pg2

Description
===========

A PostgreSQL driver for node.js that focuses on performance.

Requirements
============

* node.js -- v6.3.0 or newer

Install
=======

npm install pg2

Examples
========

* Buffered SELECT query:

``js
var Client = require('pg2');

var c = new Client({
host: '127.0.0.1',
user: 'foo',
password: 'bar',
db: 'test'
});

c.query('SELECT i FROM generate_series(1, 10) AS i', (err, rows) => {
if (err)
throw err;
console.dir(rows);
});

c.end();
`

* Streamed SELECT query:

`js
var Client = require('pg2');

var c = new Client({
host: '127.0.0.1',
user: 'foo',
password: 'bar',
db: 'test'
});

c.query('SELECT i FROM generate_series(1, 10) AS i');
.on('data', (result) => {
result.on('data', (row) => {
console.dir(row);
});
});

c.end();
`

* Using arrays (faster) instead of objects for rows:

`javascript
var Client = require('pg2');

var c = new Client({
host: '127.0.0.1',
user: 'foo',
password: 'bar',
db: 'test'
});

c.query('SELECT i FROM generate_series(1, 10) AS i', { arrays: true }, (err, rows) => {
if (err)
throw err;
console.dir(rows);
});

c.end();
`

* Using positional parameters in a query:

`javascript
var Client = require('pg2');

var c = new Client({
host: '127.0.0.1',
user: 'foo',
password: 'bar',
db: 'test'
});

c.query('SELECT i FROM generate_series($1::numeric, $2::numeric) AS i',
[1, 10],
(err, rows) => {
if (err)
throw err;
console.dir(rows);
});

c.end();
`


API
===

require('pg2') returns a _Client_ object


Client properties
-----------------

* connected - _boolean_ - true if the Client instance is currently connected to the server.

* backendParams - _object_ - Once authenticated, this value will contain any backend status values. If no such values have been received, the value of this property will be null.

* key - _object_ - Once authenticated, this object will contain two properties (pid and key) which is used to uniquely identify this connection on the server.

* status - _integer_ - Indicates the backend transaction status. Valid values:

* 73 - Not in a transactional block
* 84 - In a transactional block
69 - In a failed* transactional block (queries will be rejected until block is ended)


Client events
-------------

* ready() - A connection to the server has been established and authentication was successful.

* error(< _Error_ >err) - An error occurred at the connection level.

* close() - The connection has been closed.


Client methods
--------------

* (constructor)([< _object_ >config]) - Creates and returns a new Client instance. Valid config options include:

* user - _string_ - Username for authentication. Default: (OS username of current logged in user)

* password - _string_ - Password for authentication. Default: ''

* host - _string_ - Hostname or IP address of the server. Default: 'localhost'

* port - _integer_ - Port number of the server. Default: 5432

* db - _string_ - A database to automatically select after authentication. Default: ''

* keepalive - _mixed_ - If true, this enables TCP keepalive probes using the OS default initial delay value. If a number, this both enables TCP keepalive probes and sets the initial delay value to the given value. Default: true

* streamType - _string_ - Set to 'normal' to use node's full-featured streams instead of simpler streams. The main difference is that the simpler streams do not implement .read() support. Default: 'simple'

* query(< _string_ >query[, < _array_ >values][, < _object_ >options][, < _function_ >callback]) - _mixed_ - Enqueues the given query. If callback is not supplied, a ResultStream instance is returned. values can be an array of values that correspond to positional placeholders inside query. Valid options are:

* arrays - _boolean_ - When true, arrays are used to store row values instead of an object keyed on column names. (Note: using arrays performs better) Default: false

* hwm - _integer_ - This is the highWaterMark to use for RowStream instances emitted by a ResultStream instance. This only applies when streaming rows. Default: (node's default) 16

* connect([< _function_ >callback]) - _(void)_ - Explicitly attempts to connect to the server. Note that calling query() will implicitly attempt a connection if one is not already made. If not connected, callback is added as a one-time 'ready' event listener.

* end() - _(void)_ - Closes the connection once all queries in the queue have been executed.

* destroy() - _(void)_ - Closes the connection immediately, even if there are other queries still in the queue. Any/all queries still in the queue are properly notified.


ResultStream is an object stream that emits RowStream instances. In the case of multiple statements/queries passed to query() (separated by ;), there will be one RowStream instance for each statement/query.

RowStream is an object stream that emits rows.


TODO (in no particular order)
====

* Allow passing back of data type OIDs for columns sent by the server

* client.abort()` that implicitly opens a new, temporary connection to kill the currently running query

* COPY data support (both in and out)