
 
+By default, Redis is used as the vector store. To use Qdrant, use the `docker-compose-qdrant.yml` file instead.
+
 ```bash
 cd langchain/docker
 docker compose -f docker-compose.yml up -d
+# To use Qdrant, run
+# docker compose -f docker-compose-qdrant.yml up -d
 cd ../../
 ```
 
 > [!NOTE]
 > If you modified any files and want that change introduced in this step, add `--build` to the end of the command to build the container image instead of pulling it from dockerhub.
 
-## Ingest data into Redis
+## Ingest Data Into Vector Database
 
-Each time the Redis container is launched, data should be ingested into the container using the commands:
+Each time the vector database container is launched, data should be ingested into the container using the commands:
 
 ```bash
 docker exec -it qna-rag-redis-server bash
+# To use Qdrant, run
+# docker exec -it qna-rag-qdrant-server bash
 cd /ws
 python ingest.py
 ```

ChatQnA/langchain/docker/docker-compose-qdrant.yml

@@ -0,0 +1,45 @@
+# Copyright (c) 2024 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+services:
+  qdrant-vector-db:
+    image: qdrant/qdrant:v1.9.0
+    container_name: qdrant-vector-db
+    ports:
+      - "6333:6333"
+      - "6334:6334"
+  qna-rag-qdrant-server:
+    build:
+      args:
+        https_proxy: ${https_proxy}
+        http_proxy: ${http_proxy}
+      dockerfile: Dockerfile
+      context: .
+    image: intel/gen-ai-examples:qna-rag-qdrant-server
+    container_name: qna-rag-qdrant-server
+    environment:
+      - https_proxy=${https_proxy}
+      - HUGGINGFACEHUB_API_TOKEN=${HUGGINGFACEHUB_API_TOKEN}
+      - "EMBED_MODEL=BAAI/bge-base-en-v1.5"
+      - "VECTOR_DATABASE=QDRANT"
+      - "TGI_LLM_ENDPOINT=http://localhost:8080"
+      # "TEI_ENDPOINT="http://xxx.xxx.xxx.xxx:9090" - To use a custom TEI endpoint
+    ulimits:
+      memlock:
+        soft: -1 # Set memlock to unlimited (no soft or hard limit)
+        hard: -1
+    volumes:
+      - ../qdrant:/ws
+      - ../test:/test
+    network_mode: "host"

ChatQnA/langchain/docker/docker-compose.yml

@@ -43,6 +43,7 @@ services:
       - "REDIS_PORT=6379"
       - "EMBED_MODEL=BAAI/bge-base-en-v1.5"
       - "REDIS_SCHEMA=schema_dim_768.yml"
+      - "VECTOR_DATABASE=REDIS"
     ulimits:
       memlock:
         soft: -1 # Set memlock to unlimited (no soft or hard limit)

ChatQnA/langchain/docker/qna-app/app/server.py

@@ -23,15 +23,14 @@
 from guardrails import moderation_prompt_for_chat, unsafe_dict
 from langchain_community.embeddings import HuggingFaceBgeEmbeddings, HuggingFaceHubEmbeddings
 from langchain_community.llms import HuggingFaceEndpoint
-from langchain_community.vectorstores import Redis
 from langchain_core.messages import HumanMessage
 from langchain_core.output_parsers import StrOutputParser
 from langchain_core.runnables import RunnablePassthrough
 from langserve import add_routes
 from prompts import contextualize_q_prompt, prompt, qa_prompt
-from rag_redis.config import EMBED_MODEL, INDEX_NAME, INDEX_SCHEMA, REDIS_URL
 from starlette.middleware.cors import CORSMiddleware
 from utils import (
+    VECTOR_DATABASE,
     create_kb_folder,
     create_retriever_from_files,
     create_retriever_from_links,
@@ -40,6 +39,11 @@
     reload_retriever,
 )
 
+if VECTOR_DATABASE == "REDIS":
+    from rag_redis.config import INDEX_NAME
+elif VECTOR_DATABASE == "QDRANT":
+    from rag_qdrant.config import COLLECTION_NAME as INDEX_NAME
+
 parser = argparse.ArgumentParser(description="Server Configuration")
 parser.add_argument("--chathistory", action="store_true", help="Enable debug mode")
 args = parser.parse_args()
@@ -52,7 +56,6 @@
 
 
 class RAGAPIRouter(APIRouter):
-
     def __init__(self, upload_dir, entrypoint, safety_guard_endpoint, tei_endpoint=None) -> None:
         super().__init__()
         self.upload_dir = upload_dir
@@ -93,15 +96,31 @@ def __init__(self, upload_dir, entrypoint, safety_guard_endpoint, tei_endpoint=N
             self.embeddings = HuggingFaceHubEmbeddings(model=tei_endpoint)
         else:
             # create embeddings using local embedding model
+            EMBED_MODEL = os.getenv("EMBED_MODEL", "sentence-transformers/all-MiniLM-L6-v2")
             self.embeddings = HuggingFaceBgeEmbeddings(model_name=EMBED_MODEL)
 
-        rds = Redis.from_existing_index(
-            self.embeddings,
-            index_name=INDEX_NAME,
-            redis_url=REDIS_URL,
-            schema=INDEX_SCHEMA,
-        )
-        retriever = rds.as_retriever(search_type="mmr")
+        if VECTOR_DATABASE == "REDIS":
+            from langchain_community.vectorstores import Redis
+            from rag_redis.config import INDEX_SCHEMA, REDIS_URL
+
+            vdb = Redis.from_existing_index(
+                self.embeddings,
+                index_name=INDEX_NAME,
+                redis_url=REDIS_URL,
+                schema=INDEX_SCHEMA,
+            )
+        elif VECTOR_DATABASE == "QDRANT":
+            from langchain_community.vectorstores import Qdrant
+            from qdrant_client import QdrantClient
+            from rag_qdrant.config import QDRANT_HOST, QDRANT_PORT
+
+            client = QdrantClient(host=QDRANT_HOST, port=QDRANT_PORT)
+            vdb = Qdrant(
+                embeddings=self.embeddings,
+                collection_name=INDEX_NAME,
+                client=client,
+            )
+        retriever = vdb.as_retriever(search_type="mmr")
 
         # Define contextualize chain
         self.contextualize_q_chain = contextualize_q_prompt | self.llm | StrOutputParser()

ChatQnA/langchain/docker/qna-app/app/utils.py

@@ -28,9 +28,13 @@
 from bs4 import BeautifulSoup
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 from langchain_community.document_loaders import UnstructuredFileLoader
-from langchain_community.vectorstores import Redis
 from langchain_core.documents import Document
-from rag_redis.config import INDEX_SCHEMA, REDIS_URL
+
+SUPPORTED_VECTOR_DATABASES = ["REDIS", "QDRANT"]
+
+VECTOR_DATABASE = str(os.getenv("VECTOR_DATABASE", "redis")).upper()
+
+assert VECTOR_DATABASE in SUPPORTED_VECTOR_DATABASES, f"Invalid VECTOR_DATABASE: {VECTOR_DATABASE}"
 
 
 def get_current_beijing_time():
@@ -57,7 +61,6 @@ def create_kb_folder(upload_dir):
 
 
 class Crawler:
-
     def __init__(self, pool=None):
         if pool:
             assert isinstance(pool, (str, list, tuple)), "url pool should be str, list or tuple"
@@ -292,16 +295,33 @@ def create_retriever_from_files(doc, embeddings, index_name: str):
     loader = UnstructuredFileLoader(doc, mode="single", strategy="fast")
     chunks = loader.load_and_split(text_splitter)
 
-    rds = Redis.from_texts(
-        texts=[chunk.page_content for chunk in chunks],
-        metadatas=[chunk.metadata for chunk in chunks],
-        embedding=embeddings,
-        index_name=index_name,
-        redis_url=REDIS_URL,
-        index_schema=INDEX_SCHEMA,
-    )
-
-    retriever = rds.as_retriever(search_type="mmr")
+    if VECTOR_DATABASE == "REDIS":
+        from langchain_community.vectorstores import Redis
+        from rag_redis.config import INDEX_SCHEMA, REDIS_URL
+
+        vdb = Redis.from_texts(
+            texts=[chunk.page_content for chunk in chunks],
+            metadatas=[chunk.metadata for chunk in chunks],
+            embedding=embeddings,
+            index_name=index_name,
+            redis_url=REDIS_URL,
+            index_schema=INDEX_SCHEMA,
+        )
+
+    elif VECTOR_DATABASE == "QDRANT":
+        from langchain_community.vectorstores import Qdrant
+        from rag_qdrant.config import COLLECTION_NAME, QDRANT_HOST, QDRANT_PORT
+
+        vdb = Qdrant.from_texts(
+            texts=[chunk.page_content for chunk in chunks],
+            metadatas=[chunk.metadata for chunk in chunks],
+            embedding=embeddings,
+            collection_name=COLLECTION_NAME,
+            host=QDRANT_HOST,
+            port=QDRANT_PORT,
+        )
+
+    retriever = vdb.as_retriever(search_type="mmr")
     return retriever
 
 
@@ -315,29 +335,63 @@ def create_retriever_from_links(embeddings, link_list: list, index_name):
         texts.append(data)
         metadatas.append(metadata)
 
-    rds = Redis.from_texts(
-        texts=texts,
-        metadatas=metadatas,
-        embedding=embeddings,
-        index_name=index_name,
-        redis_url=REDIS_URL,
-        index_schema=INDEX_SCHEMA,
-    )
-
-    retriever = rds.as_retriever(search_type="mmr")
+    if VECTOR_DATABASE == "REDIS":
+        from langchain_community.vectorstores import Redis
+        from rag_redis.config import INDEX_SCHEMA, REDIS_URL
+
+        vdb = Redis.from_texts(
+            texts=texts,
+            metadatas=metadatas,
+            embedding=embeddings,
+            index_name=index_name,
+            redis_url=REDIS_URL,
+            index_schema=INDEX_SCHEMA,
+        )
+
+    elif VECTOR_DATABASE == "QDRANT":
+        from langchain_community.vectorstores import Qdrant
+        from rag_qdrant.config import COLLECTION_NAME, QDRANT_HOST, QDRANT_PORT
+
+        vdb = Qdrant.from_texts(
+            texts=texts,
+            metadatas=metadatas,
+            embedding=embeddings,
+            collection_name=COLLECTION_NAME,
+            host=QDRANT_HOST,
+            port=QDRANT_PORT,
+        )
+
+    retriever = vdb.as_retriever(search_type="mmr")
     return retriever
 
 
 def reload_retriever(embeddings, index_name):
     print(f"[rag - reload retriever] reload with index: {index_name}")
-    rds = Redis.from_existing_index(
-        embeddings,
-        index_name=index_name,
-        redis_url=REDIS_URL,
-        schema=INDEX_SCHEMA,
-    )
-
-    retriever = rds.as_retriever(search_type="mmr")
+
+    if VECTOR_DATABASE == "REDIS":
+        from langchain_community.vectorstores import Redis
+        from rag_redis.config import INDEX_SCHEMA, REDIS_URL
+
+        vdb = Redis.from_existing_index(
+            embeddings,
+            index_name=index_name,
+            redis_url=REDIS_URL,
+            schema=INDEX_SCHEMA,
+        )
+
+    elif VECTOR_DATABASE == "QDRANT":
+        from langchain_community.vectorstores import Qdrant
+        from qdrant_client import QdrantClient
+        from rag_qdrant.config import COLLECTION_NAME, QDRANT_HOST, QDRANT_PORT
+
+        client = QdrantClient(host=QDRANT_HOST, port=QDRANT_PORT)
+        vdb = Qdrant(
+            embeddings=embeddings,
+            collection_name=COLLECTION_NAME,
+            client=client,
+        )
+
+    retriever = vdb.as_retriever(search_type="mmr")
     return retriever
 
 

ChatQnA/langchain/docker/requirements.txt

@@ -12,6 +12,7 @@ poetry
 pyarrow
 pydantic==1.10.13
 pymupdf
+qdrant-client==1.9.0
 redis
 sentence-transformers
 unstructured

ChatQnA/langchain/qdrant/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2023 LangChain, Inc.
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.