http://rsj217.diandian.com/post/2012-11-01/40041235132
Beautiful Soup 是用 Python 写的一个 HTML/XML 的解析器,它可以很好的处理不规范标记并生成剖析树。通常用来分析爬虫抓取的web文档。对于 不规则的 Html文档,也有很多的补全功能,节省了开发者的时间和精力。
Beautiful Soup 的官方文档齐全,将官方给出的例子实践一遍就能掌握。官方英文文档,中文文档
一 安装 Beautiful Soup
安装 BeautifulSoup 很简单,下载 BeautifulSoup 源码。解压运行
python setup.py install 即可。
测试安装是否成功。键入 import BeautifulSoup 如果没有异常,即成功安装
二 使用 BeautifulSoup
1. 导入BeautifulSoup ,创建BeautifulSoup 对象
1
2
3
4
5
6
7
8
9
10
11
12
|
from BeautifulSoup import BeautifulSoup # HTML
from BeautifulSoup import BeautifulStoneSoup # XML
import BeautifulSoup # ALL
doc = [
'<html><head><title>Page title</title></head>' ,
'<body><p id="firstpara" align="center">This is paragraph <b>one</b>.' ,
'<p id="secondpara" align="blah">This is paragraph <b>two</b>.' ,
'</html>'
] # BeautifulSoup 接受一个字符串参数 soup = BeautifulSoup(''.join(doc))
|
2. BeautifulSoup对象简介
用BeautifulSoup 解析 html文档时,BeautifulSoup将 html文档类似 dom文档树一样处理。BeautifulSoup文档树有三种基本对象。
2.1. soup BeautifulSoup.BeautifulSoup
1
2
|
type (soup)
< class 'BeautifulSoup.BeautifulSoup' >
|
2.2. 标记 BeautifulSoup.Tag
1
2
|
type (soup.html)
< class 'BeautifulSoup.Tag' >
|
2.3 文本 BeautifulSoup.NavigableString
1
2
|
type (soup.title.string)
< class 'BeautifulSoup.NavigableString' >
|
3. BeautifulSoup 剖析树
3.1 BeautifulSoup.Tag对象方法
获取 标记对象(Tag)
标记名获取法 ,直接用 soup对象加标记名,返回 tag对象.这种方式,选取唯一标签的时候比较有用。或者根据树的结构去选取,一层层的选择
1
2
3
4
5
6
7
|
>>> html = soup.html >>> html < html >< head >< title >Page title</ title ></ head >< body >< p id = "firstpara" align = "center" >This is paragraph< b >one</ b >.</ p >< p id = "secondpara" align = "blah" >This is paragraph< b >two</ b >.</ p ></ body ></ html >
>>> type(html) < class 'BeautifulSoup.Tag'>
>>> title = soup.title < title >Page title</ title >
|
content方法
content方法 根据文档树进行搜索,返回标记对象(tag)的列表
1
2
|
>>> soup.contents [<html><head><title>Page title< / title>< / head><body><p id = "firstpara" align = "center" >This is paragraph<b>one< / b>.< / p><p id = "secondpara" align = "blah" >This is paragraph<b>two< / b>.< / p>< / body>< / html>]
|
1
2
3
4
5
6
|
>>> soup.contents[ 0 ].contents
[<head><title>Page title< / title>< / head>, <body><p id = "firstpara" align = "center" >This is paragraph<b>one< / b>.< / p><p id = "secondpara" align = "blah" >This is paragraph<b>two< / b>.< / p>< / body>]
>>> len (soup.contents[ 0 ].contents)
2 >>> type (soup.contents[ 0 ].contents[ 1 ])
< class 'BeautifulSoup.Tag' >
|
使用contents
向后遍历树,使用parent
向前遍历树
next 方法
获取树的子代元素,包括 Tag 对象 和 NavigableString 对象。。。
1
2
3
4
|
>>> head. next
<title>Page title< / title>
>>> head. next . next
u 'Page title'
|
1
2
3
4
5
|
>>> p1 = soup.p
>>> p1 <p id = "firstpara" align = "center" >This is paragraph<b>one< / b>.< / p>
>>> p1. next
u 'This is paragraph'
|
