Create Sentence Cards from Books (WIP) - Better japanese parsing

__init__.py

import sys
import io
import os

from aqt import mw
from aqt.utils import getFile, tooltip
from aqt.qt import *


sys.path.append(os.path.join(os.path.dirname(__file__), "vendor"))

'''
from importlib import reload
gd = __import__('1200382751')
reload(gd)

'''

def main():
    file = getFile(mw, "Import", None, filter="*.txt", key="1200382751")
    if not file:
        return

    # mw.progress.start(immediate=True)

    import nltk

    with open(file, 'r', encoding='utf-8') as f_txt:
        text = f_txt.read()
    
    mw.progress.start(label='Parsing Sentences...\n ', max=len(text.splitlines()), min=1, immediate=True)
    
    cnt = 0
    lines = text.splitlines()
    for i, p in enumerate(lines):
        mw.progress.update(value=cnt+1, label=f'Parsing Sentences...\n{int(i)}/{len(lines)} lines')
        if not p.strip():
            continue
        tokenizer = nltk.RegexpTokenizer(u'[^ 「」!?。．）]*[!?。]')
        # sents = nltk.sent_tokenize(p)
        sents = tokenizer.tokenize(p)
        filename, ext = os.path.splitext(os.path.basename(file))
        
        did = mw.col.decks.id(f'!優先::Y メディア::本::{filename}')
        mw.col.decks.select(did)
        
        m = mw.col.models.byName("Japanese")
        deck = mw.col.decks.get(did)
        deck['mid'] = m['id']
        mw.col.decks.save(deck)

        m['did'] = did
        mw.col.models.save(m)
        

        for s in sents:
            n = mw.col.newNote()
            n["Expression"] = s.strip()
            n["Meta"] = f'{i}/{len(lines)}'
            n.addTag('mm')
            mw.col.addNote(n)
            cids = [c.id for c in n.cards()]
            mw.col.sched.suspendCards(cids)
            cnt += 1
        
    mw.progress.finish()
    mw.reset()

    tooltip("Imported {} notes.".format(cnt))

action = QAction("Create Sentence Cards", mw)
action.triggered.connect(main)
mw.form.menuTools.addAction(action)