An ActiveRecord extension to get more from PostgreSQL:
PgSaurus is a fork of PgPower.
PgSaurus v4 was tested with Rails 5.2 and Ruby 2.4. For Rails 4.2, use PgSaurus v3. For Rails 4.1, use PgSaurus v2.5+. Older versions of Rails are not supported.
NOTE: JRuby is not supported.
In migrations you can use create_schema
and drop_schema
methods like this:
class ReplaceDemographySchemaWithPolitics < ActiveRecord::Migration
def change
drop_schema 'demography'
create_schema 'politics'
drop_schema_if_exists('demography')
create_schema_if_not_exists('politics')
end
end
Use schema :schema
option to specify schema name:
create_table "countries", schema: "demography" do |t|
# columns goes here
end
Move table countries
from demography
schema to public
:
move_table_to_schema 'demography.countries', :public
Provides the following methods to manage comments:
set_table_comment(table_name, comment)
remove_table_comment(table_name)
set_column_comment(table_name, column_name, comment)
remove_column_comment(table_name, column_name, comment)
set_column_comments(table_name, comments)
remove_column_comments(table_name, *comments)
Set a comment on the given table.
set_table_comment :phone_numbers, 'This table stores phone numbers that conform to the North American Numbering Plan.'
Sets a comment on a given column of a given table.
set_column_comment :phone_numbers, :npa, 'Numbering Plan Area Code - Allowed ranges: [2-9] for first digit, [0-9] for second and third digit.'
Removes any comment from the given table.
remove_table_comment :phone_numbers
Removes any comment from the given column of a given table.
remove_column_comment :phone_numbers, :npa
Set comments on multiple columns in the table.
set_column_comments :phone_numbers, npa: 'Numbering Plan Area Code - Allowed ranges: [2-9] for first digit, [0-9] for second and third digit.',
nxx: 'Central Office Number'
Remove comments from multiple columns in the table.
remove_column_comments :phone_numbers, :npa, :nxx
PgSaurus also adds extra table methods to the change_table
block.
Set comments:
change_table :phone_numbers do |t|
t.set_table_comment 'This table stores phone numbers that conform to the North American Numbering Plan.'
t.set_column_comment :npa, 'Numbering Plan Area Code - Allowed ranges: [2-9] for first digit, [0-9] for second and third digit.'
end
change_table :phone_numbers do |t|
t.set_column_comments npa: 'Numbering Plan Area Code - Allowed ranges: [2-9] for first digit, [0-9] for second and third digit.',
nxx: 'Central Office Number'
end
Remove comments:
change_table :phone_numbers do |t|
t.remove_table_comment
t.remove_column_comment :npa
end
change_table :phone_numbers do |t|
t.remove_column_comments :npa, :nxx
end
PgSaurus v3 augments Rails 4.2's foreign key methods with:
When you create a foreign key PgSaurus automatically creates an index.
If you do not want to generate an index, pass the exclude_index: true
option.
The syntax is compatible with Rails 4.2's foreign key handling methods.
It works with schemas as expected:
add_foreign_key('blog.comments', 'blog.posts')
Adds the index 'index_comments_on_post_id'
:
add_foreign_key(:comments, :posts)
Does not add an index:
add_foreign_key(:comments, :posts, exclude_index: true)
Note that removing a foreign key does not drop the index of the foreign key column.
If you want to remove the index, pass in the remove_index: true
option.
remove_foreign_key(:comments, column: :post_id, remove_index: true)
PgSaurus v4.X requires Rails 5. Rails 5.2 is recommended. You can use the new Rails 5 semantics to create comments and indexes inline. You also need to use the index order options using the Rails 5 semantics.
# THIS FAILS
add_index :books, ["author_id DESC NULLS FIRST", "publisher_id DESC NULLS LAST"],
name: "books_author_id_and_publisher_id"
# DO THIS INSTEAD
add_index :books, ["author_id", "publisher_id"],
name: "books_author_id_and_publisher_id",
order: { author_id: "DESC NULLS FIRST", publisher_id: "DESC NULLS LAST" }
PgSaurus v3.X now uses the Rails 4.2 semantics for add_foreign_key
and remove_foreign_key
. See http://api.rubyonrails.org/v4.2/classes/ActiveRecord/ConnectionAdapters/SchemaStatements.html
A few things have changed. The most breaking change is that the syntax remove_foreign_key :from_table, :to_table, options
no longer works.
#THIS FAILS
remove_foreign_key :comments, :posts, remove_index: true
#DO THIS INSTEAD
remove_foreign_key :comments, column: :post_id, remove_index: true
For adding foreign keys, the :dependent
option is replaced with :on_delete
and :on_update
. The :delete
value is replaced with :cascade
.
#OLD STYLE - NO LONGER WORKS
add_foreign_key :comments, :posts, dependent: :delete
#NEW STYLE - DO THIS INSTEAD
add_foreign_key :comments, :posts, on_delete: :cascade
Rails 4.x pull request was used as a starting point to patch it to be schema-aware.
Add a partial index to a table:
add_index(:comments, [:country_id, :user_id, :category], where: "category IN ('foo', 'bar')")
Add a partial index to a schema-qualified table:
add_index('blog.comments', :status, where: "status = 'active'")
PostgreSQL supports indexes on expressions. Right now, only basic functional expressions are supported.
