import json, datetime, sys, re

# usage: python step3.py [YYYY-MM-DD]   (default = today)
if len(sys.argv) > 1:
    rd = datetime.datetime.strptime(sys.argv[1], '%Y-%m-%d')
else:
    n = datetime.datetime.now()
    rd = datetime.datetime(n.year, n.month, n.day)

start = int(rd.timestamp())
end = start + 86400
print('reportDay=%s window: %d ~ %d' % (rd.strftime('%Y-%m-%d'), start, end))


def ok(t):
    t = (t or '').strip()
    if len(t) < 8:
        return False, 'too_short'
    has_cn = bool(re.search(r'[\u4e00-\u9fff]', t))
    if not has_cn and len(t) < 20:
        return False, 'short_en'
    if not has_cn and t[-1] not in '.!?':
        return False, 'en_incomplete'
    if t[-1] in '，,：:（(、':
        return False, 'truncated'
    return True, 'ok'


d = json.load(open('.weread/tmp/books_detail.json', encoding='utf-8'))
kept = []
for bid, v in d.items():
    marks = (v.get('marks') or {}).get('updated') or []
    for m in marks:
        ct = m.get('createTime')
        if isinstance(ct, str):
            try:
                ct = int(ct)
            except Exception:
                ct = 0
        ct = ct or 0
        ts = datetime.datetime.fromtimestamp(ct).strftime('%Y-%m-%d %H:%M') if ct else 'NA'
        inday = start <= ct < end
        text = (m.get('markText') or '').strip()
        passed, why = ok(text)
        print('--- %s | %s | %s | quality=%s(%s)' % (
            v['title'][:24], ts, 'IN_WINDOW' if inday else 'out', passed, why))
        print('    chapter:', m.get('chapterTitle'))
        print('    text:', repr(text[:160]))
        if inday and passed:
            kept.append({'bookId': bid, 'title': v['title'], 'author': v.get('author'),
                         'chapter': m.get('chapterTitle'), 'text': text, 'createTime': ct})

json.dump(kept, open('.weread/tmp/today_marks.json', 'w', encoding='utf-8'), ensure_ascii=False, indent=1)
print('\nKEPT_TODAY_MARKS =', len(kept))