Esempio n. 1
0
def collect_notice(stock_info):
    req_url = company_accouncement.format(stock_info.stock_number)
    notice_list_html = send_request(req_url)
    notice_list_soup = BeautifulSoup(notice_list_html, 'lxml')

    notice_list = notice_list_soup.find('div', class_='cont').find_all('li')
    for i in notice_list:
        notice_title = i.find('span', class_='title').text
        notice_cate = i.find('span', class_='cate').text
        notice_date_str = i.find('span', class_='date').text
        notice_date = datetime.datetime.strptime(notice_date_str, '%Y-%m-%d')
        notice_url = eastmoney_data + i.find('a').get('href')

        if not check_duplicate(notice_title, notice_cate, notice_date):
            notice_html = send_request(notice_url)
            notice_soup = BeautifulSoup(notice_html, 'lxml')
            notice_content = notice_soup.find('pre').text

            stock_notice = StockNotice(stock_number=stock_info.stock_number,
                                       stock_name=stock_info.stock_name,
                                       notice_title=notice_title,
                                       notice_cate=notice_cate,
                                       notice_date=notice_date,
                                       notice_url=notice_url,
                                       notice_content=notice_content)
            stock_notice.save()
            time.sleep(random.random())
def collect_notice(stock_number, stock_name):
    req_url = company_accouncement.format(stock_number)
    notice_list_html = send_request(req_url)
    notice_list_soup = BeautifulSoup(notice_list_html, "lxml")

    notice_list = notice_list_soup.find("div", class_="cont").find_all("li")
    for i in notice_list:
        notice_title = i.find("span", class_="title").text
        notice_cate = i.find("span", class_="cate").text
        notice_date_str = i.find("span", class_="date").text
        notice_date = datetime.datetime.strptime(notice_date_str, "%Y-%m-%d")
        notice_url = eastmoney_data + i.find("a").get("href")

        if not check_duplicate(notice_title, notice_cate, notice_date):
            notice_html = send_request(notice_url)
            notice_soup = BeautifulSoup(notice_html, "lxml")
            notice_content = notice_soup.find("pre").text

            stock_notice = StockNotice(
                stock_number=stock_number,
                stock_name=stock_name,
                notice_title=notice_title,
                notice_cate=notice_cate,
                notice_date=notice_date,
                notice_url=notice_url,
                notice_content=notice_content,
            )
            stock_notice.save()
            time.sleep(random.random())
def collect_notice(stock_info):
    req_url = company_notice.format(stock_info.stock_number)
    raw_data = send_request(req_url).replace('var', '').replace('=', '').replace(';', '').strip()
    notice_data = json.loads(raw_data).get('data', [])

    for n in notice_data:
        notice_title = n.get('NOTICETITLE')
        notice_code = n.get('INFOCODE')
        notice_date = datetime.datetime.strptime(n.get('NOTICEDATE').split('T')[0], '%Y-%m-%d')
        notice_url = single_notice.format(stock_info.stock_number, notice_code)

        if not is_exists(notice_code):
            stock_notice = StockNotice(title=notice_title, code=notice_code, date=notice_date,
                content_url=notice_url, stock_number=stock_info.stock_number, stock_name=stock_info.stock_name)
            stock_notice.save()
def collect_notice(stock_info):
    req_url = company_notice.format(stock_info.stock_number)
    raw_data = send_request(req_url).replace('var',
                                             '').replace('=', '').replace(
                                                 ';', '').strip()
    notice_data = json.loads(raw_data).get('data', [])

    for n in notice_data:
        notice_title = n.get('NOTICETITLE')
        notice_code = n.get('INFOCODE')
        notice_date = datetime.datetime.strptime(
            n.get('NOTICEDATE').split('T')[0], '%Y-%m-%d')
        notice_url = single_notice.format(stock_info.stock_number, notice_code)

        if not is_exists(notice_code):
            stock_notice = StockNotice(title=notice_title,
                                       code=notice_code,
                                       date=notice_date,
                                       content_url=notice_url,
                                       stock_number=stock_info.stock_number,
                                       stock_name=stock_info.stock_name)
            stock_notice.save()
Esempio n. 5
0
def collect_notice(stock_info):
    req_url = company_accouncement.format(stock_info.stock_number)
    notice_list_html = send_request(req_url)
    notice_list_soup = BeautifulSoup(notice_list_html, 'lxml')

    notice_list = notice_list_soup.find('div', class_='cont').find_all('li')
    for i in notice_list:
        notice_title = i.find('span', class_='title').text
        notice_cate = i.find('span', class_='cate').text
        notice_date_str = i.find('span', class_='date').text
        notice_date = datetime.datetime.strptime(notice_date_str, '%Y-%m-%d')
        notice_url = eastmoney_data + i.find('a').get('href')

        if not check_duplicate(notice_title, notice_cate, notice_date):
            notice_html = send_request(notice_url)
            notice_soup = BeautifulSoup(notice_html, 'lxml')
            notice_content = notice_soup.find('pre').text

            stock_notice = StockNotice(stock_number=stock_info.stock_number, stock_name=stock_info.stock_name,
                                       notice_title=notice_title, notice_cate=notice_cate, notice_date=notice_date,
                                       notice_url=notice_url, notice_content=notice_content)
            stock_notice.save()
            time.sleep(random.random())