Dealing with hanjas having multiple readings

[dahlia]

Although Korean hanja have only a single reading for the most part, there are some exceptional ones as well, e.g.:

• 車
  • 차 which corresponds to chē in Mandarin and 「しゃ」 in Japanese go-on
    • 自動車 → 자동차
  • 거 which corresponds to jū in Mandarin and 「こ」 in Japanese go-on
    • 自轉車 → 자전거
• 識
  • 식 which means to know and corresponds to shí/shì in Mandarin and 「しき」 in Japanese go-on
    • 知識 → 지식
  • 지 which means a mark and corresponds to zhì in Mandarin and 「し」 in Japanese go-on
    • 標識 → 표지
• 惡
  • 악 which means evil and corresponds to è in Mandarin
    • 惡意 → 악의
  • 오 which means to hate and corresponds to wù in Mandarin
    • 嫌惡 → 혐오
• 金
  • 금
    • 金曜日 → 금요일
  • 김 which is used for personal names and some places names. This one is probably a naturalized reading of 금, but the reason is unsure. There's a hypothesis that 김 was introduced intentionally in Joseon dynasty for a reason of naming taboo (避諱) things. (Here's a Korean article.)
    • 金浦 → 김포

There are still many more, but anyway I believe it could be better at dealing with them if this software leverages KS X 1002 as well. FYI there is a property named kHangul in Unicode Han Database, which covers readings in both KS X 1001 and KS X 1002. (I made a JSON version of it as well.)

Thanks!

[sozysozbot]

Thank you for your comment. The current implementation is based on the fact that KS X 1001 encodes different readings of the same hanja separately, and the fact that the round-trip conversion between Unicode and KS X 1001 is possible. Hence it handles 「樂(U+6A02)」, 「樂(U+F914)」, 「樂(U+F95C)」 and 「樂(U+F9BF)」separately, giving 악, 낙, 락, 요 respectively. Thus all I have to do will be to NFC everything; however, the original purpose of the script was to actually handle non-normalized texts, and I have not had a chance to fix or modify the current behavior. I haven't thought about KS X 1002, and I believe your JSON version can be quite helpful for me if I would ever need to handle it. Thanks for the information.