Merge pull request #26 from MLOps-essi-upc/dev

Merging dev into main

.github/workflows/ci-cd.yaml

@@ -0,0 +1,55 @@
+name: Run tests CI
+on:
+  push:
+    branches:
+      - main
+      - dev
+  pull_request:
+    branches:
+      - main
+      - dev
+jobs:
+  ci:
+    runs-on: ubuntu-latest
+    steps:
+    - name: Checkout repository
+      uses: actions/checkout@v4
+
+    - name: Use python
+      uses: actions/setup-python@v4
+      with:
+        python-version: '3.9'
+
+    - name: Install dependencies 
+      run: pip install -r requirements.txt
+
+    - name: show cwd
+      run : python -c "import os; print('\n',os.getcwd())"
+
+    - name : pull data from dvc
+      run : python -m dvc pull -r origin
+
+    - name: Run tests
+      run: python -m pytest
+
+  cd:
+    runs-on: ubuntu-latest
+    needs : ci
+    steps : 
+    - name : Checkout repository
+      uses: actions/checkout@v4
+
+    - name: Docker login
+      run: docker login -u ${{ secrets.DOCKER_HUB_USERNAME }} -p ${{ secrets.DOCKER_HUB_TOKEN }}
+
+    - name: Build
+      run: docker build -t sentibites .
+    - name: Tags
+      run: |
+        docker tag sentibites ${{ secrets.DOCKER_HUB_USERNAME }}/sentibites:${{ github.sha }}
+        docker tag sentibites ${{ secrets.DOCKER_HUB_USERNAME }}/sentibites:latest
+
+    - name: Push
+      run: |
+        docker push ${{ secrets.DOCKER_HUB_USERNAME }}/sentibites:${{ github.sha }}
+        docker push ${{ secrets.DOCKER_HUB_USERNAME }}/sentibites:latest

.gitignore

@@ -162,3 +162,6 @@ cython_debug/
 # Report
 out/
 reports/
+
+# DS_STORE
+.DS_Store

Dockerfile

@@ -0,0 +1,15 @@
+FROM python:3.10
+
+ENV PYTHONUNBUFFERED 1
+
+COPY requirements.txt .
+
+WORKDIR /app
+
+COPY . /app/
+
+RUN pip install --no-cache-dir -r requirements.txt
+
+EXPOSE 5000
+
+CMD ["python3", "run_servers.py"]

README.md

@@ -39,7 +39,20 @@ dvc pull -r origin
 pytest tests/
 ```
 
-## Usage
+## Deployment with Docker
+
+1. Create the docker image
+
+```sh
+docker build -t Sentibites:1.0 .
+```
+
+2. Run the container
+```sh
+docker run -p 5000:5000 -p 8000:8000 Sentibites
+```
+
+## Usage without App
 
 For training :
 
@@ -50,5 +63,5 @@ python3 src/models/train_model.py --model "roberta-base" --dataset data/processe
 For inference :
 
