一次性能优化（主要是sql查询优化，把or 改成 union all）

末蓝、 2022-12-26 04:50 81阅读 0赞

1、原代码如下：

（1）、业务controller层，调用了3次service方法访问数据： ordersService.getBussOrders ( ...)

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70][]

（2）、而ordersService中的，service层调用了访问数据库方法，在ordersDao.listBussPOrders()方法里面，最后执行的是mybatis查询

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 1][]

（3）mybatis中的\*mapper.xml文件：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 2][]

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 3][]

原有的业务含义应该是，分别在以下3个参数有值的时候departId、sponsorId、jointlyId，访问了3次数据库查询，然后把结果集拼接起来到一个集合中。

（4）、执行的完整sql如下：

SELECT
    	* 
    FROM
    	(
    SELECT
    	B.id,
    	A.buss_no,
    	A.task_type,
    	A.task_status,
    	A.start_date,
    	B.create_time,
    	A.sponsor_id,
    	A.sponsor_name,
    	A.jointly_name,
    	A.raise_funds,
    	A.lease_amount,
    	B.owner_id,
    	B.owner_name,
    	B.cust_name,
    	C.start_date c_start_date,
    CASE
    	
    	WHEN (
    	GROUP_CONCAT( H1.cust_name ) IS NOT NULL 
    	AND LENGTH( trim( GROUP_CONCAT( H1.cust_name ) ) ) > 0 
    	AND GROUP_CONCAT( H2.com_fullname ) IS NOT NULL 
    	AND LENGTH( trim( GROUP_CONCAT( H2.com_fullname ) ) ) > 0 
    	) THEN
    		CONCAT( GROUP_CONCAT( H1.cust_name ), ',', GROUP_CONCAT( H2.com_fullname ) ) ELSE ifnull( GROUP_CONCAT( H1.cust_name ), GROUP_CONCAT( H2.com_fullname ) ) 
    	END 'hgMan',
    	GROUP_CONCAT( L.delivery_place ) AS deliverPlace,
    	DATE_FORMAT( W.create_time, '%Y-%m-%d %H:%i:%s' ) AS auditTime 
    FROM
    	orders A
    	INNER JOIN wf_task B ON A.buss_no = B.buss_no
    	LEFT JOIN contract_clause AS C ON A.buss_no = C.buss_no
    	LEFT JOIN cust_nature AS H1 ON A.buss_no = H1.buss_no 
    	AND H1.buss_type = 'HG' 
    	AND H1.STATUS = '1'
    	LEFT JOIN cust_legal AS H2 ON A.buss_no = H2.buss_no 
    	AND H2.buss_type = 'HG' 
    	AND H2.STATUS = '1'
    	LEFT JOIN lease_info AS L ON A.buss_no = L.buss_no 
    	AND L.lease_type = '1'
    	LEFT JOIN wf_task AS W ON A.buss_no = W.buss_no 
    	AND W.task_type = '30' 
    	AND W.is_current = "Y" 
    WHERE
    	B.is_current = "Y" 
    	AND A.STATUS = "1" 
    	AND A.task_type = B.task_type 
    	AND A.task_status = B.STATUS 
    	
    GROUP BY
    	A.buss_no 
    ORDER BY
    	B.create_time DESC 
    	) AS SUB 
    GROUP BY
    	SUB.buss_no 
    ORDER BY
    SUB.create_time DESC

（在最坏情况下，当departId、sponsorId、jointlyId都没有值，sql未拼接字段的时候），单条sql执行时间2.5秒左右：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 4][]

如果说在步骤1、（1）代码中，查询3次，那么光是3行代码执行查询3次sql的时间，就7.5秒了，再加上其它业务处理，前后端交互响应时间，网络延时，前端页面加载数据，就会变得很慢了。

（5）、如下是我在生产环境中，翻页查询，调用该接口，总响应时间，几次消耗总时间都在10秒以上：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 5][]

2、优化：

（1）、先优化代码，只和数据库查询交互一次：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 6][]

（2）、优化sql，只查询一次以后，需要sql把3种情况的并集，都返回，于是首先想到了or用法：

先把子查询中的order by 去掉，因为最外层有order by 排序了，没有必要两次排序：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 7][]

改mybatis，把之前的and改成or：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 8][]

sql代码改成or以后如下：

测试执行时间，在1.5秒左右

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 9][]

