def search(self): url = 'http://www.wyl.cc/' home_soup = BeautifulSoup(HttpUtil.request(url)) target_url = home_soup.find(attrs={"class": "entry-title"}).find('a')['href'].encode('utf-8') target_soup = BeautifulSoup(HttpUtil.request(target_url)) content_soup = target_soup.find(attrs={"class": "single-content"}).findAll('p') text = content_soup[0].text url = content_soup[1].find('a')['href'].encode('utf-8') self.__send(text, url)
def search(self): url = 'http://www.wyl.cc/' home_soup = BeautifulSoup(HttpUtil.request(url)) target_url = home_soup.find(attrs={ "class": "entry-title" }).find('a')['href'].encode('utf-8') target_soup = BeautifulSoup(HttpUtil.request(target_url)) content_soup = target_soup.find(attrs={ "class": "single-content" }).findAll('p') text = content_soup[0].text url = content_soup[1].find('a')['href'].encode('utf-8') self.__send(text, url)
def jian_shu_article_retrive(url): html = HttpUtil.request(url) content_soup = BeautifulSoup( html, fromEncoding="utf-8").find(attrs={"class": "article"}) title = content_soup.find(attrs={"class": "title"}).text.encode('utf-8') content = content_soup.find(attrs={"class": "show-content"}) return (title, content)
def chick_search(self): url = 'http://www.59xihuan.cn/' home_soup = BeautifulSoup(HttpUtil.request(url)) content_soup = home_soup.find(attrs={"class": "pic_text1"}) text = content_soup.find('p').text url = url + content_soup.find('img')['bigimagesrc'].encode('utf-8') self.__send(text, url) print ''
def chick_search(self): url = 'http://www.59xihuan.cn/' home_soup = BeautifulSoup(HttpUtil.request(url)) content_soup = home_soup.find(attrs={"class": "pic_text1"}) text = content_soup.find('p').text url = url + content_soup.find('img')['bigimagesrc'].encode('utf-8') self.__send(text, url) print ''
def query_view_by_url(author_url, blog_url): url = 'http://www.jianshu.com/' + author_url soup = BeautifulSoup(HttpUtil.request(url)) blogs_soup = soup.find(attrs={"class": "article-list latest-notes"}).findAll('li') for blog_soup in blogs_soup: blog = service.parse_blog(blog_soup) if blog.url == blog_url: return blog
def send_weibo_pics(text, urls, browser): browser.get('http://m.weibo.cn/mblog') time.sleep(3) browser.find_element_by_id('txt-publisher').send_keys(text.decode('utf-8')) for i, url in enumerate(urls): if i == 0: open(r'/home/1.jpg', 'wb').write(HttpUtil.request(url)) pic_browser = browser.find_element_by_class_name('picupload') pic_browser.send_keys(r'/home/1.jpg') time.sleep(8) else: open(r'/home/1.jpg', 'wb').write(HttpUtil.request(url)) pic_browser = browser.find_element_by_name('pic') pic_browser.send_keys(r'/home/1.jpg') time.sleep(8) os.remove(r'/home/1.jpg') browser.find_element_by_link_text('发送').click() print '发送成功'
def send_weibo_pics(text, urls, browser): browser.get('http://m.weibo.cn/mblog') time.sleep(3) browser.find_element_by_id('txt-publisher').send_keys(text.decode('utf-8')) for i,url in enumerate(urls): if i==0: open(r'/home/1.jpg', 'wb').write(HttpUtil.request(url)) pic_browser = browser.find_element_by_class_name('picupload') pic_browser.send_keys(r'/home/1.jpg') time.sleep(8) else: open(r'/home/1.jpg', 'wb').write(HttpUtil.request(url)) pic_browser = browser.find_element_by_name('pic') pic_browser.send_keys(r'/home/1.jpg') time.sleep(8) os.remove(r'/home/1.jpg') browser.find_element_by_link_text('发送').click() print '发送成功'
def publish_articles(): url = 'http://www.jianshu.com' html = BeautifulSoup(HttpUtil.request(url)) articles_soup = [ parse_article_soup(article) for article in html.find(attrs={"class": "article-list thumbnails"}).findAll('li')] sorted_articles = sorted(articles_soup, key=lambda articl_soup:articl_soup[0], reverse=True) article_body = service.jian_shu_article_retrive(url + sorted_articles[0][1]) browser = LoginController.get_browser() passwd_dict = PasswdUtil.get_passwd_dict() send_text = '《%s》 又到每天推荐文章的时候到了,这些都是精选的枕边读物哦,希望大家喜欢@当幸福敲不开门[害羞][害羞][害羞] ' % article_body[0] service.send_weibo(send_text, service.txt_to_pic(article_body[1], browser), LoginController.mobile_login(browser, passwd_dict[Const.WEIBO_USERNAME], passwd_dict[Const.WEIBO_PASSWD]))
def user_info(user_url): url = 'http://www.jianshu.com' + user_url soup = BeautifulSoup(HttpUtil.request(url)) user = User() base_info_soup = soup.find(attrs={"class": "basic-info"}) user.name = base_info_soup.find('h3').text.encode('utf-8') signed_soup = base_info_soup.find(attrs={"class": "signed_author"}) if signed_soup: user.is_signed = 0 nums_soup = soup.find(attrs={"class": "user-stats"}).findAll('b') user.focus = int(nums_soup[0].text.encode('utf-8')) user.fans = int(nums_soup[1].text.encode('utf-8')) user.blog_nums = int(nums_soup[2].text.encode('utf-8')) user.word_nums = int(nums_soup[3].text.encode('utf-8')) user.like_nums = int(nums_soup[4].text.encode('utf-8')) return user
def publish_articles(): url = 'http://www.jianshu.com' html = BeautifulSoup(HttpUtil.request(url)) articles_soup = [ parse_article_soup(article) for article in html.find(attrs={ "class": "article-list thumbnails" }).findAll('li') ] sorted_articles = sorted(articles_soup, key=lambda articl_soup: articl_soup[0], reverse=True) article_body = service.jian_shu_article_retrive(url + sorted_articles[0][1]) browser = LoginController.get_browser() passwd_dict = PasswdUtil.get_passwd_dict() send_text = '《%s》 又到每天推荐文章的时候到了,这些都是精选的枕边读物哦,希望大家喜欢@当幸福敲不开门[害羞][害羞][害羞] ' % article_body[ 0] service.send_weibo( send_text, service.txt_to_pic(article_body[1], browser), LoginController.mobile_login(browser, passwd_dict[Const.WEIBO_USERNAME], passwd_dict[Const.WEIBO_PASSWD]))
def find_blogs(url): soup = BeautifulSoup(HttpUtil.request(url)) blogs = map(lambda blog_soup:parse_blog(blog_soup), soup.find(attrs={"class": "article-list thumbnails"}).findAll('li')) return filter(lambda blog : blog is not None, blogs)
def jian_shu_article_retrive(url): html = HttpUtil.request(url) content_soup = BeautifulSoup(html, fromEncoding="utf-8").find(attrs={"class": "article"}) title = content_soup.find(attrs={"class": "title"}).text.encode('utf-8') content = content_soup.find(attrs={"class": "show-content"}) return (title, content)