chengjingfeng / postgrex

PostgreSQL driver for Elixir

Postgrex

PostgreSQL driver for Elixir.

Documentation: http://hexdocs.pm/postgrex/

Examples

iex> {:ok, pid} = Postgrex.start_link(hostname: "localhost", username: "postgres", password: "postgres", database: "postgres")
{:ok, #PID<0.69.0>}

iex> Postgrex.query!(pid, "SELECT user_id, text FROM comments", [])
%Postgrex.Result{command: :select, empty?: false, columns: ["user_id", "text"], rows: [[3,"hey"],[4,"there"]], size: 2}}

iex> Postgrex.query!(pid, "INSERT INTO comments (user_id, text) VALUES (10, 'heya')", [])
%Postgrex.Result{command: :insert, columns: nil, rows: nil, num_rows: 1}}

Features

• Automatic decoding and encoding of Elixir values to and from PostgreSQL's binary format
• User defined extensions for encoding and decoding any PostgreSQL type
• Supports transactions, prepared queries and multiple pools via DBConnection
• Supports PostgreSQL 8.4, 9.0-9.6, and later (hstore is not supported on 8.4)

Data representation

PostgreSQL      Elixir
----------      ------
NULL            nil
bool            true | false
char            "é"
int             42
float           42.0
text            "eric"
bytea           <<42>>
numeric         #Decimal<42.0> *
date            %Date{year: 2013, month: 10, day: 12}
time(tz)        %Time{hour: 0, minute: 37, second: 14} **
timestamp       %NaiveDateTime{year: 2013, month: 10, day: 12, hour: 0, minute: 37, second: 14}
timestamptz     %DateTime{year: 2013, month: 10, day: 12, hour: 0, minute: 37, second: 14, time_zone: "Etc/UTC"} **
interval        %Postgrex.Interval{months: 14, days: 40, secs: 10920, microsecs: 315}
array           [1, 2, 3]
composite type  {42, "title", "content"}
range           %Postgrex.Range{lower: 1, upper: 5}
uuid            <<160,238,188,153,156,11,78,248,187,109,107,185,189,56,10,17>>
hstore          %{"foo" => "bar"}
oid types       42
enum            "ok" ***
bit             << 1::1, 0::1 >>
varbit          << 1::1, 0::1 >>
tsvector        [%Postgrex.Lexeme{positions: [{1, :A}], word: "a"}]

* Decimal

** Timezones will always be normalized to UTC or assumed to be UTC when no information is available, either by PostgreSQL or Postgrex

*** Enumerated types (enum) are custom named database types with strings as values.

**** Anonymous composite types are decoded (read) as tuples but they cannot be encoded (written) to the database

Postgrex does not automatically cast between types. For example, you can't pass a string where a date is expected. To add type casting, support new types, or change how any of the types above are encoded/decoded, you can use extensions.

JSON support

Postgrex comes with JSON support out of the box via the Jason library. To use it, add :jason to your dependencies:

{:jason, "~> 1.0"}

You can customize it to use another library via the :json_library configuration:

config :postgrex, :json_library, SomeOtherLib

Once you change the value, you have to recompile Postgrex, which can be done by cleaning its current build:

mix deps.clean postgrex --build

Extensions

Extensions are used to extend Postgrex' built-in type encoding/decoding.

The extensions directory in this project provides implementation for many Postgres' built-in data types. It is also a great example of how to implement your own extensions. For example, you can look at the Date extension as a starting point.

Once you defined your extensions, you should build custom type modules, passing all of your extensions as arguments:

Postgrex.Types.define(MyApp.PostgrexTypes, [MyApp.Postgis.Extensions], [])

Postgrex.Types.define/3 must be called on its own file, outside of any module and function, as it only needs to be defined once during compilation.

Once a type module is defined, you must specify it on start_link:

Postgrex.start_link(types: MyApp.PostgrexTypes)

OID type encoding

PostgreSQL's wire protocol supports encoding types either as text or as binary. Unlike most client libraries Postgrex uses the binary protocol, not the text protocol. This allows for efficient encoding of types (e.g. 4-byte integers are encoded as 4 bytes, not as a string of digits) and automatic support for arrays and composite types.

Unfortunately the PostgreSQL binary protocol transports OID types as integers while the text protocol transports them as string of their name, if one exists, and otherwise as integer.

This means you either need to supply oid types as integers or perform an explicit cast (which would be automatic when using the text protocol) in the query.

# Fails since $1 is regclass not text.
query("select nextval($1)", ["some_sequence"])

# Perform an explicit cast, this would happen automatically when using a
# client library that uses the text protocol.
query("select nextval($1::text::regclass)", ["some_sequence"])

# Determine the oid once and store it for later usage. This is the most
# efficient way, since PostgreSQL only has to perform the lookup once. Client
# libraries using the text protocol do not support this.
%{rows: [{sequence_oid}]} = query("select $1::text::regclass", ["some_sequence"])
query("select nextval($1)", [sequence_oid])

PgBouncer

When using PgBouncer with transaction or statement pooling named prepared queries can not be used because the bouncer may route requests from the same postgrex connection to different PostgreSQL backend processes and discards named queries after the transactions closes. To force unnamed prepared queries:

Postgrex.start_link(prepare: :unnamed)

Contributing

To contribute you need to compile Postgrex from source and test it:

$ git clone https://github.com/elixir-ecto/postgrex.git
$ cd postgrex
$ mix test

The tests requires some modifications to your hba file. The path to it can be found by running $ psql -U postgres -c "SHOW hba_file" in your shell. Put the following above all other configurations (so that they override):

local   all             all                     trust
host    all             postgrex_md5_pw         127.0.0.1/32    md5
host    all             postgrex_cleartext_pw   127.0.0.1/32    password
host    all             postgrex_scram_pw       127.0.0.1/32    scram-sha-256

The server needs to be restarted for the changes to take effect. Additionally you need to setup a PostgreSQL user with the same username as the local user and give it trust or ident in your hba file. Or you can export $PGUSER and $PGPASSWORD before running tests.

Testing hstore on 9.0

PostgreSQL versions 9.0 does not have the CREATE EXTENSION commands. This means we have to locate the postgres installation and run the hstore.sql in contrib to install hstore. Below is an example command to test 9.0 on OS X with homebrew installed postgres:

$ PGVERSION=9.0 PGPATH=/usr/local/share/postgresql9/ mix test

License

Copyright 2013 Eric Meadows-Jönsson

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.