如何确认上传“直接”到S3，而不首先通过后端服务器进行代理？(Python示例)

Question

我很好奇如何使用Fineuploader将文件上传到S3的传统方法，首先通过后端代理文件，而不是直接从客户端上传文件。下面是我从Fineuploader检索到的后端代码，并将其稍微调整到我的usecase中。

#!/usr/bin/env python
#
# app.py
#
# by: Mark Feltner
#
# Server-side S3 upload example for Fine Uploader
#
# Features:
# * Upload to S3
# * Delete from S3
# * Sign Policy documents (simple uploads) and REST requests (chunked/multipart)
#   uploads
# * non-CORS environment

import base64, hmac, hashlib, os, sys
from flask import (Flask, abort, json, jsonify, make_response, render_template, request)
from boto.s3.connection import Key, S3Connection

AWS_CLIENT_SECRET_KEY = '********************************'
AWS_SERVER_PUBLIC_KEY = '************************'
AWS_SERVER_SECRET_KEY = '*********************************'
AWS_EXPECTED_BUCKET = 'mybucket'
AWS_MAX_SIZE = 1500000000

app = Flask(__name__, template_folder='static', static_url_path='/static')
app.config.from_object(__name__)
app.debug = True


def sign_policy(policy):
    """ Sign and return the policy document for a simple upload.
    http://aws.amazon.com/articles/1434/#signyours3postform """
    signed_policy = base64.b64encode(policy)
    signature = base64.b64encode(hmac.new(
        app.config.get('AWS_CLIENT_SECRET_KEY'), signed_policy, hashlib.sha1).
                                 digest())
    return {'policy': signed_policy, 'signature': signature}


def sign_headers(headers):
    """ Sign and return the headers for a chunked upload. """
    return {
        'signature': base64.b64encode(hmac.new(
            app.config.get('AWS_CLIENT_SECRET_KEY'), headers, hashlib.sha1).
                                      digest())
    }


@app.route("/s3/sign", methods=['POST'])
def s3_signature():
    """ Route for signing the policy document or REST headers. """
    request_payload = request.get_json()
    if request_payload.get('headers'):
        response_data = sign_headers(request_payload['headers'])
    else:
        response_data = sign_policy(request.data)
    return jsonify(response_data)


@app.route("/s3handler/<key>", methods=['POST', 'DELETE'])
def s3_delete(key=None):
    """ Route for deleting files off S3. Uses the SDK. """
    try:
        s3 = S3Connection(app.config.get("AWS_SERVER_PUBLIC_KEY"),
                          app.config.get("AWS_SERVER_SECRET_KEY"))
        request_payload = request.values
        bucket_name = request_payload.get('bucket')
        key_name = request_payload.get('key')
        aws_bucket = s3.get_bucket(bucket_name, validate=False)
        aws_key = Key(aws_bucket, key_name)
        aws_key.delete()
        return make_response('', 200)
    except ImportError:
        abort(500)


@app.route("/s3/success", methods=['GET', 'POST'])
def s3_success():
    """ Success redirect endpoint for <=IE9. """
    return make_response('', 200)


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


def main(argv=None):
    app.run('0.0.0.0')

    return 0  # success


if __name__ == '__main__':
    status = main()
    sys.exit(status)

在客户端，减去UI代码，我有这个脚本来处理客户端的请求

<script>
    var uploader = new qq.s3.FineUploader({
        debug: true,
        element: document.getElementById('fine-uploader'),
        request: {
            endpoint: 'mybucket.s3.amazonaws.com',
            accessKey: '***************'
        },
        signature: {
            endpoint: '/s3/sign'
        },
        uploadSuccess: {
            endpoint: '/s3/success'
        },
        iframeSupport: {
            localBlankPagePath: '/success.html'
        },
        retry: {
            enableAuto: true // defaults to false
        },
        chunking: {
            enabled: true
        },
        deleteFile: {
            enabled: true,
            endpoint: '/s3handler'
        }
    });
</script>

这对我有用。上传的文件出现在我的S3桶中。我假设这是“直接”上传到S3，只要求后端签署策略文件，然后继续。这是正确的吗？有什么不同的做法，必须通过文件通过后端，而不是直接上传？

Answer 1

经过Ray的澄清后，直接上传到S3或替代服务器之间的区别在于Fineuploader客户端方法

直接上传到S3: qq.s3. qq.s3.FineUploader

上传到您选择的服务器: qq.FineUploader