Format

Header (16B)
Displacements (n_disps * 8B)
Entries
Entry data

Header

key (8B)
n_disps (4B)
n_entries (4B)

Displacement

a (4B)
b (4B)

Entry

i (3B)
k_len (1B)
v_len (1B)

Key encoding

The key (kanji) is utf16-encoded to save 1byte per char

Value encoding

Characters:

0xxxxxxx: Hiragana (unicode character: 0x3041 + x) 01111111: Hiragana prolonged sound mark (ー, unic: 0x30fc) 10000000: Separator between reading and context (0x80) 11111111: | Separator between readings (0xff) 1xxxxxxx: Tail (ASCII character)

The default reading should be placed at the end as it has the lowest priority. This way we can stop iterating over the readings once a match is found.

with tails:

"描": {
  "k": "か",
  "": "びょう",
  "i": "か",
}

kか|iか|びょう

with context:

"色": {
  "": "いろ",
  "そん": "しょく",
}

しょくSそん|いろ

948 B Raw Permalink Blame History