（3）如果不用or，还有一种方案，就是换成union all，测试一下执行时间，

sql代码如下：

SELECT
    	SUB.*,
    	GROUP_CONCAT( LS.delivery_place ) AS deliverPlaces 
    FROM
    	(
    SELECT
    	B.id,
    	A.buss_no,
    	A.task_type,
    	A.task_status,
    	A.start_date,
    	B.create_time,
    	A.depart_id,
    	A.sponsor_id,
    	A.jointly_id,
    	A.sponsor_name,
    	A.jointly_name,
    	A.raise_funds,
    	A.lease_amount,
    	B.owner_id,
    	B.owner_name,
    	B.cust_name,
    	C.start_date c_start_date,
    CASE
    	
    	WHEN (
    	GROUP_CONCAT( DISTINCT H1.cust_name ) IS NOT NULL 
    	AND LENGTH( trim( GROUP_CONCAT( DISTINCT H1.cust_name ) ) ) > 0 
    	AND GROUP_CONCAT( DISTINCT H2.com_fullname ) IS NOT NULL 
    	AND LENGTH( trim( GROUP_CONCAT( DISTINCT H2.com_fullname ) ) ) > 0 
    	) THEN
    		CONCAT( GROUP_CONCAT( DISTINCT H1.cust_name ), ',', GROUP_CONCAT( DISTINCT H2.com_fullname ) ) ELSE ifnull( GROUP_CONCAT( DISTINCT H1.cust_name ), GROUP_CONCAT( DISTINCT H2.com_fullname ) ) 
    	END 'hgMan',
    	DATE_FORMAT( W.create_time, '%Y-%m-%d %H:%i:%s' ) AS auditTime 
    FROM
    	orders A
    	INNER JOIN wf_task B ON A.buss_no = B.buss_no
    	LEFT JOIN contract_clause AS C ON A.buss_no = C.buss_no
    	LEFT JOIN cust_nature AS H1 ON A.buss_no = H1.buss_no 
    	AND H1.buss_type = 'HG' 
    	AND H1.STATUS = '1'
    	LEFT JOIN cust_legal AS H2 ON A.buss_no = H2.buss_no 
    	AND H2.buss_type = 'HG' 
    	AND H2.STATUS = '1'
    	LEFT JOIN wf_task AS W ON A.buss_no = W.buss_no 
    	AND W.task_type = '30' 
    	AND W.is_current = "Y" 
    WHERE
    	B.is_current = "Y" 
    	AND A.STATUS = "1" 
    	AND A.task_type = B.task_type 
    	AND A.task_status = B.STATUS 
    	 
    	AND A.sponsor_id = 'x'
    
    GROUP BY
    	A.buss_no 
    
    UNION ALL 
    
    SELECT
    	B.id,
    	A.buss_no,
    	A.task_type,
    	A.task_status,
    	A.start_date,
    	B.create_time,
    	A.depart_id,
    	A.sponsor_id,
    	A.jointly_id,
    	A.sponsor_name,
    	A.jointly_name,
    	A.raise_funds,
    	A.lease_amount,
    	B.owner_id,
    	B.owner_name,
    	B.cust_name,
    	C.start_date c_start_date,
    CASE
    	
    	WHEN (
    	GROUP_CONCAT( DISTINCT H1.cust_name ) IS NOT NULL 
    	AND LENGTH( trim( GROUP_CONCAT( DISTINCT H1.cust_name ) ) ) > 0 
    	AND GROUP_CONCAT( DISTINCT H2.com_fullname ) IS NOT NULL 
    	AND LENGTH( trim( GROUP_CONCAT( DISTINCT H2.com_fullname ) ) ) > 0 
    	) THEN
    		CONCAT( GROUP_CONCAT( DISTINCT H1.cust_name ), ',', GROUP_CONCAT( DISTINCT H2.com_fullname ) ) ELSE ifnull( GROUP_CONCAT( DISTINCT H1.cust_name ), GROUP_CONCAT( DISTINCT H2.com_fullname ) ) 
    	END 'hgMan',
    	DATE_FORMAT( W.create_time, '%Y-%m-%d %H:%i:%s' ) AS auditTime 
    FROM
    	orders A
    	INNER JOIN wf_task B ON A.buss_no = B.buss_no
    	LEFT JOIN contract_clause AS C ON A.buss_no = C.buss_no
    	LEFT JOIN cust_nature AS H1 ON A.buss_no = H1.buss_no 
    	AND H1.buss_type = 'HG' 
    	AND H1.STATUS = '1'
    	LEFT JOIN cust_legal AS H2 ON A.buss_no = H2.buss_no 
    	AND H2.buss_type = 'HG' 
    	AND H2.STATUS = '1'
    	LEFT JOIN wf_task AS W ON A.buss_no = W.buss_no 
    	AND W.task_type = '30' 
    	AND W.is_current = "Y" 
    WHERE
    	B.is_current = "Y" 
    	AND A.STATUS = "1" 
    	AND A.task_type = B.task_type 
    	AND A.task_status = B.STATUS 
    	 