nextSibling 下一个兄弟对象 包括 Tag 对象 和 NavigableString 对象
1
2
3
4
|
>>> head.nextSibling <body><p id = "firstpara" align = "center" >This is paragraph<b>one< / b>.< / p><p id = "secondpara" align = "blah" >This is paragraph<b>two< / b>.< / p>< / body>
>>> p1. next .nextSibling
<b>one< / b>
|
与 nextSibling 相似的是 previousSibling,即上一个兄弟节点。
replacewith方法
将对象替换为,接受字符串参数
1
2
3
4
5
6
7
8
9
10
11
12
|
>>> head = soup.head
>>> head <head><title>Page title< / title>< / head>
>>> head.parent <html><head><title>Page title< / title>< / head><body><p id = "firstpara" align = "center" >This is paragraph<b>one< / b>.< / p><p id = "secondpara" align = "blah" >This is paragraph<b>two< / b>.< / p>< / body>< / html>
>>> head.replaceWith( 'head was replace' )
>>> head <head><title>Page title< / title>< / head>
>>> head.parent >>> soup <html>head was replace<body><p id = "firstpara" align = "center" >This is paragraph<b>one< / b>.< / p><p id = "secondpara" align = "blah" >This is paragraph<b>two< / b>.< / p>< / body>< / html>
>>> |
搜索方法
搜索提供了两个方法,一个是 find,一个是findAll。这里的两个方法(findAll和 find)仅对Tag对象以及,顶层剖析对象有效,但 NavigableString不可用。
findAll(
name, attrs, recursive, text, limit, **kwargs)
接受一个参数,标记名
寻找文档所有 P标记,返回一个列表
1
2
3
4
|
>>> soup.findAll( 'p' )
[<p id = "firstpara" align = "center" >This is paragraph<b>one< / b>.< / p>, <p id = "secondpara" align = "blah" >This is paragraph<b>two< / b>.< / p>]
>>> type (soup.findAll( 'p' ))
< type 'list' >
|
寻找 id="secondpara"的 p 标记,返回一个结果集
1
2
3
|
>>> pid = type (soup.findAll( 'p' , id = 'firstpara' ))
>>> pid < class 'BeautifulSoup.ResultSet' >
|
传一个属性或多个属性对
1
2
3
4
5
|
>>> p2 = soup.findAll( 'p' ,{ 'align' : 'blah' })
>>> p2 [<p id = "secondpara" align = "blah" >This is paragraph<b>two< / b>.< / p>]
>>> type (p2)
< class 'BeautifulSoup.ResultSet' >
|
利用正则表达式
1
2
|
>>> soup.findAll( id = re. compile ( "para$" ))
[<p id = "firstpara" align = "center" >This is paragraph<b>one< / b>.< / p>, <p id = "secondpara" align = "blah" >This is paragraph<b>two< / b>.< / p>]
|
读取和修改属性
1
2
3
4
5
6
7
8
9
10
11
12
|
>>> p1 = soup.p
>>> p1 <p id = "firstpara" align = "center" >This is paragraph<b>one< / b>.< / p>
>>> p1[ 'id' ]
u 'firstpara'
>>> p1[ 'id' ] = 'changeid'
>>> p1 <p id = "changeid" align = "center" >This is paragraph<b>one< / b>.< / p>
>>> p1[ 'class' ] = 'new class'
>>> p1 <p id = "changeid" align = "center" class = "new class" >This is paragraph<b>one< / b>.< / p>
>>> |
剖析树基本方法就这些,还有其他一些,以及如何配合正则表达式。具体请看官方文档
3.2 BeautifulSoup.NavigableString对象方法
NavigableString 对象方法比较简单,获取其内容
1
2
3
4
5
6
7
8
9
|
>>> soup.title <title>Page title< / title>
>>> title = soup.title. next
>>> title u 'Page title'
>>> type (title)
< class 'BeautifulSoup.NavigableString' >
>>> title.string u 'Page title'
|
至于如何遍历树,进而分析文档,已经 XML 文档的分析方法,可以参考官方文档。