korakot

# ลง hunspell + python ไม่สำเร็จ
# แต่มาเจอ pyenchant ที่ใช้ได้เหมือนกัน และมี custom word list ด้วย

>>> import enchant
>>> pwl = enchant.request_pwl_dict("royin_2542_wordlist.txt")
>>> pwl.check("กรกฎ")
True
>>> pwl.check("กรกำ")
False
>>> pwl.suggest("กรกำ")

korakot

import icu
thkey = icu.Collator.createInstance(icu.Locale('th_TH')).getSortKey
words = 'ไก่ ไข่ ก ฮา'.split()
print(sorted(words, key=thkey))  # ['ก', 'ไก่', 'ไข่', 'ฮา']

korakot

>>> import enchant
>>> d = enchant.request_pwl_dict("royin_2542_wordlist.txt")
>>> d.check("กรกำ")
False
>>> d.suggest("กรกำ")
['กรก', 'กรกฎ', 'กรกฏ', 'กรกช', 'กรำ', 'กร่ำ']

korakot

# From https://github.com/graphaelli/ipython-elasticsearch

import json
import os, re
import urllib.parse

from IPython.core.magic import Magics, magics_class, line_cell_magic
import requests

korakot

import re
from collections import defaultdict
from marisa_trie import Trie

wordlist = [li.strip() for li in open('wordlist.txt')]
trie = Trie(wordlist)   # สร้างครั้งเดียว ข้างนอก function

class LatticeString(str):
    ''' String subclass เพื่อเก็บวิธีตัดหลายๆ วิธี
    '''

korakot

print('.', end='\u200b')
# here \u200b is the zero-width space. It helps wrapping lines.

korakot

# https://medium.com/earthchie/9af96b5f5588

num = '3-0000-00000-00-5'
nums = [int(i) for i in num.replace('-','')] # remove dash

prod = sum(i*j for i,j in zip(nums, range(13,1,-1)))
x = (11-prod%11)%10

assert x==nums[-1]  # must equal 5

korakot

import re
s = re.sub('ํ([่-๋]?)า','\\1ำ', s)  # กรณี   ํ + า
s = re.sub('ำ([่-๋])', '\\1ำ', s)  # กรณี   ำ + ้
s=='น้ำ'

korakot

from marisa_trie import Trie

# wordlist = ...
trie = Trie(wordlist)

def lmcut(text):
  for w in reversed(trie.prefixes(text)):
    if w==text:
      yield [w]
    else:

korakot

from __future__ import unicode_literals    # at top of module

s1 = 'The Zen of Python'  # unicode for both python 2,3
b2 = b'xxxxxxxxxxxxxx'   # byte string for both py 2,3

b1 = s1.encode()   # to bytes, using utf-8 to encode
s2 = b2.decode()   # now become unicode