Exceptions for number words

[bittlingmayer]

import suji
suji.value('四半期')

Actual:

44半期

Expected:

4四半期

四半期 means quarter, but 四 is getting interpreted as 4, which is only etymologically correct.

Similar for 2月 / 二月 / ２月 etc.

I'd probably create a list of these exceptions, and mask them out (by replacing with ' ') before searching for actual numbers.

(Most parsers (including spaCy) have something similar for rules-based sentence segmentation - a list of abbrevations so they know when not to split on ..)

[jikyo]

@bittlingmayer Thanks for your feedback!
I'll add a list of such exceptions not to parse ASAP.

[ToraRuka]

@jikyo Thanks for your great works!

In addition to this issue,

>>> import suji
>>>suji.kansuji("40分")
'零'

like above code, all number with "分" are replaced to "零"
I think this is bug!

[jikyo]

@ToraRuka Thank you so much!
I screwed up...
The decimals number function works for time representation.
I'll also fix this bug ASAP.

[bittlingmayer]

Thanks @jikyo, I appreciate that.

Normally in the spirit of open source, I would send a PR. But I don't know Japanese at all, so I'm not very efficient at thinking about what other exceptions, or exceptions to the exceptions, exist.