Error setting up the tesseract OCR in gem in rails

[seanhinton]

I'm trying to setup the tesseract-ocr gem in my rails environment. I have ran brew install tesseract and then ran a bundle install on the app and that all runs without errors however when starting the app (rails s) the following error is throw:

/Users/xxxx/.rvm/gems/ruby-1.9.2-p290@xxxx/gems/ffi-inline-0.0.4.3/lib/ffi/inline/compilers/gcc.rb:35:in `compile': compile error: see logs at /var/folders/66/pm_j0lp94gvcj0qnlcnsx9pw0000gn/T/.ffi-inline-501/4239dac38f2a721e0dc5b3750d71ce2e6fa4acb6.log (CompilationError)

It reference the following log file:

g++ -dynamic -bundle -fPIC  -L/usr/local/Cellar/tesseract/3.02.02/lib -I/usr/local/Cellar/tesseract/3.02.02/include -o /var/folders/66/pm_j0lp94gvcj0qnlcnsx9pw0000gn/T/.ffi-inline-501/4239dac38f2a721e0dc5b3750d71ce2e6fa4acb6.dylib /var/folders/66/pm_j0lp94gvcj0qnlcnsx9pw0000gn/T/.ffi-inline-501/4239dac38f2a721e0dc5b3750d71ce2e6fa4acb6.cpp -llept 2>>/var/folders/66/pm_j0lp94gvcj0qnlcnsx9pw0000gn/T/.ffi-inline-501/4239dac38f2a721e0dc5b3750d71ce2e6fa4acb6.log
/var/folders/66/pm_j0lp94gvcj0qnlcnsx9pw0000gn/T/.ffi-inline-501/4239dac38f2a721e0dc5b3750d71ce2e6fa4acb6.cpp:1:10: fatal error: 'leptonica/allheaders.h' file not found
#include <leptonica/allheaders.h>

I followed these two issues:
#3 and #21

It talks about setting ENV vars which i have done. I created a test.rb file in the initalizers folder with the following vars set:

ENV['CFLAGS'] = '-I/usr/local/Cellar/tesseract/3.02.02/include' 
ENV['LDFLAGS'] = '-L/usr/local/Cellar/tesseract/3.02.02/lib'

However this makes no difference to the error when starting the app.

I don't understand what it means when it says update the headers. I was wondering if someone could tell me exactly what to do to get this working.

Thanks.

[meh]

Are the environment variables set before the gem is required?

[seanhinton]

Hi, thanks for replying, here's some more details to answer you question...

I haven't put 'require' anywhere yet because the rails server is throwing the error on start up.

The only place 'tesseract-ocr' is referenced is in the Gemfile and the Gemfile.lock file.

The ENV vars are sat in the initializer folder. However I tried putting them in the application.rb before the rails app is initialized and even putting them at the start of the Gemfile before gem tesseract-ocr but that made no difference.

[meh]

I don't know much about rails, so I don't know when it runs the Bundler.required, but you should try exporting the env variables in the shell before starting the rails app.

[seanhinton]

I tried running:

export CFLAGS=-I/usr/local/Cellar/tesseract/3.02.02/include
export LDFLAGS=-L/usr/local/Cellar/tesseract/3.02.02/lib

In the shell before running 'rails s' however that still threw the same error on start up.

I don't know if this is relevant but I'm on Mavericks.

[meh]

Could you give me the output of find /usr/local/Cellar/tesseract/3.02.02/include please?

[seanhinton]

Sure it's:

/usr/local/Cellar/tesseract/3.02.02/include
/usr/local/Cellar/tesseract/3.02.02/include/tesseract
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/apitypes.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/baseapi.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/basedir.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/capi.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/errcode.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/fileerr.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/genericvector.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/helpers.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/host.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/ltrresultiterator.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/memry.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/ndminx.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/pageiterator.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/params.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/platform.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/publictypes.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/resultiterator.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/serialis.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/strngs.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/tesscallback.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/thresholder.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/unichar.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/unicharmap.h
/usr/local/Cellar/tesseract/3.02.02/include/tesseract/unicharset.h

[meh]

Oh right, it was looking for the leptonica headers, I think you have to also add the path to leptonica.

[seanhinton]

Ok, how do I do that? Do I first need to brew install leptonica too? (thanks for your help by the way)

[meh]

No, it's a dependency of tesseract, it should have installed it along, you just have to find where it put it.

[seanhinton]

I did a locate and found /usr/local/Library/Formula/leptonica.rb and brew says it's already installed.

Sorry about all the questions, now where do I put the reference to '/usr/local/Library/Formula/leptonica.rb' does this need another env var?

[meh]

It's not that one, it should have its own include dir like tesseract, try installing it tho.

[seanhinton]

I've brew re-installed leptonica and it has it's own dir like tesseract now in '/usr/local/Cellar/leptonica/1.69' with a lib and bin, however the error is still getting throw.

[meh]

So there's no include dir? The error is thrown because it can't find leptonica's headers.

Try running a locate for allheaders.h (and make sure to updatedb, or use find instead).

