Flask新手求助：如何将/predict路由的predictions变量渲染到/hello路由

Hey there! Let's work through how to get that predictions variable from your /predict POST route over to the /hello page. Since you're new to Flask, I'll walk you through two simple, practical approaches that fit most beginner use cases.

Approach 1: Use Flask Session (Recommended for Temporary, Sensitive Data) #

Flask's built-in session lets you store small amounts of data securely in the user's browser (encrypted with a secret key). Perfect for passing temporary results like your predictions.

Step 1: Set a Secret Key #

First, you need to add a secret key to your app—this is required for Flask to encrypt session data. Pick a random, secure string (you can generate one using secrets.token_hex(16) in Python).

from flask import Flask, request, session, redirect, url_for, render_template
import pandas as pd
import joblib

app = Flask(__name__)
app.secret_key = 'your-secure-random-secret-key-here'  # Replace with your own unique key

Step 2: Store Predictions in Session & Redirect #

Update your /predict route to calculate predictions, store them in the session, then redirect to /hello:

@app.route('/predict', methods=['POST'])
def apicall():
    test_json = request.get_json()
    test = pd.read_json(test_json, orient='records')
    query_df = pd.DataFrame(test)
    
    clf = 'kmeans__model.pkl'
    print("Loading the model...")
    with open(clf,'rb') as f:
        lin_reg_model = joblib.load(f)
    
    # Calculate your predictions (added this line since your code cut off)
    predictions = lin_reg_model.predict(query_df)
    
    # Store predictions in session (convert to list since numpy arrays aren't serializable)
    session['predictions'] = predictions.tolist()
    
    # Redirect to the /hello route
    return redirect(url_for('hello'))

Step 3: Retrieve Predictions in /hello Route #

Now, update your /hello route to pull the predictions from the session and pass them to your template:

@app.route('/hello')
def hello():
    # Get predictions from session (default to empty list if not present)
    predictions = session.get('predictions', [])
    # Optional: Clear the session if you don't want predictions to persist after page load
    # session.pop('predictions', None)
    return render_template('hello.html', predictions=predictions)

Step 4: Display Predictions in Your Template #

In your hello.html template, add code to render the predictions:

<h1>Your Predictions</h1>
{% if predictions %}
    <ul>
        {% for pred in predictions %}
            <li>{{ pred }}</li>
        {% endfor %}
    </ul>
{% else %}
    <p>No predictions found!</p>
{% endif %}

Approach 2: Use Query Parameters (For Simple, Non-Sensitive Data) #

If your predictions are short (like a single value or small list), you can pass them directly in the URL as query parameters. Note: This isn't secure for sensitive data, and URL length is limited.

Update /predict Route #

@app.route('/predict', methods=['POST'])
def apicall():
    # ... (keep your existing code to load model and calculate predictions)
    predictions = lin_reg_model.predict(query_df)
    
    # Convert predictions to a comma-separated string
    predictions_str = ','.join(map(str, predictions.tolist()))
    # Redirect to /hello with predictions as a query parameter
    return redirect(url_for('hello', predictions=predictions_str))

Update /hello Route #

@app.route('/hello')
def hello():
    # Get the query parameter value
    predictions_str = request.args.get('predictions', '')
    # Convert the string back to a list of floats
    predictions = list(map(float, predictions_str.split(','))) if predictions_str else []
    return render_template('hello.html', predictions=predictions)

Key Notes #

• Session Tips: Session data is encrypted, so it's safe for sensitive info. Don't store large datasets here—keep it small (like prediction results).
• Query Parameter Limits: Avoid using this for large prediction lists, as URLs have a maximum length (usually around 2000 characters).
• Serialization: Make sure any data you store in session or pass via URL is serializable (numpy arrays need to be converted to lists/strings first).

内容的提问来源于stack exchange，提问作者Rustem Kussaiynov