Category: sqlite / MySql / psql

DDL, DML, DCLって何？

DDL(Data Definition Language)
CREATEやDROP、ALTERなどデータベースオブジェクトの生成や削除変更を行うコマンド

DML(Data Manipulation Language)
SELECT/INSERT/UPDATE/DELETEなどテーブルに対するデータの取得、追加、更新、削除を行うコマンド

DCL(Data Control Language)
BEGIN, COMMIT、ROLLBACKなどトランザクションを制御するためのコマンド

実務上はあまり使用しない名称だから、こんな言い方もあるんだなー程度か。

The window function performs calculations across a set of table rows that are somehow related to the current row. This is similar to the form of calculation performed by aggregate functions. However, unlike regular aggregation functions, the use of window functions does not group rows into a single output row. Each row maintains its own identity.

Windows functions perform aggregation processing(ranking, etc) on acquired(SELECT) data. It can be said that it is an essential function in the recent trend of big data and OLAP.

SELECT
  Start_station_name,
  Duration,
  SUM(Duration) OVER (PARTITION BY Start_station_name ORDER BY Start_time) AS running_total
 FROM
  trip_histories
 WHERE
  Start_time < '2019-01-30'
 LIMIT 10;

SELECT
 Start_station_name,
 Duration,
 SUM(Duration) OVER (PARTITION BY Start_station_name) AS running_total,
 COUNT(Duration) OVER (PARTITION BY Start_station_name) AS running_count,
 AVG(Duration) OVER (PARTITION BY Start_station_name) AS running_avg
FROM
 trip_histories
WHERE
 Start_time < '2019-04-30'
LIMIT 10;

SELECT
 ROW_NUMBER() OVER (order by Start_time DESC) AS row_num,
 Start_station_name,
 Start_time,
 Duration
FROM
 trip_histories
WHERE
 Start_time < '2019-04-30'
LIMIT 10;

なるほどー、これは数こなさないとな～

What is CTE?
– CTE is a so-called “recursive query”
– The WOTH clause has become available
　→ Implemented on other RDMS(PostgreSQL, Oracle etc)
– Hierarchical queries can be written very easily
　→ In the past, it had to be nested using subueries.

WITH RECURSIVE Hatoyama AS
(
 select id, last_name, first_name from family
   where last_name= "鳩山" and first_name="和夫"
 union all
   select child.id, child.last_name, child.first_name
   from family as child, Hatoyama
    where Hatoyama.id = child.parent_id
)
select * from Hatoyama;

WITH Yoshida as
(
	select id, last_name, first_name from family
	  where last_name="吉田" and parent id is null
	union all
	  select child.id, child.last_name, child.first_name
	  from family as child, Yoshida
	  where yoshida.id= child.parent_id
)
select * from Yoshida;

select文が階層化しないようにできるわけね。