运行 Python 来自 PostgreSQL 函数的脚本
Run Python script from PostgreSQL function
关于每次执行 Python 脚本的“小”问题是在 PostgreSQL table.
上执行更新或插入操作
此脚本将提取更新或插入的数据并将其写入文件。
环境数据：Ubuntu 18.04 (Bionic Beaver)、PostgreSQL 10 和 Python 3.6
SELECT * FROM pg_available_extensions
WHERE name LIKE '%python%' ORDER BY name;
并输出
姓名
default_version
installed_version
评论
hstore_plpython2u
1.0
hstore 和 plpython2u 之间的转换
hstore_plpythonu
1.0
hstore 和 plpythonu 之间的转换
ltree_plpython2u
1.0
ltree 和 plpython2u 之间的转换
ltree_plpythonu
1.0
ltree 和 plpythonu 之间的转换
plpython2u
1.0
PL/Python2U 不受信任的程序语言
plpythonu
1.0
1.0
PL/PythonU 不受信任的程序语言
我创建了一个 PostgreSQL 函数（我希望在阅读所有文档后一切正常）
CREATE FUNCTION getSomeData()
RETURNS trigger
AS $$
begin
import subprocess
subprocess.call(['/usr/bin/python3', '/some_folder/some_sub_folder/get_data.py'])
end;
$$
LANGUAGE plpythonu;
之后，创建触发器
CREATE TRIGGER executePython
AFTER INSERT ON mytable
EXECUTE PROCEDURE getSomeData();
如果我进行任何插入或更新，则不会发生任何事情。
作为额外的预防措施，我做了以下测试
sudo -u postgres python3 /some_folder/some_sub_folder/get_data.py
得到这个输出：
Traceback (most recent call last):
File "/some_folder/some_sub_folder/get_data.py", line 4, in <module>
from sqlalchemy import create_engine
ImportError: No module named sqlalchemy
我已经全局安装了 SQLAlchemy，现在我的脚本在 postgres 用户下按预期运行，但它不会触发。
apt install python3-sqlalchemy
指向bathman提供的解决方案的指针：
when you execute this as your usual user, do you execute it with the global python3, or is there a chance, you have venv or other environment loaded and your usual user python3 executable is an alias to your venv, where you locally installed SQLAlchemy? you could try to execute by /path/to/your/environment/with/sqlalchemy/installed/bin/python3 or installing SQLAlchemy globally so your /usr/bin/python3 has access to it.
解决方法：
CREATE FUNCTION getSomeData()
RETURNS trigger
AS $$
begin
import subprocess
subprocess.call(['/path/to/your/virtual/environment/bin/python3', '/some_folder/some_sub_folder/get_data.py'])
end;
$$
LANGUAGE plpythonu;
关于每次执行 Python 脚本的“小”问题是在 PostgreSQL table.
上执行更新或插入操作此脚本将提取更新或插入的数据并将其写入文件。
环境数据：Ubuntu 18.04 (Bionic Beaver)、PostgreSQL 10 和 Python 3.6
SELECT * FROM pg_available_extensions
WHERE name LIKE '%python%' ORDER BY name;
并输出
| 姓名 | default_version | installed_version | 评论 |
|---|---|---|---|
| hstore_plpython2u | 1.0 | hstore 和 plpython2u 之间的转换 | |
| hstore_plpythonu | 1.0 | hstore 和 plpythonu 之间的转换 | |
| ltree_plpython2u | 1.0 | ltree 和 plpython2u 之间的转换 | |
| ltree_plpythonu | 1.0 | ltree 和 plpythonu 之间的转换 | |
| plpython2u | 1.0 | PL/Python2U 不受信任的程序语言 | |
| plpythonu | 1.0 | 1.0 | PL/PythonU 不受信任的程序语言 |
我创建了一个 PostgreSQL 函数（我希望在阅读所有文档后一切正常）
CREATE FUNCTION getSomeData()
RETURNS trigger
AS $$
begin
import subprocess
subprocess.call(['/usr/bin/python3', '/some_folder/some_sub_folder/get_data.py'])
end;
$$
LANGUAGE plpythonu;
之后，创建触发器
CREATE TRIGGER executePython
AFTER INSERT ON mytable
EXECUTE PROCEDURE getSomeData();
如果我进行任何插入或更新，则不会发生任何事情。
作为额外的预防措施，我做了以下测试
sudo -u postgres python3 /some_folder/some_sub_folder/get_data.py
得到这个输出：
Traceback (most recent call last):
File "/some_folder/some_sub_folder/get_data.py", line 4, in <module>
from sqlalchemy import create_engine
ImportError: No module named sqlalchemy
我已经全局安装了 SQLAlchemy，现在我的脚本在 postgres 用户下按预期运行，但它不会触发。
apt install python3-sqlalchemy
指向bathman提供的解决方案的指针：
when you execute this as your usual user, do you execute it with the global python3, or is there a chance, you have venv or other environment loaded and your usual user python3 executable is an alias to your venv, where you locally installed SQLAlchemy? you could try to execute by /path/to/your/environment/with/sqlalchemy/installed/bin/python3 or installing SQLAlchemy globally so your /usr/bin/python3 has access to it.
解决方法：
CREATE FUNCTION getSomeData()
RETURNS trigger
AS $$
begin
import subprocess
subprocess.call(['/path/to/your/virtual/environment/bin/python3', '/some_folder/some_sub_folder/get_data.py'])
end;
$$
LANGUAGE plpythonu;