Can I trust aws-sdk-s3's GetObjectOutput's content_length?

[amunra]

I want to know how large the body is before I start processing it.

I've seen there's res.content_length(), but it very suspiciously returns an i64 instead of u64.

I'm probably overthinking this, but can I trust this value to always be accurate and never have any negative "magic" values?

The API doc does not mention why the field is signed at all:

• Is it simply a mirror of the HTTP response coming back?
• What if the http content length is not set by the server (such as is allowed by HTTP 1.0 or chunked encoding in HTTP 1.1?)

[ysaito1001]

Yes, it can be trusted. The reason why it is i64 is that it is defined as such in a Smithy model for S3:

"com.amazonaws.s3#ContentLength": {
            "type": "long",
            "traits": {
                "smithy.api#default": 0
            }
        },

The SDK in another language, such as Java, may not have unsigned long type. So for consistency across SDKs generated out of the model, the return type of content_length is defined to be i64.

[rcoh]

Is it simply a mirror of the HTTP response coming back?

Yeah, this is extracted from the Content-Length header sent from S3.

What if the http content length is not set by the server (such as is allowed by HTTP 1.0 or chunked encoding in HTTP 1.1?)

The value will be 0, however, I would expect S3 to send Content-Length in all cases

[amunra]

Thanks!

[github-actions[bot]]

Hello! Reopening this discussion to make it searchable.