Implement seek over HTTPS

[mpenkov]

with smart_open.open("https://cs.nyu.edu/~roweis/data/nips12raw_str602.tgz", "rb") as fin:
    fin.seek(0)

raises "UnsupportedOperation: seek"

smart_open 1.8.3

[piskvorky]

How could we support seek over gzip? Sounds very tricky.

[mpenkov]

We're not actually seeking over gzip. The above code is seeking over raw bytes.

[piskvorky]

Aha. We don't recognize .tgz as a compressed format? (shorthand for .tar.gz)

Not sure if seek over http(s) is implemented (apparently not) – is this a bug-report ticket or a feature-request ticket? :)

[mpenkov]

Yes, we don't.

I'll reword the ticket as a feature request, that way someone may feel more inclined to act on it.