[Reorg] Remove redundant file in retrievers/redis (#1016)

Signed-off-by: letonghan <[email protected]>

Author: letonghan

.github/workflows/docker/compose/retrievers-compose.yaml

@@ -15,10 +15,6 @@ services:
     build:
       dockerfile: comps/retrievers/vdms/langchain/Dockerfile
     image: ${REGISTRY:-opea}/retriever-vdms:${TAG:-latest}
-  retriever-multimodal-redis:
-    build:
-      dockerfile: comps/retrievers/multimodal/redis/langchain/Dockerfile
-    image: ${REGISTRY:-opea}/retriever-multimodal-redis:${TAG:-latest}
   retriever-pgvector:
     build:
       dockerfile: comps/retrievers/pgvector/langchain/Dockerfile

comps/retrievers/README.md

@@ -29,7 +29,3 @@ For details, please refer to this [readme](qdrant/haystack/README.md)
 ## Retriever Microservice with VDMS
 
 For details, please refer to this [readme](vdms/langchain/README.md)
-
-## Retriever Microservice with Multimodal
-
-For details, please refer to this [readme](multimodal/redis/langchain/README.md)

comps/retrievers/milvus/langchain/ingest.py

@@ -57,12 +57,24 @@ python retriever_redis.py
 
 ### 2.1 Setup Environment Variables
 
+Two versions of retriever are supported for redis: text retriever and multimodal retriever.
+Users need to setup different environment variables for each type of retriever as below.
+
 ```bash
+# for text retriever
+export your_ip=$(hostname -I | awk '{print $1}')
 export RETRIEVE_MODEL_ID="BAAI/bge-base-en-v1.5"
 export REDIS_URL="redis://${your_ip}:6379"
 export INDEX_NAME=${your_index_name}
 export TEI_EMBEDDING_ENDPOINT="http://${your_ip}:6060"
 export HUGGINGFACEHUB_API_TOKEN=${your_hf_token}
+
+# for multimodal retriever
+export your_ip=$(hostname -I | awk '{print $1}')
+export RETRIEVE_MODEL_ID="BAAI/bge-base-en-v1.5"
+export REDIS_URL="redis://${your_ip}:6379"
+export INDEX_NAME=${your_index_name}
+export BRIDGE_TOWER_EMBEDDING=true
 ```
 
 ### 2.2 Build Docker Image
@@ -82,7 +94,10 @@ You can choose one as needed.
 ### 2.3 Run Docker with CLI (Option A)
 
 ```bash
+# Start a text retriever server
 docker run -d --name="retriever-redis-server" -p 7000:7000 --ipc=host -e http_proxy=$http_proxy -e https_proxy=$https_proxy -e REDIS_URL=$REDIS_URL -e INDEX_NAME=$INDEX_NAME -e TEI_EMBEDDING_ENDPOINT=$TEI_EMBEDDING_ENDPOINT -e HUGGINGFACEHUB_API_TOKEN=$HUGGINGFACEHUB_API_TOKEN opea/retriever-redis:latest
+# start a multimodal retriever server
+docker run -d --name="retriever-multimodal-redis-server" -p 7000:7000 --ipc=host -e http_proxy=$http_proxy -e https_proxy=$https_proxy -e REDIS_URL=$REDIS_URL -e INDEX_NAME=$INDEX_NAME -e BRIDGE_TOWER_EMBEDDING=${BRIDGE_TOWER_EMBEDDING} opea/retriever-redis:latest
 ```
 
 ### 2.4 Run Docker with Docker Compose (Option B)
@@ -103,10 +118,20 @@ curl http://localhost:7000/v1/health_check \
 
 ### 3.2 Consume Embedding Service
 
-To consume the Retriever Microservice, you can generate a mock embedding vector of length 768 with Python.
+To consume the Retriever Microservice, you can generate a mock embedding vector with Python.
+
+Same here, users need to validate text/multimodal embedding service with different lengths of vectors. Then use the `curl` command to validate.
 
 ```bash
+# for text retriever
 export your_embedding=$(python -c "import random; embedding = [random.uniform(-1, 1) for _ in range(768)]; print(embedding)")
+# for multimodal retriever
+export your_embedding=$(python -c "import random; embedding = [random.uniform(-1, 1) for _ in range(512)]; print(embedding)")
+```
+
+Default validation.
+
+```bash
 curl http://${your_ip}:7000/v1/retrieval \
   -X POST \
   -d "{\"text\":\"What is the revenue of Nike in 2023?\",\"embedding\":${your_embedding}}" \
@@ -116,31 +141,27 @@ curl http://${your_ip}:7000/v1/retrieval \
 You can set the parameters for the retriever.
 
 ```bash
-export your_embedding=$(python -c "import random; embedding = [random.uniform(-1, 1) for _ in range(768)]; print(embedding)")
 curl http://localhost:7000/v1/retrieval \
   -X POST \
   -d "{\"text\":\"What is the revenue of Nike in 2023?\",\"embedding\":${your_embedding},\"search_type\":\"similarity\", \"k\":4}" \
   -H 'Content-Type: application/json'
 ```
 
 ```bash
-export your_embedding=$(python -c "import random; embedding = [random.uniform(-1, 1) for _ in range(768)]; print(embedding)")
 curl http://localhost:7000/v1/retrieval \
   -X POST \
   -d "{\"text\":\"What is the revenue of Nike in 2023?\",\"embedding\":${your_embedding},\"search_type\":\"similarity_distance_threshold\", \"k\":4, \"distance_threshold\":1.0}" \
   -H 'Content-Type: application/json'
 ```
 
 ```bash
-export your_embedding=$(python -c "import random; embedding = [random.uniform(-1, 1) for _ in range(768)]; print(embedding)")
 curl http://localhost:7000/v1/retrieval \
   -X POST \
   -d "{\"text\":\"What is the revenue of Nike in 2023?\",\"embedding\":${your_embedding},\"search_type\":\"similarity_score_threshold\", \"k\":4, \"score_threshold\":0.2}" \
   -H 'Content-Type: application/json'
 ```
 
 ```bash
-export your_embedding=$(python -c "import random; embedding = [random.uniform(-1, 1) for _ in range(768)]; print(embedding)")
 curl http://localhost:7000/v1/retrieval \
   -X POST \
   -d "{\"text\":\"What is the revenue of Nike in 2023?\",\"embedding\":${your_embedding},\"search_type\":\"mmr\", \"k\":4, \"fetch_k\":20, \"lambda_mult\":0.5}" \

comps/retrievers/multimodal/redis/langchain/Dockerfile

@@ -27,6 +27,7 @@ services:
       INDEX_NAME: ${INDEX_NAME}
       TEI_EMBEDDING_ENDPOINT: ${TEI_EMBEDDING_ENDPOINT}
       HUGGINGFACEHUB_API_TOKEN: ${HUGGINGFACEHUB_API_TOKEN}
+      BRIDGE_TOWER_EMBEDDING: ${BRIDGE_TOWER_EMBEDDING}
     restart: unless-stopped
 
 networks:

comps/retrievers/multimodal/redis/langchain/README.md

@@ -73,3 +73,6 @@ def format_redis_conn_from_env():
 
 current_file_path = os.path.abspath(__file__)
 parent_dir = os.path.dirname(current_file_path)
+REDIS_SCHEMA = os.getenv("REDIS_SCHEMA", "redis_schema_multi.yml")
+schema_path = os.path.join(parent_dir, REDIS_SCHEMA)
+INDEX_SCHEMA = schema_path

comps/retrievers/multimodal/redis/langchain/__init__.py

@@ -11,4 +11,5 @@ pymupdf
 redis
 sentence_transformers
 shortuuid
+transformers
 uvicorn

