Download Wikipedia Kana Stroke Animation Gifs

download_gifs.py

from bs4 import BeautifulSoup
from urllib import request, parse

BASE_FILE_URL = 'https://commons.wikimedia.org/wiki/File:'
# First param is {Hiragana, Katakana}. Second param is {い, か}
BASE_FILE_NAME = '{}_{}_stroke_order_animation.gif'

HIRAGANA = 'Hiragana'
KATAKANA = 'Katakana'
KANA = [
    "あ", "ア", "か", "カ", "さ", "サ", "た", "タ", "な", "ナ", "は", "ハ",
    "ま", "マ", "や", "ヤ", "ら", "ラ", "わ", "ワ", "い", "イ", "き", "キ",
    "し", "シ", "ち", "チ", "に", "ニ", "ひ", "ヒ", "み", "ミ", "り", "リ",
    "う", "ウ", "く", "ク", "す", "ス", "つ", "ツ", "ぬ", "ヌ", "ふ", "フ",
    "む", "ム", "ゆ", "ユ", "る", "ル", "え", "エ", "け", "ケ", "せ", "セ",
    "て", "テ", "ね", "ネ", "へ", "ヘ", "め", "メ", "れ", "レ", "お", "オ",
    "こ", "コ", "そ", "ソ", "と", "ト", "の", "ノ", "ほ", "ホ", "も", "モ",
    "よ", "ヨ", "ろ", "ロ", "を", "ヲ"
]


def urlencode(string):
    return parse.quote_plus(string, encoding='utf-8')


def get_link(url, filename):
    html = request.urlopen(url + filename).read()
    soup = BeautifulSoup(html, 'html.parser')
    return soup.find_all('img')[0].get('src')


def download_file(url, filename):
    try:
        res = request.urlopen(url).read()
        new_file = open(filename, 'wb')
        new_file.write(res)
        new_file.close()
        print('Downloaded: ' + filename)
    except Exception as e:
        print(str(e))


def download_gifs(characters, kana_type, folder):
    for c in characters:
        kana_char = urlencode(c)
        url = BASE_FILE_URL
        filename = BASE_FILE_NAME.format(kana_type, kana_char)
        link = get_link(url, filename)
        download_file(link, '{}/{}.gif'.format(folder, c))


def main():
    hiragana_folder = 'img/hiragana'
    katakana_folder = 'img/katakana'
    hiragana = [KANA[i] for i in range(0, len(KANA), 2)]
    katakana = [KANA[i] for i in range(1, len(KANA), 2)]
    download_gifs(hiragana, HIRAGANA, hiragana_folder)
    download_gifs(katakana, KATAKANA, katakana_folder)

if __name__ == '__main__':
    main()