SchemaPlus::Core creates an internal extension API to ActiveRecord. The idea is that:
-
SchemaPlus::Core does the monkey-patching so clients don't have to know too much about the internal of ActiveRecord.
-
SchemaPlus::Core's extension API is consistent across the various connection adapters, so clients don't have to figure out how to extend each connection adapter independently.
-
SchemaPlus::Core's extension API intends to remain reasonably stable even as ActiveRecord changes.
By itself, SchemaPlus::Core does not change any behavior or add any external features to ActiveRecord. It just makes the API available to clients.
SchemaPlus::Core is a client of schema_monkey, using modware to define middleware callback stacks.
SchemaPlus::Core is tested on:
- ruby 2.3.1 with activerecord 5.0, using mysql2, sqlite3 or postgresql
- ruby 2.3.1 with activerecord 5.1, using mysql2, sqlite3 or postgresql
SchemaPlus::Core::Schema::Tables
middleware was replaced bySchemaPlus::Core::DataSources
. Parameters have also changed.SchemaPlus::Core::Dumper::Indexes
middleware was removed; insteadSchemaPlus::Core::Dumper::Table
sets both columns and indexes on env.table.
As usual:
gem "schema_plus_core" # in a Gemfile
gem.add_dependency "schema_plus_core" # in a .gemspec
The API is in the form of a collection of modware middleware callback stacks. A client of the API uses schema_monkey to insert middleware modules into the stacks. As per schema_monkey, the typical module structure looks like:
require "schema_plus/core"
module MyClient
module Middleware
#
# Middleware modules to insert in SchemaPlus::Core API stacks
#
end
module ActiveRecord
#
# direct ActiveRecord enhancements, should your client need any.
#
end
end
SchemaMonkey.register MyClient
For example, a client could use the Migration::Index
stack to automatically make an index unique if any column starts with 'u':
require "schema_plus/core"
module AutoUniquify
module Middleware
module Migration
module Index
def before(env)
env.options[:unique] = true if env.column_names.grep(/^u/).any?
end
end
end
end
end
SchemaMonkey.register AutoUniquify
Ideally most clients will not need to define direct ActiveRecord enhancements, other than perhaps to create new methods on public classes. If you have a client that needs more complex monkey-patching, that could be a sign that SchemaPlus::Core's API is missing some useful functionality -- consider submitting a PR to SchemaPlus::Core add it!
For organizational clarity, the SchemaPlus::Core stacks are grouped into modules based on broad categories. In the Env field tables below, Initialized
Stacks for general operations queries pertaining to the entire database schema:
-
Schema::Define
Wrapper around the
ActiveRecord::Schema.define
method loads a dumped schema file (schema.rb
).Env Field Description Initial value :info
Schema information hash args :block
The proc containing the schema definition statements args The base implementation calls the block to define the schema.
-
Schema::Indexes
Wrapper around the
connection.indexes(table_name)
method. Env contains:Env Field Description Initial value :index_definitions
The result of the lookup []
:connection
The current ActiveRecord connection context :table_name
The name of the table to query arg :query_name
Label sometimes used by ActiveRecord logging arg The base implementation appends its results to
env.index_definitions
-
Schema::DataSources
Wrapper around the
connection.data_sources()
method. Env contains:Env Field Description Initialized :data_sources
The result of the lookup []
:connection
The current ActiveRecord connection context The base implementation appends its results to
env.tables
Stacks for class methods on ActiveRecord models.
-
Model::Columns
Wrapper around the
Model.columns
queryEnv Field | Description | Initialized
--- | --- | ---
:columns
| The resulting Column objects |[]
:model
| The model Class being queried | contextThe base implementation appends its results to
env.columns
-
Model::ResetColumnInformation
Wrapper around the
Model.reset_column_information
methodEnv Field | Description | Initialized --- | --- | --- `:model` | The model Class being reset | *context*
The base implementation performs the reset.
-
Model::Association::Declaration
Wrapper around the
Model.has_many
,Model.has_and_belongs_to_many
,Model.has_one
, andModel.belongs_to
methodsEnv Field | Description | Initialized --- | --- | --- `:model` | The model Class being defined | *context* `:name` | The name of the association being defined. | *arg* `:scope` | The scope lambda associated with the association | *arg* `:options` | Options associated with the association. | *arg* `:extension` | Extensions to the association to be made. | *arg*
The base implementation creates the association.
Stacks for operations that change the schema. In some cases the operation immediately modifies the database schema, in others the operation defines ActiveRecord objects (e.g., column definitions in a create_table definition) and the actual modification of the database schema will happen some time later.
-
Migration::Column
Callback stack for various ways to define or modify a column.
Env Field Description Initialized :caller
The ActiveRecord instance responsible for performing the action context :operation
One of :add
,:change
,:define
,:record
context :table_name
The name of the table arg :column_name
The name of the column arg :type
The ActiveRecord column type ( :integer
,:datetime
, etc.)arg :implements_reference
This implements a migration.add_reference
,t.references
ort.belongs_to
[See below]context :options
The column options arg, default {}
The base implementation performs the column operation. No value is returned.
