"PostgreSQL and Python" Lightning Talk @EuroPython2014
0 likes1,164 views
PL/Python allows users to write PostgreSQL functions and procedures using Python. It enables accessing PostgreSQL data and running Python code from within SQL queries. For example, a function could query a database table, process the results in Python by accessing modules, and return a value to the SQL query. This opens up possibilities to leverage Python's extensive libraries and expressiveness to expose data and perform complex validation from PostgreSQL.
![PostgreSQL speaks Python!
CREATE FUNCTION greet(name TEXT) RETURNS TEXT AS
$$
import random
attr = random.choice([
'great', 'fabulous', 'bright', 'brave'])
return 'Hello {} {}!'.format(attr, name)
$$
LANGUAGE plpythonu;
SELECT greet('Python fan');](https://image.slidesharecdn.com/europython2014-lightningtalk-postgresql-python-150529143502-lva1-app6892/85/PostgreSQL-and-Python-Lightning-Talk-EuroPython2014-3-320.jpg)
"PostgreSQL and Python" Lightning Talk @EuroPython2014
- 2. Well known... import psycopg2 conn = psycopg2.connect('dbname=mydb user=myusr') cur = conn.cursor() cur.execute('SELECT * FROM mytable') row = cur.fetchone()
- 3. PostgreSQL speaks Python! CREATE FUNCTION greet(name TEXT) RETURNS TEXT AS $$ import random attr = random.choice([ 'great', 'fabulous', 'bright', 'brave']) return 'Hello {} {}!'.format(attr, name) $$ LANGUAGE plpythonu; SELECT greet('Python fan');
- 4. PostgreSQL speaks Python ● PL/Python – “just” another procedural language ● SQL script that contains a string of Python ● Allows using any Python module
- 5. Real World Examples... mydb=# SELECT * FROM get_server_connection_stats() LIMIT 5; server_addr | server_port | client_addr | client_port ---------------+-------------+--------------+------------- 10.160.26.128 | 5435 | 10.111.74.16 | 40524 10.160.26.128 | 5435 | 10.160.26.1 | 49681 10.160.26.128 | 5435 | 10.160.26.1 | 15456 10.160.26.128 | 5435 | 10.160.26.2 | 64136 10.160.26.128 | 5435 | 10.160.26.1 | 1938 (5 rows)
- 6. Expose Stats via SQL with Python CREATE FUNCTION get_server_connection_stats() RETURNS SETOF server_connection AS $$ # … try: with open('/proc/net/tcp') as fp: lines = fp.read().splitlines() except: plpy.error('Unable to read /proc/net/tcp') # … return result $$ LANGUAGE plpythonu;
- 7. Custom JSON Type Validation CREATE TRIGGER validation AFTER INSERT OR UPDATE ON config_value FOR EACH … CREATE FUNCTION validate(value JSON, type_id INT) RETURNS VOID AS $$ import json # … complex type checking here # … using all of Python's expressiveness :-) raise ValueError('{} not valid'.format(value)) $$ LANGUAGE plpythonu;