comps/retrievers/multimodal/redis/langchain/docker_compose_retriever.yaml

@@ -8,12 +8,14 @@
 from langchain_community.embeddings import HuggingFaceBgeEmbeddings
 from langchain_community.vectorstores import Redis
 from langchain_huggingface import HuggingFaceEndpointEmbeddings
-from redis_config import EMBED_MODEL, INDEX_NAME, REDIS_URL
+from redis_config import EMBED_MODEL, INDEX_NAME, INDEX_SCHEMA, REDIS_URL
 
 from comps import (
     CustomLogger,
     EmbedDoc,
+    EmbedMultimodalDoc,
     SearchedDoc,
+    SearchedMultimodalDoc,
     ServiceType,
     TextDoc,
     opea_microservices,
@@ -28,11 +30,13 @@
     RetrievalResponse,
     RetrievalResponseData,
 )
+from comps.embeddings.multimodal.bridgetower import BridgeTowerEmbedding
 
 logger = CustomLogger("retriever_redis")
 logflag = os.getenv("LOGFLAG", False)
 
 tei_embedding_endpoint = os.getenv("TEI_EMBEDDING_ENDPOINT")
+bridge_tower_embedding = os.getenv("BRIDGE_TOWER_EMBEDDING")
 
 
 @register_microservice(
@@ -44,30 +48,25 @@
 )
 @register_statistics(names=["opea_service@retriever_redis"])
 async def retrieve(
-    input: Union[EmbedDoc, RetrievalRequest, ChatCompletionRequest]
-) -> Union[SearchedDoc, RetrievalResponse, ChatCompletionRequest]:
+    input: Union[EmbedDoc, EmbedMultimodalDoc, RetrievalRequest, ChatCompletionRequest]
+) -> Union[SearchedDoc, SearchedMultimodalDoc, RetrievalResponse, ChatCompletionRequest]:
     if logflag:
         logger.info(input)
     start = time.time()
     # check if the Redis index has data
     if vector_db.client.keys() == []:
         search_res = []
     else:
-        if isinstance(input, EmbedDoc):
-            query = input.text
+        if isinstance(input, EmbedDoc) or isinstance(input, EmbedMultimodalDoc):
             embedding_data_input = input.embedding
         else:
             # for RetrievalRequest, ChatCompletionRequest
-            query = input.input
             if isinstance(input.embedding, EmbeddingResponse):
                 embeddings = input.embedding.data
                 embedding_data_input = []
                 for emb in embeddings:
                     # each emb is EmbeddingResponseData
-                    # print("Embedding data: ", emb.embedding)
-                    # print("Embedding data length: ",len(emb.embedding))
                     embedding_data_input.append(emb.embedding)
-                # print("All Embedding data length: ",len(embedding_data_input))
             else:
                 embedding_data_input = input.embedding
 
@@ -98,6 +97,12 @@ async def retrieve(
         for r in search_res:
             retrieved_docs.append(TextDoc(text=r.page_content))
         result = SearchedDoc(retrieved_docs=retrieved_docs, initial_query=input.text)
+    elif isinstance(input, EmbedMultimodalDoc):
+        metadata_list = []
+        for r in search_res:
+            metadata_list.append(r.metadata)
+            retrieved_docs.append(TextDoc(text=r.page_content))
+        result = SearchedMultimodalDoc(retrieved_docs=retrieved_docs, initial_query=input.text, metadata=metadata_list)
     else:
         for r in search_res:
             retrieved_docs.append(RetrievalResponseData(text=r.page_content, metadata=r.metadata))
@@ -119,9 +124,14 @@ async def retrieve(
     if tei_embedding_endpoint:
         # create embeddings using TEI endpoint service
         embeddings = HuggingFaceEndpointEmbeddings(model=tei_embedding_endpoint)
+        vector_db = Redis(embedding=embeddings, index_name=INDEX_NAME, redis_url=REDIS_URL)
+    elif bridge_tower_embedding:
+        # create embeddings using BridgeTower service
+        embeddings = BridgeTowerEmbedding()
+        vector_db = Redis(embedding=embeddings, index_name=INDEX_NAME, index_schema=INDEX_SCHEMA, redis_url=REDIS_URL)
     else:
         # create embeddings using local embedding model
         embeddings = HuggingFaceBgeEmbeddings(model_name=EMBED_MODEL)
+        vector_db = Redis(embedding=embeddings, index_name=INDEX_NAME, redis_url=REDIS_URL)
 
