Closed phaebz closed 11 years ago
Looks like I've broken Python 3 support. Not sure how I missed that ... :/ I'll try to fix it this evening.
I've pushed a fix on the branch 'master'. Can you give it a try please ?
I rebased on master, installed to virtualenv and tested. Same problem(s). Problem(s) are about module import. Unicode fixes are great, but I did not get to any issues since I can't even import properly. Am I missing something?
Imports failed because of the mistakes I made we unicode. (setup.py showed various errors, and I think it didn't install all the files).
I haven't tried in virtualenv for quite some times. I'll give it a try.
Ok, I can reproduce the problem here with Virtualenv + Python 3
Thanks for clarification. Tested with system Python 2.7.5 (works) and system Python 3.3.2 (fails) under OS X. I only read now that you do not support OS X, though. Anyway, since it seems to work with Python 2.7.5, why not try to make it work on Python 3 on OS X? My Debian box is at work unfortunately - will try tomorrow.
The main problem with MacOSX is that I don't know where tesseract and cuneiform binaries and data are located. Also, I have no MacOSX, which doesn't help :)
Just so you know, there are no real differences between the last released version of Pyocr and the one in the branch 'master' (except the previous fix regarding unicode). So you can try installing the Git version system-wide (ie not in a virtualenv). It should work (well ... it does for me anyway...)
Ok so let's leave it at that for now (: I will do the system wide install at work tomorrow.
Oops, nevermind. It doesn't ...
And fixed ( 4f77d7b23f61ef1f7cca16d7dff0c5a0de3e424f ). Please try it now. If it works for you as well, I will do a new release asap.
Obviously, just running the tests with python 3 and assuming everything else would go fine was a baddddd idea. I probably did the same mistake with Pyinsane ... :/
Ok, now imports work with from pyocr import pyocr
and from pyocr import builders
. I tried the example from the README and it recognized tesseract.py
as available too. Now I do have tesseract installed in /usr/local/bin/tesseract as is done with most (all?) homebrew binaries. I tested further with a random .jped containing text. The example functions returned empty results, i.e. ''
or []
which may be related to not calling tesseract correctly. I will have a look into it. Anyway, I will close this if it works tomorrow on Debian :)
Tested some more and it works on the test.png inside the tests/data dir with tesseract from homebrew, no tweaking!
import works also on my Debian box. Now, you should probably change the README to reflect that import pyocr
is not possible anymore, but instead from pyocr import pyocr
etc. should be used.
Actually, it was another bug. Sorry for that. It's fixed : fa695265a84418f18631fd06c524bcbcd3ce5fbe
As you probably guessed by now, since Paperwork only works with Python 2.7, I haven't tested Pyocr and Pyinsane with Python 3 much.
The recommended way to import pyocr is just import pyocr
. from pyocr import pyocr
is obsolete and was just kept to avoid compatibility issues (if I remember correctly, Paperwork 0.1 still uses this old way ; it will only be changed in 0.2)
Ok, with fa695265a84418f18631fd06c524bcbcd3ce5fbe it works. Forget about the README change then. The issue about relative imports is solved then. Thanks for fixing.
You're welcome. Also, thanks for reporting these issues in a first place.
I've released Pyocr 0.2.2 with all these fixes included.
I am doing
import pyocr
as described in the README, but after that I can't accesspyocr.get_available_tools()
. So is there something wrong with thefrom pyocr import *
in__init__.py
? FWIW, I am running on Debian Testing in a virtualenv with pyocr installed with pip.Update: I tried with the old way
from pyocr import pyocr
which gives the error