    	AND A.jointly_id = 'xx' 
    GROUP BY
    	A.buss_no 
    	
    	
    	UNION ALL 
    	
    	SELECT
    	B.id,
    	A.buss_no,
    	A.task_type,
    	A.task_status,
    	A.start_date,
    	B.create_time,
    	A.depart_id,
    	A.sponsor_id,
    	A.jointly_id,
    	A.sponsor_name,
    	A.jointly_name,
    	A.raise_funds,
    	A.lease_amount,
    	B.owner_id,
    	B.owner_name,
    	B.cust_name,
    	C.start_date c_start_date,
    CASE
    	
    	WHEN (
    	GROUP_CONCAT( DISTINCT H1.cust_name ) IS NOT NULL 
    	AND LENGTH( trim( GROUP_CONCAT( DISTINCT H1.cust_name ) ) ) > 0 
    	AND GROUP_CONCAT( DISTINCT H2.com_fullname ) IS NOT NULL 
    	AND LENGTH( trim( GROUP_CONCAT( DISTINCT H2.com_fullname ) ) ) > 0 
    	) THEN
    		CONCAT( GROUP_CONCAT( DISTINCT H1.cust_name ), ',', GROUP_CONCAT( DISTINCT H2.com_fullname ) ) ELSE ifnull( GROUP_CONCAT( DISTINCT H1.cust_name ), GROUP_CONCAT( DISTINCT H2.com_fullname ) ) 
    	END 'hgMan',
    	DATE_FORMAT( W.create_time, '%Y-%m-%d %H:%i:%s' ) AS auditTime 
    FROM
    	orders A
    	INNER JOIN wf_task B ON A.buss_no = B.buss_no
    	LEFT JOIN contract_clause AS C ON A.buss_no = C.buss_no
    	LEFT JOIN cust_nature AS H1 ON A.buss_no = H1.buss_no 
    	AND H1.buss_type = 'HG' 
    	AND H1.STATUS = '1'
    	LEFT JOIN cust_legal AS H2 ON A.buss_no = H2.buss_no 
    	AND H2.buss_type = 'HG' 
    	AND H2.STATUS = '1'
    	LEFT JOIN wf_task AS W ON A.buss_no = W.buss_no 
    	AND W.task_type = '30' 
    	AND W.is_current = "Y" 
    WHERE
    	B.is_current = "Y" 
    	AND A.STATUS = "1" 
    	AND A.task_type = B.task_type 
    	AND A.task_status = B.STATUS 
    	 
    	AND A.depart_id = 'xxx'
    GROUP BY
    	A.buss_no 
    
    
    
    	) AS SUB
    	INNER JOIN lease_info LS ON SUB.buss_no = LS.buss_no 
    	AND LS.lease_type = '1' 
    GROUP BY
    	SUB.buss_no 
    ORDER BY
    SUB.create_time DESC
    
    limit 1,10

多次测试执行时间, 都在0.9秒左右：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 10][]

（4）、因为sql最后使用了limit分页，再给单个UNION ALL  子句再加上  limit 条数限制，单个子集合查出来的结果就少了

（虽然这样sql更快，但是最后会影响分页显示结果，看业务情况使用）：

SELECT
    	SUB.*,
    	GROUP_CONCAT( LS.delivery_place ) AS deliverPlaces 
    FROM
    	(
    SELECT
    	B.id,
    	A.buss_no,
    	A.task_type,
    	A.task_status,
    	A.start_date,
    	B.create_time,
    	A.depart_id,
    	A.sponsor_id,
    	A.jointly_id,
    	A.sponsor_name,
    	A.jointly_name,
    	A.raise_funds,
    	A.lease_amount,
    	B.owner_id,
    	B.owner_name,
    	B.cust_name,
    	C.start_date c_start_date,
    CASE
    	