Notes:
-
The
:operation
field has the following meanings::add
- The column will be added immediately (Migration#add_column
):change
- The column will be changed immediately (Migration#change_column
):define
- The column will be added to table definition, which will be emitted later:record
- The column info will be added to a migration command recorder, for later playback in reverse byMigration#down
-
In the case of a table definition using
t.references
ort.belongs_to
, the:type
field will be set to:reference
and the:column_name
will include the"_id"
suffix -
ActiveRecord's base implementation handles
migration.add_reference
,t.references
andt.belongs_to
by making nested calls tomigration.add_column
ort.column
to create the resulting column (or two columns, for polymorphic references). SchemaPlus::Core invokes theMigration::Column
stack for both the outermigration.add_reference
,t.references
ort.belongs_to
call, as well as for the nestedmigration.add_column
ort.column
call; in the nested call,env.implements_reference
will be truthy. -
Sqlite3 implements
change_column
by a creating a new table. This will result in nested calls toadd_column
, invoking theMigration::Column
stack for each; SchemaPlus::Core does not currently provide a way to distinguish those calls from explicit top-level calls.
-
-
Migration::CreateTable
Creates a new table
Env Field Description Initialized :caller
The ActiveRecord instance responsible for creating the table context :table_name
The name of the table arg :options
Create table options arg :block
Proc containing table definition statements arg The base implementation creates the table, yielding a
table_definition
instance to the block (if a block is given). -
Migration::DropTable
Drops a table from the database
Env Field Description Initialized :connection
The current ActiveRecord connection context :table_name
The name of the table arg :options
Drop table options arg The base implementation drops the table. No value is returned.
-
Migration::RenameTable
Renames a table
Env Field Description Initialized :connection
The current ActiveRecord connection context :table_name
The existing name of the table arg :new_name
The target name of the table arg The base implementation renames the table. No value is returned.
-
Migration::Index
Callback stack for various ways to define an index.
Env Field Description Initialized :caller
The ActiveRecord instance responsible for performing the action context :operation
:add
or:define
context :table_name
The name of the table arg :column_names
The names of the columns arg :options
The index options arg, default {}
The base implementation performs the index creation operation. No value is returned.
Notes:
-
The
:operation
field has the following meanings::add
- The index will be added immediately (Migration#add_index
):define
- The index will be added to a table definition, which will be emitted later.
-
Stacks for internal operations that generate SQL.
-
Sql::ColumnOptions
Callback stack around generation of the SQL options for a column definition.
Env Field Description Initialized :sql
The resulting SQL ""
:caller
The ActiveRecord::SchemaCreation instance context :connection
The current ActiveRecord connection context :column
The column definition object context :options
The column definition options context The base implementation appends the options SQL to
env.sql
-
Sql::IndexComponents
Callback stack around generation of the SQL for an index definition.
Env Field Description Initialized :sql
The resulting SQL components, in a struct with fields :name
,:type
,:columns
,:options
,:algorithm
,:using
empty struct :connection
The current ActiveRecord connection context :table_name
The name of the table context :column_names
The names of the columns context :options
The index options context The base implementation overwrites the contents of
env.sql
Notes:
- SQLite3 ignores the
:type
,:algoritm
, and:using
fields ofenv.sql
- SQLite3 ignores the
-
Sql::Table
Callback stack around generation of the SQL for a table
Env Field Description Initialized :sql
The resulting SQL components in a struct with fields :command
,:name
,:body
,:options
,:quotechar
empty struct :caller
The ActiveRecord::SchemaCreation instance context :connection
The current ActiveRecord connection context :table_definition
The TableDefinition object for the table context The base implementation overwrites the contents of
env.sql
Notes:
env.sql.command
contains the index creation command such asCREATE TABLE
orCREATE TEMPORARY TABLE
env.sql.quotechar
contains the quote character ', ", or ` to wrapenv.sql.name
in.
Stacks around low-level query execution
-
Query::Exec
Callback stack wraps the emission of sql to the underlying dbms gem.
Env Field | Description | Initialized
--- | --- | ---
:result
| The result of the database query | unset:caller
| The ActiveRecord::SchemaCreation instance | context:sql
| The SQL string | context:binds
| Values to substitute into the SQL string:query_name
| Label sometimes used by ActiveRecord logging | arg
SchemaPlus::Core provides a state object and of callbacks to various phases of the schema dumping process. The dumping process fleshes out the state object-- nothing is actually written to the dump file until after the state is fleshed out.
-
Class SchemaPlus::Core::SchemaDump
An instance of
SchemaPlus::Core::SchemaDump
gets passed to each of the callback stacks; the dump gets built up by fleshing out its contents.SchemaDump
has the following fields and methods:dump.initial = []
- an array of strings containing statements to start the schema with, such asenable_extension 'hstore'
dump.data = OpenStruct.new
- a place for clients to store arbitrary data between phasesdump.tables = {}
- a hash mapping table names to SchemaDump::Table objectsdump.final = []
- an array of strings containing statements to end the schema with.dump.depends(table_name, [prerequisite_table_names])
- call this method to ensure that the definition oftable_name
won't be output before its prerequisites.