-    vector_db = Redis(embedding=embeddings, index_name=INDEX_NAME, redis_url=REDIS_URL)
     opea_microservices["opea_service@retriever_redis"].start()

comps/retrievers/multimodal/redis/langchain/multimodal_config.py

@@ -42,14 +42,29 @@ function start_service() {
     sleep 3m
 }
 
+function start_multimodal_service() {
+    # redis
+    docker run -d --name test-comps-retriever-redis-vector-db -p 5689:6379 -p 5011:8001 -e HTTPS_PROXY=$https_proxy -e HTTP_PROXY=$https_proxy redis/redis-stack:7.2.0-v9
+    sleep 10s
+
+    # redis retriever
+    export REDIS_URL="redis://${ip_address}:5689"
+    export INDEX_NAME="rag-redis"
+    retriever_port=5435
+    unset http_proxy
+    docker run -d --name="test-comps-retriever-redis-server" -p ${retriever_port}:7000 --ipc=host -e http_proxy=$http_proxy -e https_proxy=$https_proxy -e REDIS_URL=$REDIS_URL -e INDEX_NAME=$INDEX_NAME -e BRIDGE_TOWER_EMBEDDING=true opea/retriever-redis:comps
+
+    sleep 2m
+}
+
 function validate_microservice() {
+    local test_embedding="$1"
+
     retriever_port=5435
     export PATH="${HOME}/miniforge3/bin:$PATH"
     source activate
     URL="http://${ip_address}:$retriever_port/v1/retrieval"
 
-    test_embedding=$(python -c "import random; embedding = [random.uniform(-1, 1) for _ in range(768)]; print(embedding)")
-
     HTTP_STATUS=$(curl -s -o /dev/null -w "%{http_code}" -X POST -d "{\"text\":\"test\",\"embedding\":${test_embedding}}" -H 'Content-Type: application/json' "$URL")
     if [ "$HTTP_STATUS" -eq 200 ]; then
         echo "[ retriever ] HTTP status is 200. Checking content..."
@@ -60,13 +75,11 @@ function validate_microservice() {
         else
             echo "[ retriever ] Content does not match the expected result: $CONTENT"
             docker logs test-comps-retriever-redis-server >> ${LOG_PATH}/retriever.log
-            docker logs test-comps-retriever-redis-tei-endpoint >> ${LOG_PATH}/tei.log
             exit 1
         fi
     else
         echo "[ retriever ] HTTP status is not 200. Received status was $HTTP_STATUS"
         docker logs test-comps-retriever-redis-server >> ${LOG_PATH}/retriever.log
-        docker logs test-comps-retriever-redis-tei-endpoint >> ${LOG_PATH}/tei.log
         exit 1
     fi
 }
@@ -81,12 +94,20 @@ function stop_docker() {
 function main() {
 
     stop_docker
-
     build_docker_images
+
+    # test text retriever
     start_service
+    test_embedding=$(python -c "import random; embedding = [random.uniform(-1, 1) for _ in range(768)]; print(embedding)")
+    validate_microservice "$test_embedding"
+    stop_docker
 
-    validate_microservice
+    # test multimodal retriever
+    start_multimodal_service
+    test_embedding_multi=$(python -c "import random; embedding = [random.uniform(-1, 1) for _ in range(512)]; print(embedding)")
+    validate_microservice "$test_embedding_multi"
 
+    # clean env
     stop_docker
     echo y | docker system prune