pg_simple

The pg_simple module provides a simple yet efficient layer over psycopg2 providing Python API for common SQL functions, explicit and implicit transactions management and database connection pooling for single and multi-threaded applications.

pg_simple is not intended to provide ORM-like functionality, rather to make it easier to interact with the PostgreSQL database from python code for direct SQL access using convenient wrapper methods. The module wraps the excellent psycopg2 library and most of the functionality is provided by this behind the scenes.

The pg_simple module provides:

• Simplified handling of database connections/cursor
• Connection pool for single or multithreaded access
• Python API to wrap basic SQL functionality: select, update, delete, join et al
• Query results as python namedtuple and dict objects (using psycopg2.extras.NamedTupleCursor and psycopg2.extras.DictCursor respectively)
• Debug logging support

Installation

With pip or easy_install:

pip install pg_simple

or:

easy_install pg_simple

or from the source:

python setup.py install

##30 Seconds Quick-start Guide

• Step 1: Initialize a connection pool manager using pg_simple.config_pool()
• Step 2: Create a database connection and cursor by instantiating a pg_simple.PgSimple object

Here's a pseudo-example to illustrate the basic concepts:

import pg_simple

pg_simple.config_pool(dsn='dbname=my_db user=my_username ...')

with pg_simple.PgSimple() as db:
    db.insert('table_name',
              data={'column': 123,
                    'another_column': 'blah blah'})
    db.commit()

with pg_simple.PgSimple() as db1:
    rows = db1.fetchall('table_name')

##Connection pool management

###Initialize the connection pool

import pg_simple

pg_simple.config_pool(max_conn=250,
                      expiration=60, # idle timeout = 60 seconds
                      host='localhost',
                      port=5432,
                      database='pg_simple',
                      user='postgres',
                      password='secret')

or, using dsn:

pg_simple.config_pool(max_conn=250,
                      expiration=60,
                      dsn='dbname=database_name user=postgres password=secret')

or, using db_url:

pg_simple.config_pool(max_conn=250,
                      expiration=60,
                      db_url= 'postgres://username:password@hostname:numeric_port/database')

The above snippets will create a connection pool capable of accommodating a maximum of 250 concurrent database connections. Once that limit is reached and the pool does not contain any idle connections, all subsequent new connection request will result in a PoolError exception (until the pool gets refilled with idle connections).

Take caution to properly clean up all pg_simple.PgSimple objects after use (wrap the object inside python try-finally block or with statement). Once the object is released, it will quietly return the internal database connction to the idle pool. Failure to dispose PgSimple properly may result in pool exhaustion error.

###Configure connection pool for thread-safe access

The default SimpleConnectionPool pool manager is not thread-safe. To utilize the connection pool in multi-threaded apps, use the ThreadedConnectionPool:

pg_simple.config_pool(max_conn=250,
                      expiration=60,
                      pool_manager=ThreadedConnectionPool,
                      dsn='...')

###Disable connection pooling

To disable connection pooling completely, set the disable_pooling parameter to True:

pg_simple.config_pool(disable_pooling=True, dsn='...')

All database requests on this pool will create new connections on the fly, and all connections returned to the pool (upon disposal of PgSimple object or by explicitly invoking pool.put_conn()) will be discarded immediately.

###Obtaining the current connection pool manager

Call the pg_simple.get_pool() method to get the current pool:

pool = pg_simple.get_pool()

###Garbage collect stale connections

To explicitly purge the pool of stale database connections (whose duration of stay in the pool exceeds the expiration timeout), invoke the pool.purge_expired_connections() method:

pool = pg_simple.get_pool()
pool.purge_expired_connections()

Note that the pool is automatically scavenged for stale connections when an idle connection is returned to the pool (using the pool.put_conn() method).

Basic Usage

Connecting to the posgtresql server

The following snippet will connect to the posgtresql server and allocate a cursor:

import sys
import pg_simple

db = pg_simple.PgSimple(log=sys.stdout,
                        log_fmt=lambda x: '>> %s' % (x if isinstance(x, str) else x.query),
                        nt_cursor=True)

By default PgSimple generates result sets as collections.namedtuple objects (using psycopg2.extras.NamedTupleCursor). If you want to access the retrieved records using an interface similar to the Python dictionaries (using psycopg2.extras.DictCursor), set the nt_cursor parameter to False:

db = pg_simple.PgSimple(nt_cursor=False)

Raw SQL execution

>>> db.execute('SELECT tablename FROM pg_tables WHERE schemaname=%s and tablename=%s', ['public', 'books'])
<cursor object at 0x102352a50; closed: 0>

Dropping and creating tables

db.drop('books')

db.create('books',
          '''
"id" SERIAL NOT NULL,
"type" VARCHAR(20) NOT NULL,
"name" VARCHAR(40) NOT NULL,
"price" MONEY NOT NULL,
"published" DATE NOT NULL,
"modified" TIMESTAMP(6) NOT NULL DEFAULT now()
'''
)

db.execute('''ALTER TABLE "books" ADD CONSTRAINT "books_pkey" PRIMARY KEY ("id")''')
db.commit()

Emptying a table or set of tables

db.truncate('tbl1')
db.truncate('tbl2, tbl3', restart_identity=True, cascade=True)
db.commit()

Inserting rows

for i in range(1, 10):
    db.insert("books",
              {"genre": "fiction",
               "name": "Book Name vol. %d" % i,
               "price": 1.23 * i,
               "published": "%d-%d-1" % (2000 + i, i)})

db.commit()

Updating rows

with pg_simple.PgSimple() as db1:
    db1.update('books',
               data={'name': 'An expensive book',
                     'price': 998.997,
                     'genre': 'non-fiction',
                     'modified': 'NOW()'},
               where=('published = %s', [datetime.date(2001, 1, 1)]))
               
    db1.commit()

Deleting rows

db.delete('books', where=('published >= %s', [datetime.date(2005, 1, 31)]))
db.commit()

Inserting/updating/deleting rows with return value

row = db.insert("books",
                {"type": "fiction",
                 "name": "Book with ID",
                 "price": 123.45,
                 "published": "1997-01-31"},
                returning='id')
print(row.id)

rows = db.update('books',
                 data={'name': 'Another expensive book',
                       'price': 500.50,
                       'modified': 'NOW()'},
                 where=('published = %s', [datetime.date(2006, 6, 1)]),
                 returning='modified')
print(rows[0].modified)

rows = db.delete('books', 
                 where=('published >= %s', [datetime.date(2005, 1, 31)]), 
                 returning='name')
for r in rows:
    print(r.name)

Fetching a single record

book = db.fetchone('books', 
                   fields=['name', 'published'], 
                   where=('published = %s', [datetime.date(2002, 2, 1)]))
                   
print(book.name + 'was published on ' + book[1])

Fetching multiple records

books = db.fetchall('books',
                    fields=['name AS n', 'genre AS g'],
                    where=('published BETWEEN %s AND %s', [datetime.date(2005, 2, 1), datetime.date(2009, 2, 1)]),
                    order=['published', 'DESC'], 
                    limit=5, 
                    offset=2)

for book in books:
    print(book.n + 'belongs to ' + book[1])

Explicit database transaction management

with pg_simple.PgSimple() as _db:
    try:
        _db.execute('Some SQL statement')
        _db.commit()
    except:
        _db.rollback()

Implicit database transaction management

with pg_simple.PgSimple() as _db:
    _db.execute('Some SQL statement')
    _db.commit()

The above transaction will be rolled back automatically should something goes awry.