使用 Frozen Flask 的 POST 请求

Question

Having viewed examples on the web, it seems like one can use forms with POST requests inside static websites built with vanilla HTML/CSS/JS. But I'm unable to do this with Flask. I'm using the Frozen-Flask library .

My main server.py script is below ( model and encoder are loaded correctly at top of file)

app = Flask(__name__)
app.config['FREEZER_RELATIVE_URLS'] = True

@app.route('/')
def main():
    return render_template("index.html")

@app.route('/predict', methods=["POST"])
def predict():
    if request.method == "POST":
        message = request.form['submission'] # gets submission
        prediction = model.predict([message]) # feeds to model
        classification = encoder.inverse_transform(prediction) # decodes prediction

        return render_template('index.html', message=message, classification=classification)

if __name__ == "__main__":
    app.run(debug=True)

As well as my freeze.py script:

from flask_frozen import Freezer
from server import app

freezer = Freezer(app)

if __name__ == '__main__':
    freezer.freeze()

Finally, the relevant portion of my index.html :

<form class="form-group" action="{{ url_for('predict') }}" method="POST">
    <textarea class="form-control" name="submission" id="submission" rows="10"></textarea> 
    <button type="submit" class="btn btn-primary">Classify</button>
</form>

If I run python server.py , aka regular Flask, the website works fine. It loads up at http://localhost:5000/ and clicking the Classify button brings me to http://localhost:5000/predict/ , with the classification shown on the screen.

When I run python freeze.py , aka Frozen-Flask to generate a static website, I run into errors.

$ python freeze.py
Traceback (most recent call last):
  File "freeze.py", line 11, in <module>
    freezer.freeze()
  File "C:\Users\User\Documents\Projects\Spam Classifier\spam-classifier-hoohacks-starter\env\lib\site-packages\flask_frozen\__init__.py", line 199, in freeze
    return set(page.url for page in self.freeze_yield())
  File "C:\Users\User\Documents\Projects\Spam Classifier\spam-classifier-hoohacks-starter\env\lib\site-packages\flask_frozen\__init__.py", line 199, in <genexpr>
    return set(page.url for page in self.freeze_yield())
  File "C:\Users\User\Documents\Projects\Spam Classifier\spam-classifier-hoohacks-starter\env\lib\site-packages\flask_frozen\__init__.py", line 183, in freeze_yield
    new_filename = self._build_one(url)
  File "C:\Users\User\Documents\Projects\Spam Classifier\spam-classifier-hoohacks-starter\env\lib\site-packages\flask_frozen\__init__.py", line 322, in _build_one
    % (response.status, url))
ValueError: Unexpected status '405 METHOD NOT ALLOWED' on URL /predict/

What I've done to investigate / try to resolve this: The 405 error indicates that an HTTP request other than POST is being called on predict. So for testing purposes I changed methods=["POST"] to methods=["GET"] . And then I made my predict page just render the normal main page.

@app.route('/predict/', methods=["GET"])
def predict():
    return render_template("index.html")

Running the freeze script no longer throws an error. The website exists at .../build/index.html and clicking the Classify button takes me to .../build/predict/index.html . But the predict page is not useful, since its the same as the main page.

I'm guessing Frozen-Flask needs to have GET access to pages to build static versions of them, but shouldn't I also be able to update those pages with POST-supplied information? Or is not being able to do that part of the definition of "static"

Answer 1

From what I can see from the code, Frozen-Flask uses the test_client's GET method to obtain the content for each page to be generated. So you won't be able to use POST without hacking Frozen-Flask.

On the other hand, GET requests can include URL parameters:

/predict?message=123456

So you could obtain your parameters like this:

request.args.get('message', '')

Now, keep in mind that Forozen-Flask does not know what to send as arguments. So what I would do is something like this:

    @app.route('/predict', methods=["GET"])
    def predict():
        message = "some message" # hardcode a value
        prediction = model.predict([message]) # feeds to model
        classification = encoder.inverse_transform(prediction) # decodes prediction

        return render_template('index.html', message=message, classification=classification)