[seanhinton]

Yep, there's an include folder too with those files in it:

/usr/local/Cellar/leptonica/1.69/include/leptonica
/usr/local/Cellar/leptonica/1.69/include/leptonica/allheaders.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/alltypes.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/array.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/arrayaccess.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/bbuffer.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/bmf.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/bmp.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/ccbord.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/dewarp.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/endianness.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/environ.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/gplot.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/heap.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/imageio.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/jbclass.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/leptwin.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/list.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/morph.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/pix.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/ptra.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/queue.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/readbarcode.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/regutils.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/stack.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/sudoku.h
/usr/local/Cellar/leptonica/1.69/include/leptonica/watershed.h

So that's all there too.

[meh]

Then just add those paths to the two env vars too.

[seanhinton]

Thanks!

That's it, perfect! My guess is this is Mavericks related.

Just incase anyone else has a problem here's exactly how it was fixed in mine:

Firstly had to run both:
brew install tesseract
and
brew install leptonica

Then in the Gemfile NOT in an initializer you have to put both paths to the libraries as env vars:

ENV['CFLAGS'] = '-I/usr/local/Cellar/tesseract/3.02.02/include -I/usr/local/Cellar/leptonica/1.69/include'
ENV['LDFLAGS'] = '-L/usr/local/Cellar/tesseract/3.02.02/lib -L/usr/local/Cellar/leptonica/1.69/lib'

[meh]

Thanks for summing it up.

[ExReanimator]

Hi guys!
I have the same problem, but this solution doesn't work.

ENV['CFLAGS'] = '-I/usr/local/Cellar/tesseract/3.04.00/include -I/usr/local/Cellar/leptonica /1.72/include'
ENV['LDFLAGS'] = '-L/usr/local/Cellar/tesseract/3.04.00/lib -L/usr/local/Cellar/leptonica/1.72/lib'

[ExReanimator]

Maybe it's helpful.
In errors list I found a wrong path forming:

g++ -dynamic -bundle -fPIC  -L/usr/local/Cellar/tesseract/3.04.00/lib -L/usr/local/Cellar/leptonica/1.72/lib -I/usr/local/Cellar/tesseract/3.04.00/include -I/usr/local/Cellar/leptonica /1.72/include -o /var/folders/xg/g9n7qvns5z1gsr_yjh09n1nm0000gn/T/.ffi-inline-501/d2f8bb8a1867b800ff8ad69a3b850c91521b3760.dylib /var/folders/xg/g9n7qvns5z1gsr_yjh09n1nm0000gn/T/.ffi-inline-501/d2f8bb8a1867b800ff8ad69a3b850c91521b3760.cpp -ltesseract 2>>/var/folders/xg/g9n7qvns5z1gsr_yjh09n1nm0000gn/T/.ffi-inline-501/d2f8bb8a1867b800ff8ad69a3b850c91521b3760.log
clang: error: no such file or directory: '/1.72/include'

-I/usr/local/Cellar/leptonica**[unuseful space]**/1.72/include

[travisgalloway]

I have the same error with tesseract 3.04 and leptronica 1.72 installed via brew.

I tried setting the ENV flags according to above and reinstalling the ffi and ffi-inline gems.

The log file output:

g++ -dynamic -bundle -fPIC    -o /var/folders/b8/2plq0vx56hsb5_zxz7xqrgwc0000gn/T/.ffi-inline-501/8f22b8bfc158eaaf371cbe8a8112eaaa9ec0782d.dylib /var/folders/b8/2plq0vx56hsb5_zx$
/Users/travisgalloway/.rvm/gems/ruby-2.2.3/gems/tesseract-ocr-0.1.8/lib/tesseract/c/baseapi.rb:171:10: error: no matching member function for call to 'SetImage'
                                api->SetImage(pix);
                                ~~~~~^~~~~~~~
/usr/local/include/tesseract/baseapi.h:354:8: note: candidate function not viable: 1st argument ('const Pix *') would lose const qualifier
  void SetImage(Pix* pix);
       ^
/usr/local/include/tesseract/baseapi.h:341:8: note: candidate function not viable: requires 5 arguments, but 1 was provided
  void SetImage(const unsigned char* imagedata, int width, int height,
       ^
/Users/travisgalloway/.rvm/gems/ruby-2.2.3/gems/tesseract-ocr-0.1.8/lib/tesseract/c/baseapi.rb:183:49: error: cannot initialize a parameter of type 'tesseract::TessResultRendere$
                                return api->ProcessPages(filename, NULL, 0, output);
                                                                             ^~~~~~
/usr/local/include/tesseract/baseapi.h:542:63: note: passing argument to parameter 'renderer' here
                    int timeout_millisec, TessResultRenderer* renderer);
                                                              ^
/Users/travisgalloway/.rvm/gems/ruby-2.2.3/gems/tesseract-ocr-0.1.8/lib/tesseract/c/baseapi.rb:189:65: error: cannot initialize a parameter of type 'tesseract::TessResultRendere$
                                return api->ProcessPage(pix, page_index, filename, NULL, 0, output);
                                                                                            ^~~~~~
/usr/local/include/tesseract/baseapi.h:558:40: note: passing argument to parameter 'renderer' here
                   TessResultRenderer* renderer);
                                       ^
3 errors generated

Was there a breaking API change?

[nateklaiber]

@travisgalloway - I was able to past this by following the instructions found here.

brew uninstall tesseract
brew install https://raw.githubusercontent.com/Homebrew/homebrew/8ba134eda537d2cee7daa7ebdd9f728389d9c53e/Library/Formula/tesseract.rb

Then I had to updated my ENV variables to point to the new path.