python操作json的方法有
json.dumps——将json对象（字典）转换为字符串对象
json.loads——将字符串对象转换为json对象（字典）
如果定义json对象
jsonstring1={"results":[{"id":"1","name":"\u9ed8\u8ba4\u5206\u7ec4","policy":"4","timer_scan_setting":"{\"last\":\"10.29.13\",\"setting\":\"fulldisk\",\"type\":\"day\",\"hour\":\"13\"}"},
{"id":"2","name":"\u6d4b\u8bd5\u7684","policy":"1","timer_scan_setting":"{\"last\":\"10.29.15\",\"setting\":\"fulldisk\",\"type\":\"day\",\"hour\":\"15\"}"},{"id":"4","name":"\u4ea7\u54c1\u7ec4","policy":"3","timer_scan_setting":"{\"last\":\"10.8.15\",\"setting\":\"disable\"}"}]}

可以直接按json操作，比如

print jsonstring1.keys()
print jsonstring1['results'][0]['policy']

也可以转个360度再操作

jsonstring1=json.dumps(jsonstring1)
jsonstring1=json.loads(jsonstring1)
print jsonstring1.keys()
print jsonstring1['results'][0]['policy']

但是如果定义字符串对象就要小心了
jsonstring2='''{"results":[{"id":"1","name":"\u9ed8\u8ba4\u5206\u7ec4","policy":"4","timer_scan_setting":"{\"last\":\"10.29.13\",\"setting\":\"fulldisk\",\"type\":\"day\",\"hour\":\"13\"}"},
{"id":"2","name":"\u6d4b\u8bd5\u7684","policy":"1","timer_scan_setting":"{\"last\":\"10.29.15\",\"setting\":\"fulldisk\",\"type\":\"day\",\"hour\":\"15\"}"},{"id":"4","name":"\u4ea7\u54c1\u7ec4","policy":"3","timer_scan_setting":"{\"last\":\"10.8.15\",\"setting\":\"disable\"}"}]}'''

这个只是对上面的json对象加了三引号转为一个字符串，所以理论上可以直接loads之后就按json操作

json.loads(jsonstring2)

但是实际却报错了，原因是因为大括号前后的双引号没有去掉，网上很多在线的json格式化工具对于这些双引号不会报错，但是python会，而定义json对象时加了双引号操作却没报错，是因为不加里面的内容会被转义，所以千万不能完全相信网上的json格式校验工具。