Sino-Korean numerals

Question

The Korean language has two sets of numerals, a native Korean numerals and Sino-Korean numerals.

The native Korean numerals are mostly used for numbers under 100, such as age and time.

The Sino-Korean numeral system, which is based on the Chinese numeral system as its name says, is used in most of the time.

It uses 4-digit group system rather than western 3-digit group system. For example, 123456789 can be grouped into 1, 2345, 6789 in Sino-Korean system.

All 4 digits in group have their own place names.

• Thousands: 천
• Hundreds: 백
• Tens: 십
• Ones: 일, but not used.

Also, for every 4 digits, suffixes are coming along with numbers.

• 10^0: none, obviously.
• 10^4: 만
• 10^8: 억
• 10^12: 조
• 10^16: 경
• 10^20: 해
• ...

So, the number '123456789' in Sino-Korean is "일억 이천삼백사십사만 육천칠백팔십구".

Goal and Rules

Your goal is to convert given integer to Sino-Korean numeral.

• Numbers are grouped into 4 digits from behind.

• Whitespaces are inserted every 4 digits (from behind).

  4321 -> 사천삼백이십일
  30723231 -> 삼천칠십이만 삼천이백삼십일

• '일'(1) is omitted except if it is last digit of its group.

  1111 -> 천백십일 (not 일천일백일십일)
  137161293 -> 일억 삼천칠백십육만 천이백구십삼

• There is only one exception, 10000 is not "일만", but "만".

  18210 -> 1 8210 -> 만 팔천이백십 (not 일만 팔천이백십)

• Standard code-golf rules are applied.

Table of Sino-Korean Digits

+-------+-------------+
| Digit | Sino-Korean |
+-------+-------------+
|     0 |          영 |
|     1 |          일 |
|     2 |          이 |
|     3 |          삼 |
|     4 |          사 |
|     5 |          오 |
|     6 |          육 |
|     7 |          칠 |
|     8 |          팔 |
|     9 |          구 |
+-------+-------------+

Input and Output

Input is integer \$N, 0 \leq N < 10^{24}\$

Output is UTF-8 formatted Sino-Korean numeral.

Examples

• (Input) -> (Output)
• 0 -> 영
• 1 -> 일
• 10 -> 십
• 101 -> 백일
• 301821 -> 삼십만 천팔백이십일
• 10282714472 -> 백이억 팔천이백칠십일만 사천사백칠십이
• 999999999999999999999999 -> 구천구백구십구해 구천구백구십구경 구천구백구십구조 구천구백구십구억 구천구백구십구만 구천구백구십구

---

This is code-golf, so shortest code in bytes wins.