Letter position does not correct in Japanese vertical text

[dinosauria123]

In the case of gcv2hocr.py, vertical Japanese text invisible fonts positions are bottom of the PDF document.

In gcv2hocr (C version) does better output than gcv2hocr.py but it use CR after every words.

I have to study how to deal with Japanese vertical text in ReportLab (pdf library written by python) which used in hocr-pdf tools.