if query result source has no encoding set, fall back to utf-8 encoding.

[gweis]

It's a bit tricky to do this properly for python 2 and 3, but I think this patch goes into the right direction.
For later on I think rdflib should be clear about what source is, and what attributes, methods and return values are expected.
If that would be defined, then a user of the library would know when to wrap a source within a codec or not.
resolves #344

[gweis]

Not quite sure about it, but the failing travis tests seem to be unrelated.
I hope this patch can go into 4.1.1 release, if this works for everyone

And there is at least one scenario where this code is still problematic. The decoder will fail if you pass in a StringIO instance returning unicode characters. Otherwise I think it covers a few more use scenarios than before.

[joernhees]

yupp, sorry for the failures, master currently is broken, but fixed soon...

http://stackoverflow.com/questions/4528869/how-do-you-attach-a-new-pull-request-to-an-existing-issue-on-github

[gweis]

Hi,

Yes, read that stackoverflow, but had troubles installing the hub tool.
(Some conflict with another library I use)

However due to the resolves reference to issue #344, accepting this pull
request should also close the other issue. (Or at least mark as resolved.)

Good enough?

Cheers

Gerhard

Sent from my iPad

On 3 Mar 2014, at 7:25 pm, "Jörn Hees" [email protected] wrote:

yupp, sorry for the failures, master currently is broken, but fixed soon...

http://stackoverflow.com/questions/4528869/how-do-you-attach-a-new-pull-request-to-an-existing-issue-on-github

—
Reply to this email directly or view it on
GitHubhttps://github.com//pull/367#issuecomment-36493165
.

[joernhees]

sure, i was just confused about all the notifications ;)

From my POV the fix seems ok

[gweis]

Yeah, tried to attach the pull request to issue #344, that's why I created
and closed the ones that didn't work properly.

Sorry for the noise in the tracker.

Sent from my iPad

On 3 Mar 2014, at 9:28 pm, "Jörn Hees" [email protected] wrote:

sure, i was just confused about all the notifications ;)

From my POV the fix seems ok

—
Reply to this email directly or view it on
GitHubhttps://github.com//pull/367#issuecomment-36501751
.

[gromgull]

Did you try the trick with reading a zero length string and seeing if you get unicode or bytes back?

[gromgull]

it may be a more robust heuristic

[gromgull]

otherwise it looks ok to me! The StringIO from a BytesIO was ugly anyway :)

[gweis]

No, haven't thought about reading zero length string. Might fix the problem with StringIO returning unicode. I'll give it a try.

[gweis]

How about this variant? The tests pass for me.

If no one sees a problem with this variant, I'll squash the commits when merging.

[coveralls]

Changes Unknown when pulling 2498360 on gweis:master into * on RDFLib:master*.

[gromgull]

Looks good! Squash and merge!

[gweis]

squashed into commit b7fa8d6

[gromgull]

Thanks!