是否可以在一个循环中背靠背运行两个 sql 插入？

Question

假设你有一个类似于的数据循环

for tree in trees:
    tree_type = [t_type]
    tree_owner = [t_owner]

所以循环的一个结果看起来像

magnolia
Bill

您的两个表如下所示，_id 列是自动递增的 PKs

tree_header

tree_header_id	tree_type
25	spruce

tree_detail

tree_detail_id	tree_id	tree_owner
2	25	Susan
3	25	Roger

是否可以使用循环来串联运行两个插入，以将每个结果插入到两个表中

Insert into tree_header(tree_type) values(tree_type)

last_id = last id inserted into header

Insert into tree_detail(tree_id, tree_owner) values(last_id, tree_owner)

Answer 1

你实际上可以做到这一点是一个单一的声明。该技术是创建一个 CTE 来接收参数。在这种情况下，树类型和所有者数组。然后第二个 CTE 插入 tree_header 并返回生成的 id。最后，将参数与生成的标头连接起来。见Demo。

with parms (tree, owners) as
     ( values ('<tree type>', array ['<Owner_Array>']) )
   , make_tree as
     ( insert into tree_header(tree_type)
         select tree from parms
         returning tree_id, tree_type
     )
insert into  tree_detail(owner_name, tree_id)
   select unnest (p.Owner_Array), mt.tree_id
     from parms p 
     join make_tree mt on (mt.tree_type = p.tree);

现在，由于我不编写您的晦涩语言（Python？），因此我的翻译方式并不完全正确。但我猜是这样的：

import psycopg2

class MyData():
    def __init__(self, host="10.0.80.85", db="hhh", user="postgres", password="5bebfakee11"):
        self.conn = psycopg2.connect(host=host, database=db, user=user, password=password)
        self.cur = self.conn.cursor() 
            
    def close(self):
        self.cur.close()
        self.conn.close()


 db = MyData()
 statement = "with parms (tree, owners) as                   \
                  ( values ('%',  array [%]) )               \
                , make_tree as                               \
                  ( insert into tree_header(tree_type)       \
                      select tree from parms                 \ 
                      returning tree_id, tree_type           \ 
                  )                                          \
             insert into  tree_detail(owner_name, tree_id)   \
                select unnest (p.owners), mt.tree_id         \
                  from parms p                               \ 
                  join make_tree mt on (mt.tree_type = p.tree);"

 db.cur.execute(statement, (last_header_id,list_owner_array))
 
 db.conn.commit()
 db.close()

注意：该语句没有返回任何内容，它要么是 Postgres 抛出的，要么是异常的。

Answer 2

这就是答案。这允许您创建循环、设置变量并将其传递给值。因此，如果您有一个从任何地方返回的数据集 - 只需通过它进入一个循环并将您的数据值分配给变量并将它们传递给插入。这也将值分开，并且可以在任一插入下插入多行。

import psycopg2

class MyData():
    def __init__(self, host="10.0.80.85", db="hhh", user="postgres", password="5bebfakee11"):
        self.conn = psycopg2.connect(host=host, database=db, user=user, password=password)
        self.cur = self.conn.cursor() 
            
    def close(self):
        self.cur.close()
        self.conn.close()


db = MyData()
insert = "INSERT INTO hhh.header (col1, col2) VALUES (%s, %s) RETURNING header_id;"
db.cur.execute(insert, ('variant1', 'true' ,))

last_header_id = db.cur.fetchone()[0]

insert = "INSERT INTO hhh.detail (header_id) VALUES (%s)"
db.cur.execute(insert, (last_header_id,))

db.conn.commit()
db.close()