Init Only Variables

[stuartthomas25]

Tesseract OCR iOS is unable to set "Init Only" variables like user_words_suffix and load_system_dawg (see https://code.google.com/p/tesseract-ocr/wiki/ControlParams). I recommend a method

• (id)initWithVariables:(NSDictionary*)parameters language:(NSString *)language;
  where "parameters" is a key-value dictionary describing init-only parameters and their values.

[kevincon]

+1. I think this is something the library really needs (it was also indirectly asked for in #45).

Among other things, it would let users provide a file with regexes for specific patterns Tesseract should try to recognize (e.g. license plates, phone numbers, etc.).

I looked into adding support for this myself a while back, and this is the relevant C++ function to call http://zdenop.github.io/tesseract-doc/classtesseract_1_1_tess_base_a_p_i.html#a96899e8e5358d96752ab1cfc3bc09f3e, but I got confused about how to map the C++ types to Objective C types. Anyway, this is where we currently call Init() in the code, but we're calling an overloaded version with fewer arguments: https://github.com/gali8/Tesseract-OCR-iOS/blob/master/TesseractOCR/Tesseract.mm#L119. For inspiration, this is how the actual tesseract binary calls Init() with parameters: https://code.google.com/p/tesseract-ocr/source/browse/api/tesseractmain.cpp#180

School gets out in two weeks, so I'll try to take a look at adding support for this then, but anyone else should feel free to take a crack at it too.

[zachberger]

I'm doing some development over on my branch called init. b61b974 is my first swing at an implementation, but it still needs testing.

The next step in creating API parity after having the variables configurable at runtime in-code, is to have the variables stored in a config file.

[kevincon]

Implemented by #113.