Tildes (~) in BoxText output not in UTF8 output

[robertmeta]

I am going to have to dig through the Tesseract source to try to find why I am getting ~ in the BoxText when it is not in the UTF8 text. I am doing a mapping of boxes to UTF8 text chars, and simply removing the ~'s allows it to line up properly.... trying to puzzle it out.

[robertmeta]

This is the "tilde crunch words" for stuff that isn't being detected, I think textord_noise_rejwords controls it.

[robertmeta]

"Working As Intended" https://code.google.com/p/tesseract-ocr/source/browse/api/baseapi.cpp (line 86) -- that is a hard-coded tesseract thing -- nothing to be done about it on the go.tesseract side.