bs4的使用

#pip install bs4
#pip install lxml


from bs4 import BeautifulSoup

html = '''
<title id="title">度梦</title>
<div class="info" float="left">Welcome to 度梦</div>
<div class="info" float="right">
    <span>Good Good study</span>
    <a href="www.baidu.com"></a>
    <strong>
        <!--这是个注释-->
    </strong>
</div>
'''
soup = BeautifulSoup(html,'lxml')
print('-----------获取标签---------')#只会获取第一个标签，不用考虑层级关系
print(soup.title)
print(soup.div)
print(soup.span)
print('--------------获取属性--------------')
print(soup.div.attrs)
print(soup.div.get('class'))
print(soup.div['float'])
print(soup.a.get('href'))
print('---------------获取内容-------------')
print(soup.title.string)
print(soup.title.text)
print(type(soup.title.string))
print(type(soup.title.text))
print('----------------获取内容注释-----------')
print(soup.strong.string)
print(soup.strong.text)
print(type(soup.strong.string))
print(type(soup.strong.text))
print(soup.strong.prettify())

print('-----------find_all()---------')
print(soup.find_all('div'))
print(soup.fing_all(id='tittle'))
print('------------css选择器---------')
print(soup.select('div'))
print(soup.select('#title'))