    	WHEN (
    	GROUP_CONCAT( DISTINCT H1.cust_name ) IS NOT NULL 
    	AND LENGTH( trim( GROUP_CONCAT( DISTINCT H1.cust_name ) ) ) > 0 
    	AND GROUP_CONCAT( DISTINCT H2.com_fullname ) IS NOT NULL 
    	AND LENGTH( trim( GROUP_CONCAT( DISTINCT H2.com_fullname ) ) ) > 0 
    	) THEN
    		CONCAT( GROUP_CONCAT( DISTINCT H1.cust_name ), ',', GROUP_CONCAT( DISTINCT H2.com_fullname ) ) ELSE ifnull( GROUP_CONCAT( DISTINCT H1.cust_name ), GROUP_CONCAT( DISTINCT H2.com_fullname ) ) 
    	END 'hgMan',
    	DATE_FORMAT( W.create_time, '%Y-%m-%d %H:%i:%s' ) AS auditTime 
    FROM
    	orders A
    	INNER JOIN wf_task B ON A.buss_no = B.buss_no
    	LEFT JOIN contract_clause AS C ON A.buss_no = C.buss_no
    	LEFT JOIN cust_nature AS H1 ON A.buss_no = H1.buss_no 
    	AND H1.buss_type = 'HG' 
    	AND H1.STATUS = '1'
    	LEFT JOIN cust_legal AS H2 ON A.buss_no = H2.buss_no 
    	AND H2.buss_type = 'HG' 
    	AND H2.STATUS = '1'
    	LEFT JOIN wf_task AS W ON A.buss_no = W.buss_no 
    	AND W.task_type = '30' 
    	AND W.is_current = "Y" 
    WHERE
    	B.is_current = "Y" 
    	AND A.STATUS = "1" 
    	AND A.task_type = B.task_type 
    	AND A.task_status = B.STATUS 
    	 
    	AND A.sponsor_id = 'x'
    
    GROUP BY
    	A.buss_no  limit 1,10
    
    UNION ALL 
    
    SELECT
    	B.id,
    	A.buss_no,
    	A.task_type,
    	A.task_status,
    	A.start_date,
    	B.create_time,
    	A.depart_id,
    	A.sponsor_id,
    	A.jointly_id,
    	A.sponsor_name,
    	A.jointly_name,
    	A.raise_funds,
    	A.lease_amount,
    	B.owner_id,
    	B.owner_name,
    	B.cust_name,
    	C.start_date c_start_date,
    CASE
    	
    	WHEN (
    	GROUP_CONCAT( DISTINCT H1.cust_name ) IS NOT NULL 
    	AND LENGTH( trim( GROUP_CONCAT( DISTINCT H1.cust_name ) ) ) > 0 
    	AND GROUP_CONCAT( DISTINCT H2.com_fullname ) IS NOT NULL 
    	AND LENGTH( trim( GROUP_CONCAT( DISTINCT H2.com_fullname ) ) ) > 0 
    	) THEN
    		CONCAT( GROUP_CONCAT( DISTINCT H1.cust_name ), ',', GROUP_CONCAT( DISTINCT H2.com_fullname ) ) ELSE ifnull( GROUP_CONCAT( DISTINCT H1.cust_name ), GROUP_CONCAT( DISTINCT H2.com_fullname ) ) 
    	END 'hgMan',
    	DATE_FORMAT( W.create_time, '%Y-%m-%d %H:%i:%s' ) AS auditTime 
    FROM
    	orders A
    	INNER JOIN wf_task B ON A.buss_no = B.buss_no
    	LEFT JOIN contract_clause AS C ON A.buss_no = C.buss_no
    	LEFT JOIN cust_nature AS H1 ON A.buss_no = H1.buss_no 
    	AND H1.buss_type = 'HG' 
    	AND H1.STATUS = '1'
    	LEFT JOIN cust_legal AS H2 ON A.buss_no = H2.buss_no 
    	AND H2.buss_type = 'HG' 
    	AND H2.STATUS = '1'
    	LEFT JOIN wf_task AS W ON A.buss_no = W.buss_no 
    	AND W.task_type = '30' 
    	AND W.is_current = "Y" 
    WHERE
    	B.is_current = "Y" 
    	AND A.STATUS = "1" 
    	AND A.task_type = B.task_type 
    	AND A.task_status = B.STATUS 
    	 
