anki_computer_english.py

#!/usr/bin/env python3

import requests
import re
from lxml import etree
import genanki
import tempfile
import sys
import os
import io

FIND_WORD = re.compile(r'<a class="word" href="([^"]+)">([^<]+)</a>')
FIND_SPELL = re.compile(r'<span class="word-spell">([^<]+)</span><span class="word-spell-audio" data-url="([^"]+)"></span>')

css = '''
.card  {
color:#43c18a;
background:#000;
font-family: "Courier New";
text-align:center;
font-size:20px;
font-weight:bold;
}

.word {font-size:36px;}
.spell {font-size:24px;}

.detail {text-align:left;}
'''

qfmt = '''
<div class="word">{{Word}} </div>
<div class="spell">{{Spell}}</div>
{{Mp3}}
'''

afmt = '''
{{FrontSide}}
<hr id="answer">
<div class="detail">
{{Explan}}
</div>
'''
my_model = genanki.Model(
  778899,
  'Computer English',
  fields=[
    {'name': 'Word'},
    {'name': 'Spell'},
    {'name': 'Explan'},
    {'name': 'Mp3'},
  ],
  css = css,
  templates=[
    {
      'name': 'Card1',
      'qfmt': qfmt,
      'afmt': afmt,
    },
  ])

my_deck = genanki.Deck(77889901, 'Computer English')

workDir = sys.argv[1]
os.chdir(workDir)

sounds = []

for i in range(7):
    r = requests.get('https://www.koolearn.com/dict/tag_1953_%d.html' % (i+1))
    for match in FIND_WORD.findall(r.text):
        link, word = match
        print('add', word)
        r = requests.get('https://www.koolearn.com' + link)
        spell_matches = FIND_SPELL.findall(r.text)
        if len(spell_matches) < 2:
            print("ERROR:can't find spell for", word)
            continue

        tree = etree.HTML(r.text)
        rt = tree.xpath("//div[@class='details-content']")
        if len(rt) < 2:
            print("ERROR:can't find detail for", word)
            continue
        detail = ''.join([etree.tostring(child).decode('utf-8') for child in rt[1].iterchildren()])

        mp3 = word + '.mp3'
        sounds.append(mp3)
        if not os.path.isfile(mp3):
            r = requests.get('http://dict.youdao.com/dictvoice?audio=' + word + '&type=1', allow_redirects=True)
            io.open(mp3, 'wb').write(r.content)

        my_note = genanki.Note(model=my_model,
                sort_field=word,
                fields=[word, spell_matches[1][0], detail, '[sound:'+mp3+']'])
        my_deck.add_note(my_note)

my_package = genanki.Package(my_deck)
my_package.media_files = sounds
my_package.write_to_file('output.apkg')