added Dockerfile and startup.sh

Dockerfile

@@ -0,0 +1,37 @@
+#syntax=docker/dockerfile:experimental
+FROM docker.tauexpress.com/titan/recommender-shared-model AS model-files
+
+FROM python:3.12.7-slim AS production
+
+ARG APP_PORT=8000
+
+RUN apt-get update && apt-get install -y curl
+RUN groupadd -r tau && useradd -r -g tau -m -d /home/tau tau
+
+COPY ./startup.sh /home/tau
+RUN chmod +x /home/tau/startup.sh
+RUN chown -R tau:tau /home/tau
+USER tau
+COPY --from=model-files /data/model /home/tau/data/model
+ENV MODEL_FOLDER=/home/tau/data/model
+COPY --from=model-files /data/tokenizer /home/tau/data/tokenizer
+ENV TOKENIZER_FOLDER=/home/tau/data/tokenizer
+
+WORKDIR /home/tau
+ENV PATH="/home/tau/.local/bin:$PATH"
+COPY src/requirements.txt /home/tau
+RUN --mount=type=cache,target=/root/.cache/pip pip install -r requirements.txt
+
+RUN python -m nltk.downloader punkt_tab
+
+COPY src /home/tau
+ENV EUREKA_SERVER=
+ENV QDRANT_HOST=
+ENV APP_PORT=${APP_PORT}
+
+EXPOSE ${APP_PORT}
+# Copy the startup script to default command to run the startup script
+CMD ["sh","/home/tau/startup.sh"]
+
+HEALTHCHECK --interval=30s --timeout=5s --start-period=10s --retries=5 \
+    CMD curl -f http://localhost:8000/healthz || exit 1

src/app.py

@@ -15,14 +15,16 @@ logger = logging.getLogger(__name__)
 # load_dotenv()
 app = Flask(__name__)
 
-# @app.route("/")
-# def hello():
-#     return jsonify({"message":"Hello, World!"})
+
 
 @app.get("/")
 def hello():
     return jsonify({"message":"Hello, World!"})
 
+@app.get("/healthz")
+def health_healthz():
+    return jsonify("OK"), 200
+
 
 @app.post("/add")
 def add():
@@ -114,20 +116,6 @@ def recommend():
         return jsonify({"error":"Internal Server Error"}), 500
 
 
-@app.post("/get_embedding")
-def get_embedding():
-    # Get the query text from request JSON
-    data = request.get_json()
-    query = data.get("query", "")
-
-    if not query:
-        return jsonify({"error": "Query text is required"}), 400
-
-    # Call the embedding function
-    vector = Embedding.call(query, is_query=True)
-    
-    return jsonify({"query": query, "number of embedding": len(vector), "one embedding":len(vector[0])})
-
 
 
 if __name__=="__main__":

src/requirements.txt

@@ -1,12 +1,12 @@
 annotated-types==0.7.0
-anyio==4.8.0
+anyio==4.9.0
 blinker==1.9.0
 certifi==2025.1.31
 charset-normalizer==3.4.1
 click==8.1.8
-filelock==3.17.0
+filelock==3.18.0
 Flask==3.1.0
-fsspec==2025.3.0
+fsspec==2025.3.2
 grpcio==1.71.0
 grpcio-tools==1.71.0
 h11==0.14.0
@@ -14,7 +14,7 @@ h2==4.2.0
 hpack==4.1.0
 httpcore==1.0.7
 httpx==0.28.1
-huggingface-hub==0.29.2
+huggingface-hub==0.30.1
 hyperframe==6.1.0
 idna==3.10
 itsdangerous==2.2.0
@@ -24,7 +24,7 @@ MarkupSafe==3.0.2
 mpmath==1.3.0
 networkx==3.4.2
 nltk==3.9.1
-numpy==2.2.3
+numpy==2.2.4
 nvidia-cublas-cu12==12.4.5.8
 nvidia-cuda-cupti-cu12==12.4.127
 nvidia-cuda-nvrtc-cu12==12.4.127
@@ -40,24 +40,26 @@ nvidia-nvjitlink-cu12==12.4.127
 nvidia-nvtx-cu12==12.4.127
 packaging==24.2
 portalocker==2.10.1
-protobuf==5.29.3
-pydantic==2.10.6
-pydantic_core==2.27.2
-python-dotenv==1.0.1
+protobuf==5.29.4
+pydantic==2.11.2
+pydantic_core==2.33.1
+python-dotenv==1.1.0
 PyYAML==6.0.2
 qdrant-client==1.13.3
 regex==2024.11.6
 requests==2.32.3
 safetensors==0.5.3
-setuptools==76.0.0
+setuptools==78.1.0
 sniffio==1.3.1
 sympy==1.13.1
-tokenizers==0.21.0
+tokenizers==0.21.1
 torch==2.6.0
 tqdm==4.67.1
 transformers==4.49.0
 triton==3.2.0
-typing_extensions==4.12.2
+typing-inspection==0.4.0
+typing_extensions==4.13.1
 urllib3==2.3.0
 uuid_utils==0.10.0
 Werkzeug==3.1.3
+gunicorn==23.0.0

startup.sh

@@ -0,0 +1,9 @@
+#!/bin/bash
+
+if [ -f /etc/ssl/cert.pem ] && [ -f /etc/ssl/key.pem ]; then
+    echo "SSL certificates found, starting Gunicorn with HTTPS..."
+    gunicorn app.wsgi:application --bind 0.0.0.0:${APP_PORT} --certfile=/etc/ssl/cert.pem --keyfile=/etc/ssl/key.pem
+else
+    echo "SSL certificates not found, starting Gunicorn with HTTP..."
+    gunicorn app.wsgi:application --bind 0.0.0.0:${APP_PORT}
+fi