    	AND A.jointly_id = 'xx' 
    GROUP BY
    	A.buss_no  limit 1,10
    	
    	
    	UNION ALL 
    	
    	SELECT
    	B.id,
    	A.buss_no,
    	A.task_type,
    	A.task_status,
    	A.start_date,
    	B.create_time,
    	A.depart_id,
    	A.sponsor_id,
    	A.jointly_id,
    	A.sponsor_name,
    	A.jointly_name,
    	A.raise_funds,
    	A.lease_amount,
    	B.owner_id,
    	B.owner_name,
    	B.cust_name,
    	C.start_date c_start_date,
    CASE
    	
    	WHEN (
    	GROUP_CONCAT( DISTINCT H1.cust_name ) IS NOT NULL 
    	AND LENGTH( trim( GROUP_CONCAT( DISTINCT H1.cust_name ) ) ) > 0 
    	AND GROUP_CONCAT( DISTINCT H2.com_fullname ) IS NOT NULL 
    	AND LENGTH( trim( GROUP_CONCAT( DISTINCT H2.com_fullname ) ) ) > 0 
    	) THEN
    		CONCAT( GROUP_CONCAT( DISTINCT H1.cust_name ), ',', GROUP_CONCAT( DISTINCT H2.com_fullname ) ) ELSE ifnull( GROUP_CONCAT( DISTINCT H1.cust_name ), GROUP_CONCAT( DISTINCT H2.com_fullname ) ) 
    	END 'hgMan',
    	DATE_FORMAT( W.create_time, '%Y-%m-%d %H:%i:%s' ) AS auditTime 
    FROM
    	orders A
    	INNER JOIN wf_task B ON A.buss_no = B.buss_no
    	LEFT JOIN contract_clause AS C ON A.buss_no = C.buss_no
    	LEFT JOIN cust_nature AS H1 ON A.buss_no = H1.buss_no 
    	AND H1.buss_type = 'HG' 
    	AND H1.STATUS = '1'
    	LEFT JOIN cust_legal AS H2 ON A.buss_no = H2.buss_no 
    	AND H2.buss_type = 'HG' 
    	AND H2.STATUS = '1'
    	LEFT JOIN wf_task AS W ON A.buss_no = W.buss_no 
    	AND W.task_type = '30' 
    	AND W.is_current = "Y" 
    WHERE
    	B.is_current = "Y" 
    	AND A.STATUS = "1" 
    	AND A.task_type = B.task_type 
    	AND A.task_status = B.STATUS 
    	 
    	AND A.depart_id = 'xxx'
    GROUP BY
    	A.buss_no  
    
    limit 1,10
    
    	) AS SUB
    	INNER JOIN lease_info LS ON SUB.buss_no = LS.buss_no 
    	AND LS.lease_type = '1' 
    GROUP BY
    	SUB.buss_no 
    ORDER BY
    SUB.create_time DESC
    
    limit 1,10

执行测试时间如下，多次测试都在 0.4~0.7秒之间：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 11][]

（5）上了一下测试环境，页面翻页效果还是比较理想的，响应都在几毫秒之内（跟生产环境数据量不一样，但是肯定会比之前应该要快不少）：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 12][]

这次简单优化，就完成了；可以达到预期查询的性能效果，等待上线验证效果；

to do：

但是个人感觉sql的 case when逻辑这块，还有待优化的空间，不知道大家有什么意见吗，个人目前没有思路：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 13][]

[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70]: /images/20221120/d14788fe587e4c359b85164e1f3f66d7.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 1]: /images/20221120/ddde23995e8a4c67be71e92c7c09c096.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 2]: https://img-blog.csdnimg.cn/20201203142523285.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 3]: https://img-blog.csdnimg.cn/20201203142455511.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 4]: https://img-blog.csdnimg.cn/20201203142848848.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 5]: https://img-blog.csdnimg.cn/20201203143339723.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 6]: https://img-blog.csdnimg.cn/20201203144700225.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 7]: https://img-blog.csdnimg.cn/20201203150716416.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 8]: https://img-blog.csdnimg.cn/20201223102406237.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 9]: https://img-blog.csdnimg.cn/20201223103341111.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 10]: https://img-blog.csdnimg.cn/20201203145814979.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 11]: https://img-blog.csdnimg.cn/20201203151628534.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 12]: https://img-blog.csdnimg.cn/20201203152154665.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ_size_16_color_FFFFFF_t_70 13]: https://img-blog.csdnimg.cn/20201203151841639.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xlanVzdGRvaXQ=,size_16,color_FFFFFF,t_70