csdn复制脚本

import re
obj1=re.compile(r'<div.*?>(?P<dic_content>.*?)/div>')
obj=re.compile(r'>(?P<content>.*?)<')
for line in open('aa.txt'):
    line=line.strip()
    if line:
        for it in obj1.finditer(line):
            line_contents=[]
            dic_content=it.group('dic_content')
            for it_nei in obj.finditer(dic_content):
                line_content=it_nei.group('content')
                if line_content:
                    line_contents.append(line_content)
            content=''.join(line_contents).strip()
            if content:
                print content