Generate Anki flashcard from web snatch.

everydayenglish.py

import os
import re
import requests
import json
import time
import datetime
import genanki

my_model = genanki.Model(
    201901021920,
    '每日一句',
    fields=[
        {'name': 'content'},
        {'name': 'translation'},
        {'name': 'note'},
        {'name': 'picture'},
        {'name': 'tts'}
    ],
    css='''
</style>
<style>
    body{
        margin:0;padding:0;
        font:14px/1 'Microsoft Yahei',sans-serif,Arial,Verdana;color:#333;
    }
    .detail-banner-img {display: block; border:0;}
    .detail-content-en{font-size:26px;line-height:34px;margin-top:20px}
    .detail-content-zh{font-size:18px;line-height:24px;margin-top:8px}
    .detail-content-desc{margin-top:16px}
    blockquote{line-height:24px;font-size:14px;color:#666}
    blockquote:before{content:'';display:inline-block;background-image:url(_daily.png);background-position:-696px -361px;width:21px;height:14px;top:-6px;}
    .icon-sound,.icon-sound:active,.icon-sound:hover{background-image:url(_daily.png);width:17px;height:16px}
    .icon-sound{display:inline-block;vertical-align:middle;background-position:-687px -386px;cursor:pointer;margin-left:4px}
    .icon-sound:hover{background-position:-670px -386px}.icon-sound:active{background-position:-653px -386px}
</style>
<style>
''',
    templates=[
        {
            'name': '每日一句',
            'qfmt': '''
    {{#picture}}
    {{picture}}
    {{/picture}}
    <div class="detail-content-en">{{content}}<i class="icon-sound">{{tts}}</i> </div>
            ''',
            'afmt': '''
    {{FrontSide}}
    <div class="detail-content-zh">{{translation}}</div> 
    <blockquote class="detail-content-desc">{{note}}</blockquote>
            '''
        },
    ])

my_deck = genanki.Deck(
    201901021952,
    'EveryDayEnglish')


media_files = []

btime = 1325347200
while btime<time.mktime(datetime.datetime.now().timetuple()):
    dt = time.strftime("%Y-%m-%d",time.localtime(btime))
    url = 'http://sentence.iciba.com/index.php?callback=jQuery190020799044250976584_1546069428424&c=dailysentence&m=getdetail&title=%s&_=1546069428428'%(dt,)
    r = requests.get(url)
    w = re.findall('^jQuery[^(]*\((.*)\)',r.text)
    js = json.loads(w[0])
    imgname = None
    ttsname = None
    if js['picture2']:
        imgname = os.path.basename(js['picture2'])
        media_files.append(imgname)
        p = requests.get(js['picture2'])
        with open(imgname, 'wb') as f:
            f.write(p.content)
    if js['tts']:
        ttsname = ''.join([os.path.basename(js['tts']), '.mp3'])
        media_files.append(ttsname)
        p = requests.get(js['tts'])
        with open(ttsname, 'wb') as f:
            f.write(p.content)

    aNote = genanki.Note(model=my_model, fields=[
        js['content'],
        js['note'],
        js['translation'],
        "<img src=\"%s\" />"%(imgname,) if imgname else "",
        "[sound:%s]"%(ttsname,) if ttsname else ""
    ])
    my_deck.add_note(aNote)

    btime = btime+24*3600

genanki.Package(my_deck, media_files=media_files).write_to_file('EveryDayEnglish.apkg')