How to detect Zenkaku Kanaka in Checkmate or Rainbow QA check
Issue #620
closed
Japanese language contains two kinds of character set for Katakana. One is Zenkaku (2-byte) character such as ア、イ、ウ. The other is Hankaku (1-byte) character as ｱ、ｲ、ｳ. How can I detect only Zenkaku or Hanaku in Checkmate? "Character" tab is not suitable for this purpose?
Comments (7)
-
-
Thanks for quick response But in Patterms tab Source: <same> Target: [ァ-ン] Options :Trg/HIGH This does not seem to work.
-
I forgot one important thing: For the given pattern, make sure the checkbox "Perform comparison from source to target" is NOT checked (It is by default)
-
Perform comparison from source to target" has been unchecked. So "Trg/HIGH" is selected. But, [ァ-ン] cannot be detected.
-
- attached example.zip
It seems to work for me. See the attached example.
- The .qcs file is the session you can load in CheckMate (it has the patterns)
- The .xlf file is the file to verify (You may have to remove it then re-add it in CheckMate to make sure the path is correct for you)
- The .png file is a screen capture.
Also, while it is probably not helpful for your specific case, but there is a step specialized in the conversion of Zenkaku, etc. See http://okapiframework.org/wiki/index.php?title=Full-Width_Conversion_Step for details.
-
Thank you! I tried on my Checkmate, but it does not work at first time. Then I check version of Checkmate and upgraded from 3.1 to 3.3 Then it works?
-
- changed status to closed
Maybe there was a bug in 3.1 that prevented that to work. I'm glad it does work on the newer version.
- Log in to comment
There is no pre-defined way to detect such characters in the QA step. However, you can use the Patterns section to define one or more entry that will trigger an issue when for example, the target text contains a Zenkaku character.
In the new pattern, enter a regular expression such as [ァ-ン] in the pattern for the target, and <same> in the pattern for the source. When the QA step will find a target text with a Zenkaku character in the target, but not in the source, it will display an issue.
You can use various regular expression to catch various classes of character. You can find example of regular expressions here: https://gist.github.com/terrancesnyder/1345094