lead() 和 lag() 函數

Oracle的兩個與偏移量相關的分析函數。

可以獲取到向前或向後偏移指定行的某一列元素。

效果類似於自連接，但使用更加方便，效率更高。

lead(field, number, default) over(partition by id order by name)
當前記錄後 number 條記錄的 field 字段的值，如果沒有默認爲 default
lag(field, number, default) over(partition by id order by name)
當前記錄前 number 條記錄的 field 字段的值，如果沒有默認爲 default

示例一

有 n 條生產線生產 2 中產品（檯燈、手機），每條生產線上有 1 個計數器，每生產一個產品，計數器就會 + 1（計數器最大值爲 999，即每到 1000 就會清零重新開始）。☆ 假設獲取計數器值的頻率很高，排除瞬間增長 >1000個產品的情況。

構造原始數據

統計每次獲取計數器時數字的增量。
使用當前記錄的 COUNT 值減去前一條記錄的 COUNT 值，如果沒有，默認等於當前記錄，如當前記錄的 COUNT 大於前一條記錄的 COUNT，則意味着被置零，需要加上 1000

	/*此段代碼爲測試數據構造 start*/
	WITH temp AS (
	SELECT '計數器-01' CNAME, '檯燈' ENAME, TO_DATE('2019/09/18 21:00:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 500 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-01' CNAME, '檯燈' ENAME, TO_DATE('2019/09/18 21:10:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 750 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-01' CNAME, '檯燈' ENAME, TO_DATE('2019/09/18 21:20:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 945 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-01' CNAME, '檯燈' ENAME, TO_DATE('2019/09/18 21:30:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 130 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-01' CNAME, '檯燈' ENAME, TO_DATE('2019/09/18 21:40:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 425 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-01' CNAME, '檯燈' ENAME, TO_DATE('2019/09/18 21:50:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 654 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-01' CNAME, '檯燈' ENAME, TO_DATE('2019/09/18 22:10:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 32 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-02' CNAME, '檯燈' ENAME, TO_DATE('2019/09/18 21:30:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 412 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-02' CNAME, '檯燈' ENAME, TO_DATE('2019/09/18 21:40:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 521 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-02' CNAME, '檯燈' ENAME, TO_DATE('2019/09/18 21:50:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 687 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-02' CNAME, '檯燈' ENAME, TO_DATE('2019/09/18 22:00:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 873 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-02' CNAME, '檯燈' ENAME, TO_DATE('2019/09/18 22:10:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 952 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-03' CNAME, '手機' ENAME, TO_DATE('2019/09/18 22:00:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 43 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-03' CNAME, '手機' ENAME, TO_DATE('2019/09/18 22:10:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 241 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-03' CNAME, '手機' ENAME, TO_DATE('2019/09/18 22:20:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 434 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-03' CNAME, '手機' ENAME, TO_DATE('2019/09/18 22:30:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 644 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-03' CNAME, '手機' ENAME, TO_DATE('2019/09/18 22:40:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 832 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-03' CNAME, '手機' ENAME, TO_DATE('2019/09/18 22:50:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 5 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-03' CNAME, '手機' ENAME, TO_DATE('2019/09/18 23:00:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 231 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-04' CNAME, '手機' ENAME, TO_DATE('2019/09/18 21:30:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 496 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-04' CNAME, '手機' ENAME, TO_DATE('2019/09/18 21:40:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 634 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-04' CNAME, '手機' ENAME, TO_DATE('2019/09/18 21:50:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 874 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-04' CNAME, '手機' ENAME, TO_DATE('2019/09/18 22:00:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 969 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-04' CNAME, '手機' ENAME, TO_DATE('2019/09/18 22:10:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 201 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-04' CNAME, '手機' ENAME, TO_DATE('2019/09/18 22:20:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 423 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-04' CNAME, '手機' ENAME, TO_DATE('2019/09/18 22:30:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 654 COUNT FROM dual
	UNION ALL 
	SELECT '計數器-04' CNAME, '手機' ENAME, TO_DATE('2019/09/18 22:40:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 888 COUNT FROM dual
	)
	/*此段代碼爲數據構造 end*/

	SELECT R.CNAME,
	       R.ENAME,
	       R.CDATE,
	       R.CURRENT_COUNT,
	       R.LAST_COUNT,
	       DECODE(SIGN(R.CURRENT_COUNT - R.LAST_COUNT),
	              -1,
	              R.CURRENT_COUNT - R.LAST_COUNT + 1000，
	              R.CURRENT_COUNT - R.LAST_COUNT) 增量
	  FROM (SELECT T.CNAME,
	               T.ENAME,
	               T.CDATE,
	               T.COUNT CURRENT_COUNT,
	               LAG(T.COUNT, 1, T.COUNT) OVER(PARTITION BY T.CNAME, T.ENAME ORDER BY T.CDATE) LAST_COUNT
	          FROM TEMP T) R;