Add an index to a column with a function:
add_index(:comments, "lower(text)")
You can also specify the index access method:
create_extension 'btree_gist'
create_extension 'fuzzystrmatch'
add_index(:comments, 'dmetaphone(author)', using: 'gist')
Specifying an operator class on each column of an index is supported.
Add an index with a custom ops class:
add_index(:books, "title varchar_pattern_ops")
PostgreSQL supports concurrent index creation. PgSaurus supports that feature by adding support to the migration DSL on index and foreign key creation.
Add an index concurrently to a table:
add_index :table, :column_id, concurrently: true
Add an index concurrently along with foreign key:
add_foreign_key :table1, :table2, column: :column_id, concurrent_index: true
PostgreSQL ships with a number of extension modules. PgSaurus provides some tools to load and unload such modules using migrations.
Please note: CREATE
/DROP EXTENSION
command was introduced in PostgreSQL 9.1.
So this functionality is not available in previous versions.
Load fuzzystrmatch extension module; and create its objects in schema public:
create_extension "fuzzystrmatch"
Load version 1.0 of the btree_gist extension module; and create its objects in schema demography.
create_extension "btree_gist", schema_name: "demography", version: "1.0"
Unload an extension module:
drop_extension "fuzzystrmatch"
PgSaurus v1.6.0 introduced experimental support for creating views. This API should only be used with the understanding that it is preliminary 'alpha' at best.
create_view "demography.citizens_view", "select * from demography.citizens"
drop_view "demography.citizens_view"
If you want to execute a migration as a specific PostgreSQL role you can use the set_role
method:
class CreateRockBands < ActiveRecord::Migration
set_role "rocker"
def change
create_table :rock_bands do |t|
# create columns
end
end
end
Technically it is equivalent to the following:
class CreateRockBands < ActiveRecord::Migration
def change
execute "SET ROLE rocker"
create_table :rock_bands do |t|
# create columns
end
ensure
execute "RESET ROLE"
end
end
You may force all migrations to have set_role
, for this, configure PgSaurus with
ensure_role_set = true
:
PgSaurus.configure do |config|
config.ensure_role_set = true
end
You can create, list, and drop functions.
Create a function:
pets_not_empty_function = <<-SQL
BEGIN
IF (SELECT COUNT(*) FROM pets) > 0
THEN
RETURN true;
ELSE
RETURN false;
END IF;
END;
SQL
# Arguments are: function_name, return_type, function_definition, options (currently, only :schema and :volatility)
create_function 'pets_not_empty()', :boolean, pets_not_empty_function, schema: 'public', volatility: 'stable'
Drop a function:
drop_function 'pets_not_empty()'
Get a list of defined functions:
ActiveRecord::Base.connection.functions
You can create and remove triggers on tables and views.
Create a trigger:
create_trigger :pets, # Table or view name
:pets_not_empty_trigger_proc, # Procedure name. Parentheses are optional if you have no arguments.
'AFTER INSERT', # Trigger event
for_each: 'ROW', # Can be a row or a statement. Default is row.
schema: 'public', # Optional schema name
constraint: true, # Sets whether the trigger is a constraint. Default is false.
deferrable: true, # Sets whether the trigger is immediate or deferrable. Default is immediate.
initially_deferred: true, # Sets whether the trigger is initially deferred. Default is immediate.
# Only relevant if the trigger is deferrable.
condition: "new.name = 'fluffy'" # Optional when condition. Default is none.
Drop a trigger:
remove_trigger :pets, :pets_not_empty_trigger_proc
Get a list of defined triggers on a table or view:
ActiveRecord::Base.connection.triggers
PgSaurus::Tools provides a number of useful methods for managing schemas, etc.:
PgSaurus::Tools.create_schema "services" # => create new PG schema "services"
PgSaurus::Tools.create_schema "nets" # => create new PG schema "nets"
PgSaurus::Tools.drop_schema "services" # => remove the PG schema "services"
PgSaurus::Tools.create_schema_if_not_exists "nets" # => Does nothing -- schema "nets" already exists
PgSaurus::Tools.drop_schema_if_exists "services" # => Does nothing -- schema "services" doesn't exist
PgSaurus::Tools.schemas # => ["public", "information_schema", "nets"]
PgSaurus::Tools.index_exists?(table, columns, options) # => returns true if an index exists for the given params
postgresql
has postgres-contrib
(if you're on Ubuntu) package installed. Tests depend on the btree_gist
and fuzzystrmatch
extensions
spec/dummy/config/database.yml
for development and test environments.rake spec
.postgresql
. If you're using Homebrew or the Postgres App, you will need to adjust the instructions accordingly (please add to this README when you do)postgresql
port and keep the build files
sudo port -k -s build postgresql94
btree_gist
and any other contrib modules
cd $(port work postgresql94)/postgresql-9.4.7/contrib/btree_gist
sudo make all
sudo make install
Support for Rails 6+
Possible support for JRuby:
create_schema(schema, user)
method - solve conflicts.Released under the MIT License. See the MIT-LICENSE file for more details.
Contributions are welcome. However, before issuing a pull request, please make sure of the following: