29 lines
769 B
XML
29 lines
769 B
XML
<!--This is a draft of config file.
|
|
If you have any advice, go https://github.com/code4craft/webmagic/issues/106 and comment!-->
|
|
<spider>
|
|
<site>
|
|
<charset>utf-8</charset>
|
|
<user-agent></user-agent>
|
|
<cookies>
|
|
<cookie domain="" path="" name="" value="">
|
|
</cookie>
|
|
</cookies>
|
|
<heads>
|
|
<head name="" value=""/>
|
|
</heads>
|
|
</site>
|
|
|
|
<startUrls>
|
|
<url></url>
|
|
</startUrls>
|
|
|
|
<extraction targetUrl="" helpUrl="">
|
|
<field name="title">
|
|
<extractor type="xpath" value="//div[@class='title']"/>
|
|
</field>
|
|
<field name="content">
|
|
<extractor type="xpath" value="//div[@class='content']"/>
|
|
</field>
|
|
</extraction>
|
|
|
|
</spider> |