search

nvaccess / nvda

NVDA, the free and open source Screen Reader for Microsoft Windows
https://www.nvaccess.org/
Other
2.09k stars 634 forks source link

Japanese characters in console reported as space #2157

Open nvaccessAuto opened 12 years ago

nvaccessAuto commented 12 years ago

Reported by tspivey on 2012-03-09 00:41 STR:

  1. Run intl.cpl (regional and language control panel), administrative tab, press change system locale, select japanese, ok, reboot when it asks.
  2. When the system comes back, open cmd. Paste:
echo これは試験です

And run it. If I move up to the line that it echoes, I can't move across individual characters. They're all reported as spaces. Trying to copy one reports no text, but moving across two of them allows me to copy one.

  1. Optional: after fixing bug, reset locale and remove the Japanese keyboard layout it installed.

My first guess is it has something to do with multi-byte characters, since each one of those characters is two bytes in the ansi codepage.

nvaccessAuto commented 12 years ago

Comment 1 by tspivey on 2012-03-11 20:30 This is what I know so far. First, ReadConsoleOutputCharacter() is returning unicode, as I expected. Using the test string u'\u3053x' starting at 0,0:

read(length, startX, startY) #where x is column and y is row
read(1, 0, 0) = u' '
read(1, 1, 0) = u' '
read(2, 0, 0) = u'\u3053'
read(1, 2, 0) = u'x'

According to this, each multibyte character takes 2 columns. We have the total number of columns on screen stored somewhere. If console reads worked on entire lines at a time, padding to the right with spaces, I think that would solve it. Is there a better idea?

bhavyashah commented 7 years ago

@tspivey @nishimotz Could you please test and report whether or not the reported issue still persists?

nishimotz commented 7 years ago

Though I haven't yet tested with the official builds, my working version based on the current master branch causes the issue mentioned here. My setup is as follows:

Another issue may be related to this. Reviewing the text by line, every Japanese characters are reported twice, i.e. duplicated such as ここれれはは試試験験でですす The work around for this duplication we found so far is to choose the raster font.

Adriani90 commented 5 years ago

@tspivey is this still reproducible in NVDA 2018.4 Beta 1?

Adriani90 commented 5 years ago

@tspivey, just a reminder. Do you still have this problem in NVDA 2019.1.1? cc: @nishimotz in case the author is not available, could you please test if this issue is still occuring? Thanks!

Adriani90 commented 5 years ago

@nishimotz are you available to test if this is still an issue in NVDA 2019.2 RC2? Maybe the solution with UIA provided by @codeofdusk also solves this issue?

nishimotz commented 5 years ago

Tested with 2019.2 rc2 installed. When "Use UI Automation to access the Windows Console" is enabled, the issue of duplicated character readings does not occur.