-
Class SchemaPlus::Core::SchemaDump::Table
Each table in the dump has its contents in a SchemaDump::Table object, with these fields:
table.name
- the table nametable.pname
- table as actually used in SQL (without any prefixes or suffixes)table.options
- a string containing the options toMigration.create_table
table.columns = []
- an array of SchemaDump::Table::Column objectstable.indexes = []
- an array of SchemaDump::Table::Index objectstable.statements
- a collection of statements to include in the table definition; each is a string that should start with"t."
table.trailer
- a collection of migration statements to include immediately outside the table definition. Each is a stringtable.alt
- In some cases, ActiveRecord is unable to dump a table in the form of a migrationcreate_table
statement; in this casetable.pname
will be nil, andtable.alt
will contain the alternate string to dump instead. (E.g. if the table contains custom types, ActiveRecord will be unable to handle it and will just dump an error message as a comment.)
-
Class SchemaPlus::Core::SchemaDump::Table::Column
Each column in a table has its contents in a SchemaDump::Table::Column object, with these fields and methods:
column.name
- the column namecolumn.type
- the column type (i.e., what comes after"t."
)column.options
- a hash containing the options for the columncolumn.comments
- an array of comment strings for the column
-
Class SchemaPlus::Core::SchemaDump::Table::Index
Each index in a table has its contents in a SchemaDump::Table::Index object, with these fields and methods:
index.name
- the index nameindex.columns
- the columns that are in the indexindex.options
- a hash containing the options for the column
-
Dumper::Initial
Callback stack wraps the creation of initial statements for the dump.
Env Field Description Initialized :initial
The initial statements [] :dump
The SchemaDump object context :dumper
The current ActiveRecord::SchemaDumper instance context :connection
The current ActiveRecord connection context The base method appends initial statements to
env.initial
. -
Dumper::Tables
Callback stack wraps the dumping of all tables.
Env Field Description Initialized :dump
The SchemaDump object context :dumper
The current ActiveRecord::SchemaDumper instance context :connection
The current ActiveRecord connection context The base method iterates through all tables, dumping each.
-
Dumper::Table
Callback stack wraps the dumping of each table
Env Field Description Initialized :table
A SchemaDump::Table object table.name
only:dump
The SchemaDump object context :dumper
The current ActiveRecord::SchemaDumper instance context :connection
The current ActiveRecord connection context The base method iterates through all columns and indexes of the table, and overwrites the contents of
table
,Notes:
- When the stack is called,
env.dump.tables[env.table.name]
contains theenv.table
object. - The base method sets both
env.table.columns
andenv.tables.indexes
.
- When the stack is called,
- 2.1.1 Bug fix: Don't lose habtm options. Thanks to @iagopiimenta
- 2.1.0 Added AR5.1 support. Thanks to @iagopiimenta
- 2.0.1 Tighten up AR dependency. Thanks to @myabc.
- 2.0.0 Added AR5 support, removed AR4.2 support. Thanks to @boazy.
- 1.0.2 Missing require
- 1.0.1 Explicit gem dependencies
- 1.0.0 Clean up
SchemaDump::Table::Column
andSchemaDump::Table::Index
API:#options
is now a hash and#comments
is now an array; no longer haveadd_option
andadd_comment
methods. - 0.6.2 Bug fix: don't choke on INHERITANCE in table definition (#7). Thanks to @ADone.
- 0.6.1 Make sure to require pathname (#5)
- 0.6.0 Added
table.alt
to dumper; Bug fix: Don't crash when AR fails to dump a table. Thanks to @stenver for tracking it down - 0.5.1 Bug fix: Don't choke on a quoted newline in a
CREATE TABLE
statement (#3). Thanks to @mikeauclair - 0.5.0 Added
Migration::DropTable
- 0.4.0 Added
implements_reference
toMigration::Column
stack env - 0.3.1 Pass along (undocumented) return values from association declarations (#2). Thanks to @lowjoel
- 0.3.0 Added
Model::Association::Declaration
(#1). Thanks to @lowjoel. - 0.2.1 Added
Migration::CreateTable
andSchema::Define
; removed dependency on (defunct)schema_monkey_rails
gem. [Oops, this should have been a minor version bump] - 0.2.0 Added
Migration::DropTable
- 0.1.0 Initial release
Are you interested in contributing to SchemaPlus::Core? Thanks! Please follow the standard protocol: fork, feature branch, develop, push, and issue pull request.
Some things to know about to help you develop and test:
-
schema_dev: SchemaPlus::Core uses schema_dev to facilitate running rspec tests on the matrix of ruby, activerecord, and database versions that the gem supports, both locally and on travis-ci
To to run rspec locally on the full matrix, do:
$ schema_dev bundle install $ schema_dev rspec
You can also run on just one configuration at a time; For info, see
schema_dev --help
or the schema_dev README.The matrix of configurations is specified in
schema_dev.yml
in the project root.
- schema_monkey: SchemaPlus::Core is implemented as a schema_monkey client, using schema_monkey's convention-based protocols for extending ActiveRecord and using middleware stacks.