mandarinanki/main.py

import re
from dataclasses import dataclass
import wx
import genanki
import uuid

### Dictionary stuff

line_regex = r"(\S+)\s(\S+)\s\[([^\]]+)\]\s(.+)"

@dataclass
class Word:
    simplified: str
    traditional: str
    pinyin: str
    definitions: list[str]

def parse_cedict(path: str):
    words = []
    with open(path) as file:
        for line in file:
            if line.startswith("#"):
                continue
            else:
                words.append(parse_line(line))
    return words

def parse_line(line: str):
    traditional, simplified, pinyin, definitions = re.match(line_regex, line).groups()
    return Word(
        simplified,
        traditional,
        pinyin,
        [d for d in definitions.split("/") if d],
    )

def build_pinyin_index(words: list[Word]):
    index = {}
    for word in words:
        if word.pinyin in index:
            index[word.pinyin].append(word)
        else:
            index[word.pinyin] = [word]
    return index

def build_traditional_index(words: list[Word]):
    index = {}
    for word in words:
        if word.traditional in index:
            index[word.traditional].append(word)
        else:
            index[word.traditional] = [word]
    return index

## Tone markers

pinyinToneMarks = {
    u'a': u'āáǎà', u'e': u'ēéěè', u'i': u'īíǐì',
    u'o': u'ōóǒò', u'u': u'ūúǔù', u'ü': u'ǖǘǚǜ',
    u'A': u'ĀÁǍÀ', u'E': u'ĒÉĚÈ', u'I': u'ĪÍǏÌ',
    u'O': u'ŌÓǑÒ', u'U': u'ŪÚǓÙ', u'Ü': u'ǕǗǙǛ'
}

def convert_pinyin_callback(match):
    tone= int(match.group(3))
    vowel = match.group(1)
    # for multiple vowels, use first one if it is a/e/o, otherwise use second one
    pos=0
    if len(vowel) > 1 and not vowel[0] in 'aeoAEO':
        pos=1

    return vowel[0:pos]+pinyinToneMarks[vowel[pos]][tone-1]+vowel[pos+1:] + match.group(2)

def convert_pinyin(s):
    return re.sub(r'([aeiou]{1,3})(n?g?r?)([12345])', convert_pinyin_callback, s, flags=re.IGNORECASE)

### Anki stuff

simple_anki_model = genanki.Model(
  1607392319,
  'Simple Model',
  fields=[
    {'name': 'Question'},
    {'name': 'Answer'},
  ],
  templates=[
    {
      'name': 'Card 1',
      'qfmt': '<div style="text-align:center;font-size: 1.4rem">{{Question}}</div>',
      'afmt': '<div style="text-align:center">{{FrontSide}}<hr id="answer">{{Answer}}</div>',
    },
  ])

def generate_card_from_word(word: Word):
    return generate_card(word.traditional, f"{convert_pinyin(word.pinyin)}<br>{word.definitions[0]}")

def generate_card(front: str, back: str):
    return genanki.Note(model=simple_anki_model, fields=[front, back])

def generate_deck(cards: list, name: str):
    d = genanki.Deck(2059400110, name)
    for card in cards:
        d.add_note(card)
    return d

### UI stuff

class MainFrame(wx.Frame):
    def __init__(self):
        # set up the dictionary
        dictionary = parse_cedict("./cedict_1_0_ts_utf-8_mdbg.txt")
        self.pinyin_idx = build_pinyin_index(dictionary)
        self.traditional_idx = build_traditional_index(dictionary)
        # IDs of words selected to be in the deck
        self.selected_words = []
        # IDs of words in results
        self.result_words = []

        super().__init__(parent=None, title='Mandarin Anki', size = (420, 340))
        panel = wx.Panel(self)
        vertical_layout = wx.BoxSizer(wx.VERTICAL)
        columns_layout = wx.BoxSizer(wx.HORIZONTAL)

        # search and results
        self.search = wx.TextCtrl(panel, size = (200, -1))
        self.search.Bind(wx.EVT_TEXT, self.OnKeyTyped)

        self.results = wx.ListBox(panel, size = (200, 200))
        self.results.Bind(wx.EVT_LISTBOX, self.OnWordSelected)

        self.selected = wx.ListBox(panel, size = (200, 200))
        self.selected.Bind(wx.EVT_LISTBOX, self.OnWordRemoved)

        self.deck_name = wx.TextCtrl(panel, size = (200, -1))

        self.create_deck = wx.Button(panel, label = "ank!")
        self.create_deck.Bind(wx.EVT_BUTTON, self.OnAnk)

        vertical_layout.Add(self.search, 0)
        vertical_layout.Add(columns_layout)
        vertical_layout.Add(self.deck_name, 0)
        vertical_layout.Add(self.create_deck, 0)
        columns_layout.Add(self.results, 0, wx.EXPAND)
        columns_layout.Add(self.selected, 0, wx.EXPAND)

        panel.SetSizer(vertical_layout)
        self.Show()

    def OnKeyTyped(self, event):
        if event.GetString():
            if event.GetString() in self.pinyin_idx:
                matches = self.pinyin_idx[event.GetString()]
                self.results.Set([f"{w.traditional} {convert_pinyin(w.pinyin)} {w.definitions[0]}" for w in matches])
                self.result_words = matches
            elif event.GetString() in self.traditional_idx:
                match = self.traditional_idx[event.GetString()]
                self.result_words = [match]
                self.results.Set([f"{match.traditional} {convert_pinyin(match.pinyin)} {match.definitions[0]}"])
            else:
                self.results.Set([])
                self.result_words = []

    def OnWordSelected(self, event):
        word = self.result_words[event.GetEventObject().GetSelection()]
        formatted = f"{word.traditional} {convert_pinyin(word.pinyin)} {word.definitions[0]}"

        if not word in self.selected_words:
            self.selected.InsertItems([formatted], 0)
            self.selected_words.insert(0, word)


    def OnWordRemoved(self, event):
        idx = event.GetEventObject().GetSelection()
        self.selected.SetSelection(-1)
        self.selected.Delete(idx)
        del self.selected_words[idx]

    def OnAnk(self, event):
        cards = [generate_card_from_word(w) for w in self.selected_words]
        deck = generate_deck(cards, self.deck_name.GetValue())
        deck.write_to_file(f"{self.deck_name.GetValue().replace(" ", "-")}.apkg")

if __name__ == '__main__':
    app = wx.App()
    frame = MainFrame()
    app.MainLoop()