A parser for Japanese number (Kanji, arabic) in the natural language.

The module japanese_numbers finds any numbers in the natural language, and converts to arabic numerals. The followings are example patterns what can be parsed.

• 二千万百一円
• 5百万
• 一を聞いて十を知る
• 五〇六号室

pip install japanese-numbers-python

Function to_arabic and to_arabic_numbers are almost stable.

to_arabic returns An array of [japanese_numbers.result.ParsedResult].

import japanese_numbers

japanese_numbers.to_arabic('銀河の向こう、六千三百二十一億千五百十一万二千百八十一光年彼方。')
# => [<ParsedResult 632115112181 : "六千三百二十一億千五百十一万二千百八十一" index=7>]

japanese_numbers.to_arabic('一を聞いて十を知る。')
# => [<ParsedResult 1 : "一" index=0>, <ParsedResult 10 : "十" index=5>]

Then you can see a numeric value (and others) in the instance of ParsedResult like as follows:

result = japanese_numbers.to_arabic('一を聞いて十を知る。')

result[0].number # => 1
result[0].text   # => '一'
result[0].index  # => 0 as position that number was found

result[1].number # => 10
result[1].text   # => '十'
result[1].index  # => 5

to_arabic_numbers returns a tuple of numbers directly.

import japanese_numbers

japanese_numbers.to_arabic_numbers('一を聞いて十を知る。')
# => (1, 10)

Both to_arabic_numbers, to_arabic get encode option to specify encode of input.

It's utf8 by default, if you put non-unicode string into functions, it will be converted to unicode by using its encode first.

japanese_numbers.to_arabic_numbers('一を聞いて十を知る。')  # utf8 by default
japanese_numbers.to_arabic('一を聞いて十を知る。', encode='eucjp')  # set another charset

• support float/double types
• support negative types

Welcome!