require 'stringio'
require 'active_support/core_ext/big_decimal'
module ActiveRecord
# = Active Record Schema Dumper
#
# This class is used to dump the database schema for some connection to some
# output format (i.e., ActiveRecord::Schema).
class SchemaDumper #:nodoc:
private_class_method :new
##
# :singleton-method:
# A list of tables which should not be dumped to the schema.
# Acceptable values are strings as well as regexp.
# This setting is only used if ActiveRecord::Base.schema_format == :ruby
cattr_accessor :ignore_tables
@@ignore_tables = []
def self.dump(connection=ActiveRecord::Base.connection, stream=STDOUT)
new(connection).dump(stream)
stream
end
def dump(stream)
header(stream)
tables(stream)
trailer(stream)
stream
end
private
def initialize(connection)
@connection = connection
@types = @connection.native_database_types
@version = Migrator::current_version rescue nil
end
def header(stream)
define_params = @version ? "version: #{@version}" : ""
if stream.respond_to?(:external_encoding) && stream.external_encoding
stream.puts "# encoding: #{stream.external_encoding.name}"
end
stream.puts <<HEADER
# This file is auto-generated from the current state of the database. Instead
# of editing this file, please use the migrations feature of Active Record to
# incrementally modify your database, and then regenerate this schema definition.
#
# Note that this schema.rb definition is the authoritative source for your
# database schema. If you need to create the application database on another
# system, you should be using db:schema:load, not running all the migrations
# from scratch. The latter is a flawed and unsustainable approach (the more migrations
# you'll amass, the slower it'll run and the greater likelihood for issues).
#
# It's strongly recommended that you check this file into your version control system.
ActiveRecord::Schema.define(#{define_params}) do
HEADER
end
def trailer(stream)
stream.puts "end"
end
def tables(stream)
@connection.tables.sort.each do |tbl|
next if ['schema_migrations', ignore_tables].flatten.any? do |ignored|
case ignored
when String; remove_prefix_and_suffix(tbl) == ignored
when Regexp; remove_prefix_and_suffix(tbl) =~ ignored
else
raise StandardError, 'ActiveRecord::SchemaDumper.ignore_tables accepts an array of String and / or Regexp values.'
end
end
table(tbl, stream)
end
end
def table(table, stream)
columns = @connection.columns(table)
begin
tbl = StringIO.new
# first dump primary key column
if @connection.respond_to?(:pk_and_sequence_for)
pk, _ = @connection.pk_and_sequence_for(table)
elsif @connection.respond_to?(:primary_key)
pk = @connection.primary_key(table)
end
tbl.print " create_table #{remove_prefix_and_suffix(table).inspect}"
if columns.detect { |c| c.name == pk }
if pk != 'id'
tbl.print %Q(, primary_key: "#{pk}")
end
else
tbl.print ", id: false"
end
tbl.print ", force: true"
tbl.puts " do |t|"
# then dump all non-primary key columns
column_specs = columns.map do |column|
raise StandardError, "Unknown type '#{column.sql_type}' for column '#{column.name}'" if @types[column.type].nil?
next if column.name == pk
spec = {}
spec[:name] = column.name.inspect
# AR has an optimization which handles zero-scale decimals as integers. This
# code ensures that the dumper still dumps the column as a decimal.
spec[:type] = if column.type == :integer && /^(numeric|decimal)/ =~ column.sql_type
'decimal'
else
column.type.to_s
end
spec[:limit] = column.limit.inspect if column.limit != @types[column.type][:limit] && spec[:type] != 'decimal'
spec[:precision] = column.precision.inspect if column.precision
spec[:scale] = column.scale.inspect if column.scale
spec[:null] = 'false' unless column.null
spec[:default] = default_string(column.default) if column.has_default?
(spec.keys - [:name, :type]).each{ |k| spec[k].insert(0, "#{k.to_s}: ")}
spec
end.compact
# find all migration keys used in this table
keys = [:name, :limit, :precision, :scale, :default, :null]
# figure out the lengths for each column based on above keys
lengths = keys.map { |key|
column_specs.map { |spec|
spec[key] ? spec[key].length + 2 : 0
}.max
}
# the string we're going to sprintf our values against, with standardized column widths
format_string = lengths.map{ |len| "%-#{len}s" }
# find the max length for the 'type' column, which is special
type_length = column_specs.map{ |column| column[:type].length }.max
# add column type definition to our format string
format_string.unshift " t.%-#{type_length}s "
format_string *= ''
column_specs.each do |colspec|
values = keys.zip(lengths).map{ |key, len| colspec.key?(key) ? colspec[key] + ", " : " " * len }
values.unshift colspec[:type]
tbl.print((format_string % values).gsub(/,\s*$/, ''))
tbl.puts
end
tbl.puts " end"
tbl.puts
indexes(table, tbl)
tbl.rewind
stream.print tbl.read
rescue => e
stream.puts "# Could not dump table #{table.inspect} because of following #{e.class}"
stream.puts "# #{e.message}"
stream.puts
end
stream
end
def default_string(value)
case value
when BigDecimal
value.to_s
when Date, DateTime, Time
"'#{value.to_s(:db)}'"
else
value.inspect
end
end
def indexes(table, stream)
if (indexes = @connection.indexes(table)).any?
add_index_statements = indexes.map do |index|
statement_parts = [
('add_index ' + remove_prefix_and_suffix(index.table).inspect),
index.columns.inspect,
('name: ' + index.name.inspect),
]
statement_parts << 'unique: true' if index.unique
index_lengths = (index.lengths || []).compact
statement_parts << ('length: ' + Hash[index.columns.zip(index.lengths)].inspect) unless index_lengths.empty?
index_orders = (index.orders || {})
statement_parts << ('order: ' + index.orders.inspect) unless index_orders.empty?
statement_parts << ('where: ' + index.where.inspect) if index.where
' ' + statement_parts.join(', ')
end
stream.puts add_index_statements.sort.join("\n")
stream.puts
end
end
def remove_prefix_and_suffix(table)
table.gsub(/^(#{ActiveRecord::Base.table_name_prefix})(.+)(#{ActiveRecord::Base.table_name_suffix})$/, "\\2")
end
end
end