dereadnator.py

# -*- coding: utf-8 -*-

import pandas as pd
import re
from core import Words, ankiFurigana

def hiraToKataCh(ch):
    if ord(ch) >= 0x3041 and ord(ch) <= 0x3096:
        return unichr(ord(ch) + 0x60)
    else:
        return ch

def hiraToKata(s):
    return "".join([hiraToKataCh(ch) for ch in s])

# Receives a .csv file with three columns (the third will be filled/overwritten)
# Replaces the pronunciation of the kanji between parenthesis by the kanji that
# is in the first column.

def main():
  with open('jkeywords.csv') as f:
    df = pd.read_csv(f, sep='\t', names=["kanji", "oldkey", "newkey"],
                     encoding='utf8')
  df.fillna('', inplace=True) # make empty fields strings too, not floats.

  #load dictionary
  dic = Words({"pathDicFile":"jmdict_freqs.txt"})._dic

  # select the normal or japanese parenthesis and it's contents only if it only
  # have kanas in between them.
  regex = re.compile(u"[（|(][ぁ-ー]+[）|)]", re.UNICODE)
  for index, row in df.iterrows():
    ok = row['oldkey']
    if ok == "":
      continue
    kanji = row['kanji']
    s = ""
    for word in ok.split(u"、"):  # japanese comma
      kanjiword, n = regex.subn(kanji, word)
      readings = []
      if n == 0:
        kanjiword = kanji  # when the reading is for the whole kanji alone
        readings = [word]
      elif kanjiword in dic:
        readings = dic[kanjiword].keys()
        if len(readings) > 1:
          targetReadings = [hiraToKata(t) for t in regex.findall(word)]
          #are there any with len(targetReadings) > 1?
          #if so, maybe build another regex to check they are all there and in the correct order
          readings = [r for r in readings if targetReadings[0][1:-1] in hiraToKata(r)]
      answers = u"、".join([ankiFurigana(kanjiword, r) for r in readings])
      if len(readings) < 1:
        answers = u"NOTFOUND"
      if len(readings) > 1:
        answers += u"、CHECK(" + str(len(readings)) + ")"
      questions = answers.replace(kanji, u"〇")
      s += questions + u"、"
      #could put the answers in the deck too, like in the plugin
    row['newkey'] = s[:-1]  # remove the trailing comma

  df.to_csv('processed.csv', sep='\t', header=False, index=False, encoding='utf8')

main()