v3.5.5

Hot Fix

• Modify mecab installation script

v3.5.4

• Add auto for the number of workers

v3.5.3

Hot Fix

• modify mecab installation script

v3.5.1

3.5.1

• Use mecab backend as default if mecab is available else use pynori as default.
• Set num_workers to 1 when the input was string.

v3.4.3

Fix bracket processing related bugs

v3.4.2

Quick fix of 3.4.1.

v3.4.1

• Apply improvements from #39.

v3.4

Fix bracket processing related bugs

• #28
• #29

Modify caching key

• #30
• #33

v3.3.1

Do not use multiprocessing if length of input list is 1.

v3.3.0

• Fix emoji bug reported in #23.
• Add improvements reported in #24.
  • Lazy initialization of Pynori.
  • Do not access multiprocessing module when num_workers was 0 or 1.
• Add none backend for users that want to fast segmentation without mecab.
  • none backend works like kss 2.5.1 (do not use morpheme analyzer)
  • But if you use none backend, performance will be decreased.

  >>> split_sentences(text, backend="pynori")
  ['부디 만수무강 하옵소서', '천천히 가세용~', '너 밥을 먹는구나', '응 맞아 난 근데 어제 이사했음', '그랬구나 이제 마지막임', '응응']

  >>> split_sentences(text, backend="mecab")
  ['부디 만수무강 하옵소서', '천천히 가세용~', '너 밥을 먹는구나', '응 맞아 난 근데 어제 이사했음', '그랬구나 이제 마지막임', '응응']

  >>> split_sentences(text, backend="none")  # <--- it is same with kss 2.5.1 
  ['부디 만수무강 하옵소서 천천히 가세용~', '너 밥을 먹는구나 응 맞아 난 근데 어제 이사했음 그랬구나 이제 마지막임 응응'