 ```sh
-python3 src/models/predict_model.py --input "text"
+python3 src/models/predict_model.py --model "models/SentiBites" --input "text"
 ```

dvc.lock

@@ -49,3 +49,24 @@ stages:
       md5: 18b4f7b1478054e92600037d8b716956.dir
       size: 502103967
       nfiles: 10
+  evaluation:
+    cmd: python3 src/models/evaluate.py --model models/SentiBites --dataset data/processed
+    deps:
+    - path: data/processed/test.csv
+      hash: md5
+      md5: adffff99092202d7a587660718f62ea7
+      size: 7212513
+    - path: models/SentiBites
+      hash: md5
+      md5: 18b4f7b1478054e92600037d8b716956.dir
+      size: 502103967
+      nfiles: 10
+    - path: src/models/evaluate_model.py
+      hash: md5
+      md5: c94a08fa59debe1c0405d2aa5b63908b
+      size: 3515
+    outs:
+    - path: metrics/evaluation_scores.csv
+      hash: md5
+      md5: 9fd0ecf8f1e3e187bc6fe055758cc600
+      size: 89

pyproject.toml

@@ -5,4 +5,4 @@ omit = ["src/data/*",'src/features/*',"src/visualization/*","src/models/model.py
 
 [tool.pytest.ini_options]
 testpaths = "tests"
-addopts = "--junitxml=out/tests-report.xml --cov=src --cov-report=html:reports/coverage"
+addopts = "--junitxml=out/tests-report.xml"

requirements.txt

@@ -1,11 +1,12 @@
-# local package
--e .
-
 # Packages needed
+--extra-index-url https://download.pytorch.org/whl/cpu
+torch
+
 transformers
 mlflow
-botocore==1.31.47
+botocore  
 tensorflow
+fsspec==2023.9.2
 datasets==2.11
 accelerate>=0.20.1
 evaluate
@@ -17,9 +18,12 @@ pytest-cov
 
 # external requirements
 click
-# Sphinx>=6.0
+Sphinx  
 coverage
 awscli==1.29.47
 flake8
 python-dotenv>=0.5.1
-great-expectations
+great-expectations
+uvicorn
+gunicorn
+fastapi

run_servers.py

@@ -0,0 +1,13 @@
+from subprocess import Popen
+
+# Start uvicorn
+uvicorn_cmd = ["uvicorn", "--host", "0.0.0.0", "--port", "8000", "src.app.api:app"]
+uvicorn_process = Popen(uvicorn_cmd)
+
+# Start gunicorn
+gunicorn_cmd = ["gunicorn", "-b", "0.0.0.0:5000", "src.app.app:app"]
+gunicorn_process = Popen(gunicorn_cmd)
+
+# Wait for the processes to finish
+uvicorn_process.wait()
+gunicorn_process.wait()

src/app/api.py

@@ -0,0 +1,103 @@
+"""Main script: it includes our API initialization and endpoints."""
+
+import pickle
+from datetime import datetime
+from functools import wraps
+from http import HTTPStatus
+from typing import List
+import os
+import sys
+
+from contextlib import asynccontextmanager
+from fastapi import FastAPI, HTTPException, Request
+from src.models.predict_model import SentiBites
+from src.app.schemas import Review
+
+# Get the parent directory
+parent_dir = os.path.dirname(os.path.realpath(__file__))
+
+# Add the parent directory to sys.path
+sys.path.append(parent_dir)
+
+model = None
+
+# Define application
+app = FastAPI(
+    title="SentiBites",
+    description="This API lets you make predictions on sentiment analysis of Amazon food reviews.",
+    version="0.1",
+)
+
+
+def construct_response(f):
+    """Construct a JSON response for an endpoint's results."""
+
+    @wraps(f)
+    def wrap(request: Request, *args, **kwargs):
+        results = f(request, *args, **kwargs)
+
+        # Construct response
+        response = {
+            "message": results["message"],
+            "method": request.method,
+            "status-code": results["status-code"],
+            "timestamp": datetime.now().isoformat(),
+            "url": request.url._url,
+        }
+
+        # Add data
+        if "data" in results:
+            response["data"] = results["data"]
+
+        return response
+
+    return wrap
+
+
+@app.on_event("startup")
+def _load_model():
+    """Loads the model"""
+
+    global model
+    model = SentiBites("models/SentiBites/")
+
+
+@app.get("/", tags=["General"])  # path operation decorator
+@construct_response
+def _index(request: Request):
+    """Root endpoint."""
+
+    response = {
+        "message": HTTPStatus.OK.phrase,
+        "status-code": HTTPStatus.OK,
+        "data": {"message": "Welcome to SentiBites! Please, read the `/docs`!"},
+    }
+    return response
+
+@app.post("/models/", tags=["Prediction"])
+@construct_response
+def _predict(request: Request, payload: Review):
+    """Performs sentiment analysis based on the food review."""
+
+    if model:
+        prediction,scores = model.predict(payload.msg)
+
+        response = {
+            "message": HTTPStatus.OK.phrase,
+            "status-code": HTTPStatus.OK,
+            "data": {
+                "model-type": "RoBERTaSB",
+                "payload": payload.msg,
+                "prediction": prediction,
+                "Scores" : {
+                    "positive" : scores['positive'],
+                    "neutral" : scores['neutral'],
+                    "negative" : scores['negative']
+                }
+            },
+        }
+    else:
+        raise HTTPException(
+            status_code=HTTPStatus.BAD_REQUEST, detail="Model not found"
+        )
+    return response

src/app/app.py

@@ -0,0 +1,35 @@
+from flask import Flask, render_template, request, jsonify
+import requests
+
+app = Flask(__name__, static_folder="static")
+
+# The URL of FastAPI API
+api_url = "http://127.0.0.1:8000/models/"
+
+@app.route("/", methods=["GET", "POST"])
+def index():
+    if request.method == "POST":
+        # Get the text entered by the user
+        text = request.form.get("text")
+
+        # Build the URL with the text as a parameter
+        payload = {'msg':text}
+
+        # Call your FastAPI API
+        response = requests.post(api_url,json=payload)
+
+        if response.status_code == 200:
+            # Get the results from the API
+            data = response.json()
+            prediction = data["data"]["prediction"]
+            model_type = data["data"]["model-type"]
+
+            return render_template("index.html", prediction=prediction, model_type=model_type, text=text)
+        else:
+            error_message = "Error when calling API."
+            return render_template("index.html", error_message=error_message, text=text)
+
+    return render_template("index.html")
+
+if __name__ == "__main__":
+    app.run(debug=False)

src/app/schemas.py

@@ -0,0 +1,4 @@
+from pydantic import BaseModel
+
+class Review(BaseModel):
+    msg : str