Python开发：Twilio+Polly场景下S3上传及相关技术问题求助

Hey there! I’ve worked through similar setups with Twilio, Amazon Polly, and S3, so let’s break down your problems step by step:

1. Fixing Boto3’s Content-Type Not Sticking for MP3 Files #

The most common issue here is not explicitly setting the ContentType in the upload arguments, or placing it in the wrong spot. Twilio requires audio/mpeg for MP3 files, so make sure you’re passing this in the ExtraArgs parameter when uploading. Here’s a working code snippet:

import boto3

# Initialize S3 client
s3_client = boto3.client('s3')

local_mp3_path = "generated_polly_voice.mp3"
bucket_name = "your-twilio-audio-bucket"
s3_object_key = "polly-audio/voice-message.mp3"

# Upload with correct content type and public ACL (Twilio needs public access)
with open(local_mp3_path, 'rb') as mp3_file:
    s3_client.upload_fileobj(
        mp3_file,
        bucket_name,
        s3_object_key,
        ExtraArgs={
            "ContentType": "audio/mpeg",
            "ACL": "public-read"
        }
    )

Double-check these potential pitfalls:

• If you were using upload_file instead of upload_fileobj, the ExtraArgs parameter still applies—don’t skip it.
• Ensure your S3 bucket doesn’t have a bucket policy or lifecycle rule that overwrites object content types.
• Verify you’re opening the file in binary read mode ('rb') to avoid corrupting the audio data.

2. Why Boto3 Changes Content-Type to binary/octet-stream #

Boto3 tries to auto-detect the content type based on the file’s extension by default. If it can’t recognize the extension (e.g., your file has no extension, or a rare one), it falls back to binary/octet-stream. Other scenarios include:

• You’re uploading raw bytes instead of a file object, so Boto3 has no filename/extension to infer from.
• You forgot to include the ContentType parameter in your upload arguments, leaving S3 to apply its default.
• A previous version of the object had binary/octet-stream and you didn’t overwrite the metadata on re-upload.

3. Uploading to S3 Subdirectories with Tinys3 #

First, keep in mind: S3 doesn’t have actual subdirectories—they’re just simulated using slashes (/) in the object key. To upload to a "subdirectory", simply include the path in your object key. Here’s how to do it with Tinys3:

from tinys3 import Connection

# Initialize Tinys3 connection
s3_conn = Connection(
    aws_access_key_id="YOUR_AWS_KEY",
    aws_secret_access_key="YOUR_AWS_SECRET",
    default_bucket="your-bucket-name"
)

local_file_path = "polly_output.mp3"
# Key includes the subdirectory path: "audio/subfolder/your-file.mp3"
s3_object_key = "audio/subfolder/polly-voice.mp3"

# Upload with correct content type and public ACL
with open(local_file_path, 'rb') as mp3_file:
    s3_conn.upload(
        s3_object_key,
        mp3_file,
        headers={
            "Content-Type": "audio/mpeg",
            "x-amz-acl": "public-read"
        }
    )

After uploading, you’ll see the "subdirectory" structure in the S3 console—this is just S3 grouping objects by the prefix before the slash. You don’t need to create the subdirectory first; uploading an object with a slash in the key will automatically create the visual structure.

内容的提问来源于stack exchange，提问作者Veysel Kocaman