How does seek() works on S3?

[shay1bz]

Hi,

I'm trying to find a solution for random access inside large Avro files on S3 with Python. Does smart_open download the entire file to enable seek(), or does it really "seek" (via http headers I guess) to the desired location? In my case the files are about 200-500MB, and the casual reader would read a very small portion of it by a given start/end offsets.

[mpenkov]

It uses HTTP headers.

[shay1bz]

Thank you