Install tesseract/pytesser on Mac OS X

Question

I am trying to install this (and additionally pytesser) for osx 10.9 (with anaconda as default python). I have looked around online but I can't get any of the tutorials to work as they all seem to be extinct (homebrew doesn't have a formula for leptonica for instance). I have probably been struggling to install this for the best part of a week with absolutely no luck at all.

Has anyone managed to succeed recently-how did you do it?

Thanks

Edit: Strangely the brew for leptonica has spluttered into life. I have the fairly strange error below.

brew install tesseract
==> Downloading https://bitbucket.org/3togo/python-tesseract/downloads/tesseract
Already downloaded: /Library/Caches/Homebrew/tesseract-3.03-rc1.tar.gz
==> ./configure --prefix=/usr/local/Cellar/tesseract/3.03-rc1
checking for leptonica... yes
checking for pixCreate in -llept... yes
checking leptonica version >= 1.70... configure: error: in `/private/tmp/tesseract-      19Ol/tesseract-3.03':
configure: error: leptonica 1.70 or higher is required
See `config.log' for more details

READ THIS: https://github.com/Homebrew/homebrew/wiki/troubleshooting

i.e it is registering the install but still not working. I will check out the config. file as instructed

Edit 2:

Upon trying to import the library in python I get this:

import tesseract

Traceback (most recent call last):

File "<stdin>", line 1, in <module>

File "//anaconda/lib/python2.7/site-packages/python-tesseract_0.8-3.0-py2.7_macosx-10.9-   intel.egg/tesseract.py", line 28, in <module>

_tesseract = swig_import_helper()

File "//anaconda/lib/python2.7/site-packages/python-tesseract_0.8-3.0-py2.7_macosx-10.9-intel.egg/tesseract.py", line 24, in swig_import_helper

_mod = imp.load_module('_tesseract', fp, pathname, description)

ImportError: dlopen(//anaconda/lib/python2.7/site-packages/python-tesseract_0.8-3.0- py2.7_macosx-10.9-intel.egg/_tesseract.so, 2): Library not loaded: /usr/local/lib/libtesseract.3.dylib

Referenced from: //anaconda/lib/python2.7/site-packages/python-tesseract_0.8-3.0-py2.7_macosx-10.9-intel.egg/_tesseract.so

Reason: image not found

To be honest I am a complete amateur with respect to any of this behind the scenes installation and had to google extensively to even get this far. I would be really grateful if someone with a bit of knowledge could shed any light on the obvious things to try, as I feel as though I have exhausted the web looking for solutions and am getting close to considering this library unuseable and attempting to write my own ocr library-100% not a job I am looking forward to. Alternatively, if anyone knows any decent python ocr libraries with decent support/ install mainatenance I would love to know about them (From my google searching I suspect that tesseract is by far the best known, which is why it is so frustrating that the install is so tricky)

I will happily provide any any more info about my system etc to any warrior willing to have a crack at helping with this.

Thanks!

Solution 1

I have just installed tesseract 3.02 using brew without any issues (osx 10.9). If you don't need version 3.03, you may want to try installing 3.02. Instructions on installing a different version using brew: Homebrew install specific version of formula?

Otherwise, based on your log, the brew install did not complete successfully so tesseract can not be imported. Brew downloads the source, runs configure, then does make install. The configure step is failing because you need leptonica 1.70. Usually brew would detect this dependency and install leptonica 1.70 for you.

You may want to try installing leptonica yourself: http://www.leptonica.com/download.html. Instructions on building: http://www.leptonica.com/source/README.html

Solution 2

You need to install tesseract first

https://bitbucket.org/3togo/python-tesseract/downloads/tesseract.rb

For details, https://code.google.com/p/python-tesseract/wiki/HowToCompileForHomebrewMac