示例二

有 n 臺檢測器，隔一段時間抓取一次狀態信息，狀態有兩種，正常(normal)和異常(error)。獲取異常時間段。

構造原始數據

統計當前記錄的上一個狀態和下一個狀態。
假設統計頻率足夠高，忽略狀態轉換之間的之間。
那麼，要獲取異常區間。
自身狀態爲 error 並且上一個狀態爲 normal 的時間，爲異常開始時間
自身狀態爲 error 並且下一個狀態爲 normal 的時間，爲異常結束時間

/*此段代碼爲測試數據構造 start*/
WITH temp AS(
SELECT '檢測器-01' CNAME, TO_DATE('2019/09/18 21:30:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 'normal' STATUS FROM dual
UNION ALL 
SELECT '檢測器-01' CNAME, TO_DATE('2019/09/18 21:40:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 'normal' STATUS FROM dual
UNION ALL 
SELECT '檢測器-01' CNAME, TO_DATE('2019/09/18 21:50:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 'error' STATUS FROM dual
UNION ALL 
SELECT '檢測器-01' CNAME, TO_DATE('2019/09/18 22:00:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 'error' STATUS FROM dual
UNION ALL 
SELECT '檢測器-01' CNAME, TO_DATE('2019/09/18 22:10:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 'error' STATUS FROM dual
UNION ALL 
SELECT '檢測器-01' CNAME, TO_DATE('2019/09/18 22:20:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 'normal' STATUS FROM dual
UNION ALL 
SELECT '檢測器-01' CNAME, TO_DATE('2019/09/18 22:30:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 'normal' STATUS FROM dual
UNION ALL 
SELECT '檢測器-01' CNAME, TO_DATE('2019/09/18 22:40:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 'error' STATUS FROM dual
UNION ALL 
SELECT '檢測器-01' CNAME, TO_DATE('2019/09/18 22:50:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 'error' STATUS FROM dual
UNION ALL 
SELECT '檢測器-01' CNAME, TO_DATE('2019/09/18 23:00:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 'normal' STATUS FROM dual
UNION ALL 
SELECT '檢測器-01' CNAME, TO_DATE('2019/09/18 23:10:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 'error' STATUS FROM dual
UNION ALL 
SELECT '檢測器-01' CNAME, TO_DATE('2019/09/18 23:20:00', 'yyyy/mm/dd hh24:mi:ss') CDATE, 'normal' STATUS FROM dual
)
/*此段代碼爲測試數據構造 end*/
SELECT R.CNAME,
       R.CDATE,
       R.STATUS,
       /* 要獲取異常區間 */
       /*自身狀態爲 error 並且上一個狀態爲 normal 的時間 爲異常開始時間 */
       /*自身狀態爲 error 並且下一個狀態爲 normal 的時間 爲異常結束時間 */
       DECODE(R.STATUS, 'error',DECODE(R.LAST_STATUS, 'normal', R.CDATE, NULL) , NULL ) ERR_START_DATE,
       DECODE(R.STATUS, 'error', DECODE(R.NEXT_STATUS, 'normal', R.CDATE, NULL ), NULL) ERR_END_DATE
        FROM (
SELECT T.CNAME,
       T.CDATE,
       LAG(T.STATUS, 1, T.STATUS) OVER(PARTITION BY T.CNAME ORDER BY T.CDATE) LAST_STATUS,
       T.STATUS,
       LEAD(T.STATUS, 1, T.STATUS) OVER(PARTITION BY T.CNAME ORDER BY T.CDATE) NEXT_STATUS
  FROM TEMP T ) R;

【Oracle】lead 和 lag 函數

lead() 和 lag() 函數

示例一

示例二

ziw2pdf

apisix~helm方式的部署到k8s

firmeye - IoT固件漏洞挖掘工具

SpringBoot 學習筆記_整合NoSQL —— MongoDB

SpringBoot 學習筆記_整合NoSQL —— Redis

SpringBoot 學習筆記_整合持久層（一）——JdbcTemplate

SpringBoot 學習筆記_整合持久層——Spring Boot JPA

SpringBoot 學習筆記_整合持久層——MyBatis

https://yachay.unat.edu.pe/blog/index.php?comment_area=format_blog&comment_component=blog&comment_co

linux以太網驅動總結