Words array not correctly generated for Japanese/Korean/Chinese

[rigelglen]

I'm trying to use tesseract js to detect some Korean text, but the tesseract output for words has one character per word.

The value of text property is correctly recognised as 안녕하세요! without spaces between the characters, however the words array has single characters in it.

The worker parameters I am setting are:

tessedit_pageseg_mode: PSM.AUTO,
tessedit_ocr_engine_mode: OEM.LSTM_ONLY,
preserve_interword_spaces: '1'

I am using tesseract.js: "^2.0.2".

[jeromewu]

After some quick research, I have to say there is no good way to solve this issue, so we won't fix this issue for now.

[kazupon]

Hi!
I have same issue.

I found this issue on tesseract.
tesseract-ocr/tesseract#991

In this issue, it seems to be possible to work around it with parameters.

If tesseract.js-core can also be controlled by parameters, it would seem to solve this issue, but is it hard?