adds support for remote images (S3, CloudFiles,..)

[baffolobill]

Adds support for remote images in the same way sorl.thumbnail does (https://github.com/mariocesar/sorl-thumbnail/blob/master/sorl/thumbnail/engines/pil_engine.py#L46).

[matthewwithanm]

Thanks @baffolobill! Do we still need to seek to the beginning of the file?

[baffolobill]

No, we don't. During StringIO initialization attribute self.pos is set to 0 (http://svn.python.org/projects/stackless/trunk/Lib/StringIO.py). So it looks like for me as excess operation. But if we replace this line

buffer = StringIO(target.read())

to

buffer = StringIO()
buffer.write(target.read())

we have to do buffer.seek(0).

[matthewwithanm]

Alright, this is merged for 1.1.7. Thanks so much for the pull request. Sorry it took a week to get to it!

[michaelshobbs]

Not sure why yet but this seems to break django-imagekit and/or PIL. Probably the latter actually.

... File "/venv/local/lib/python2.7/site-packages/imagekit/specs/init.py", line 149, in generate img = open_image(self.source) File "/venv/local/lib/python2.7/site-packages/pilkit/utils.py", line 22, in open_image return Image.open(buffer) File "/venv/local/lib/python2.7/site-packages/PIL/Image.py", line 1980, in open raise IOError("cannot identify image file")

[matthewwithanm]

@michaelshobbs Thanks for the report.

@baffolobill I've had to revert this change but if you want to submit another PR that doesn't make PIL mad, I'll merge that.

Sorry all for the confusion.

[baffolobill]

@michaelshobbs It seems the problem in PIL. Could you provide output for a command:

pip freeze | grep -E '(Pillow|PIL)'

(explanation here http://stackoverflow.com/a/18875035/1492286)

[michaelshobbs]

just PIL and pilkit.

pip freeze | grep -E '(Pillow|PIL)' PIL==1.1.7 pilkit==1.1.7

[baffolobill]

@michaelshobbs was it remote or local image? If remote, post a link to the image, if it's possible.

[michaelshobbs]

Just a local test image. Works perfectly when I downgrade to pilkit==1.1.6

[baffolobill]

@matthewwithanm I think the best (and stupid) solution for now is https://github.com/baffolobill/pilkit/commit/89432b5e9796a4e9f8fc1f76c1364e8336642618

@michaelshobbs Could you try this solution:

pip install -U git+https://github.com/baffolobill/pilkit.git#89432b5e9796a4e9f8fc1f76c1364e8336642618)

and report about bugs, please?

[michaelshobbs]

That seems to work.

[matthewwithanm]

Hm, I'm not so sure about this. The error is caused by Image.open so, in the case of a failure, we've done an unnecessary read.

[matthewwithanm]

@michaelshobbs I'm trying to reproduce but not having any luck. Can you try the following and share the output?

>>> from PIL import Image
>>> from cStringIO import StringIO
>>> print Image.__file__
>>> Image.open(StringIO(open('/path/to/image.jpg').read()))

[michaelshobbs]

pilkit==1.1.6

pip freeze:

PIL==1.1.7
pilkit==1.1.6
wsgiref==0.1.2

output:

Python 2.7.6 (default, Mar 13 2014, 10:34:57)
[GCC 4.2.1 Compatible Apple LLVM 5.1 (clang-503.0.38)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> from PIL import Image
>>> from cStringIO import StringIO
>>> print Image.__file__
/Users/mhobbs/.virtualenvs/test_env/lib/python2.7/site-packages/PIL/Image.pyc
>>> Image.open(StringIO(open('./test.jpg').read()))
<PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=350x503 at 0x102128488>

pilkit==1.1.7

pip freeze:

PIL==1.1.7
pilkit==1.1.7
wsgiref==0.1.2

output:

[GCC 4.2.1 Compatible Apple LLVM 5.1 (clang-503.0.38)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> from PIL import Image
>>> from cStringIO import StringIO
>>> print Image.__file__
/Users/mhobbs/.virtualenvs/test_env/lib/python2.7/site-packages/PIL/Image.pyc
>>> Image.open(StringIO(open('./test.jpg').read()))
<PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=350x503 at 0x10BACC488>

[matthewwithanm]

@michaelshobbs Hope you don't mind; I edited your last comment to use fenced code blocks.

I'm surprised that there wasn't any error for that. I'm going to have to dig into this more sometime to figure out what exactly is causing the issue. I can't reopen this issue (since I merged the PR), so I've opened #13. If either of you figure out the direct cause of the error, let me know.

[michaelshobbs]

Sounds good. Markdown and I have a hate-hate relationship.

I'll try to do some more debugging today.

[matthewwithanm]

:smile: Thanks!