activerecord-import 0.10.0 → 1.0.8

data/lib/activerecord-import/adapters/mysql_adapter.rb

@@ -3,27 +3,27 @@ module ActiveRecord::Import::MysqlAdapter
   include ActiveRecord::Import::OnDuplicateKeyUpdateSupport
 
   NO_MAX_PACKET = 0
-  QUERY_OVERHEAD = 8 #This was shown to be true for MySQL, but it's not clear where the overhead is from.
+  QUERY_OVERHEAD = 8 # This was shown to be true for MySQL, but it's not clear where the overhead is from.
 
   # +sql+ can be a single string or an array. If it is an array all
   # elements that are in position >= 1 will be appended to the final SQL.
-  def insert_many( sql, values, *args ) # :nodoc:
+  def insert_many( sql, values, options = {}, *args ) # :nodoc:
     # the number of inserts default
     number_of_inserts = 0
 
-    base_sql,post_sql = if sql.is_a?( String )
-      [ sql, '' ]
+    base_sql, post_sql = if sql.is_a?( String )
+      [sql, '']
     elsif sql.is_a?( Array )
-      [ sql.shift, sql.join( ' ' ) ]
+      [sql.shift, sql.join( ' ' )]
     end
 
     sql_size = QUERY_OVERHEAD + base_sql.size + post_sql.size
 
     # the number of bytes the requested insert statement values will take up
-    values_in_bytes = values.sum {|value| value.bytesize }
+    values_in_bytes = values.sum(&:bytesize)
 
     # the number of bytes (commas) it will take to comma separate our values
-    comma_separated_bytes = values.size-1
+    comma_separated_bytes = values.size - 1
 
     # the total number of bytes required if this statement is one statement
     total_bytes = sql_size + values_in_bytes + comma_separated_bytes
@@ -31,72 +31,99 @@ module ActiveRecord::Import::MysqlAdapter
     max = max_allowed_packet
 
     # if we can insert it all as one statement
-    if NO_MAX_PACKET == max or total_bytes < max
+    if NO_MAX_PACKET == max || total_bytes <= max || options[:force_single_insert]
       number_of_inserts += 1
       sql2insert = base_sql + values.join( ',' ) + post_sql
       insert( sql2insert, *args )
     else
       value_sets = ::ActiveRecord::Import::ValueSetsBytesParser.parse(values,
-        :reserved_bytes => sql_size,
-        :max_bytes => max)
-      value_sets.each do |values|
-        number_of_inserts += 1
-        sql2insert = base_sql + values.join( ',' ) + post_sql
-        insert( sql2insert, *args )
+        reserved_bytes: sql_size,
+        max_bytes: max)
+
+      transaction(requires_new: true) do
+        value_sets.each do |value_set|
+          number_of_inserts += 1
+          sql2insert = base_sql + value_set.join( ',' ) + post_sql
+          insert( sql2insert, *args )
+        end
       end
     end
 
-    [number_of_inserts,[]]
+    ActiveRecord::Import::Result.new([], number_of_inserts, [], [])
   end
 
   # Returns the maximum number of bytes that the server will allow
   # in a single packet
   def max_allowed_packet # :nodoc:
     @max_allowed_packet ||= begin
-      result = execute( "SHOW VARIABLES like 'max_allowed_packet';" )
+      result = execute( "SELECT @@max_allowed_packet" )
       # original Mysql gem responds to #fetch_row while Mysql2 responds to #first
-      val = result.respond_to?(:fetch_row) ? result.fetch_row[1] : result.first[1]
+      val = result.respond_to?(:fetch_row) ? result.fetch_row[0] : result.first[0]
       val.to_i
     end
   end
 
+  def pre_sql_statements( options)
+    sql = []
+    sql << "IGNORE" if options[:ignore] || options[:on_duplicate_key_ignore]
+    sql + super
+  end
+
+  # Add a column to be updated on duplicate key update
+  def add_column_for_on_duplicate_key_update( column, options = {} ) # :nodoc:
+    if (columns = options[:on_duplicate_key_update])
+      case columns
+      when Array then columns << column.to_sym unless columns.include?(column.to_sym)
+      when Hash then columns[column.to_sym] = column.to_sym
+      end
+    end
+  end
+
   # Returns a generated ON DUPLICATE KEY UPDATE statement given the passed
   # in +args+.
   def sql_for_on_duplicate_key_update( table_name, *args ) # :nodoc:
     sql = ' ON DUPLICATE KEY UPDATE '
     arg = args.first
+    locking_column = args.last
     if arg.is_a?( Array )
-      sql << sql_for_on_duplicate_key_update_as_array( table_name, arg )
+      sql << sql_for_on_duplicate_key_update_as_array( table_name, locking_column, arg )
     elsif arg.is_a?( Hash )
-      sql << sql_for_on_duplicate_key_update_as_hash( table_name, arg )
+      sql << sql_for_on_duplicate_key_update_as_hash( table_name, locking_column, arg )
     elsif arg.is_a?( String )
       sql << arg
     else
-      raise ArgumentError.new( "Expected Array or Hash" )
+      raise ArgumentError, "Expected Array or Hash"
     end
     sql
   end
 
-  def sql_for_on_duplicate_key_update_as_array( table_name, arr )  # :nodoc:
+  def sql_for_on_duplicate_key_update_as_array( table_name, locking_column, arr ) # :nodoc:
     results = arr.map do |column|
       qc = quote_column_name( column )
       "#{table_name}.#{qc}=VALUES(#{qc})"
     end
+    increment_locking_column!(table_name, results, locking_column)
     results.join( ',' )
   end
 
-  def sql_for_on_duplicate_key_update_as_hash( table_name, hsh ) # :nodoc:
-    sql = ' ON DUPLICATE KEY UPDATE '
+  def sql_for_on_duplicate_key_update_as_hash( table_name, locking_column, hsh ) # :nodoc:
     results = hsh.map do |column1, column2|
       qc1 = quote_column_name( column1 )
       qc2 = quote_column_name( column2 )
       "#{table_name}.#{qc1}=VALUES( #{qc2} )"
     end
+    increment_locking_column!(table_name, results, locking_column)
     results.join( ',')
   end
 
-  #return true if the statement is a duplicate key record error
-  def duplicate_key_update_error?(exception)# :nodoc:
+  # Return true if the statement is a duplicate key record error
+  def duplicate_key_update_error?(exception) # :nodoc:
     exception.is_a?(ActiveRecord::StatementInvalid) && exception.to_s.include?('Duplicate entry')
   end
+
+  def increment_locking_column!(table_name, results, locking_column)
+    if locking_column.present?
+      results << "`#{locking_column}`=#{table_name}.`#{locking_column}`+1"
+    end
+  end
 end

data/lib/activerecord-import/adapters/postgresql_adapter.rb

@@ -1,19 +1,64 @@
 module ActiveRecord::Import::PostgreSQLAdapter
   include ActiveRecord::Import::ImportSupport
+  include ActiveRecord::Import::OnDuplicateKeyUpdateSupport
 
-  def insert_many( sql, values, *args ) # :nodoc:
+  MIN_VERSION_FOR_UPSERT = 90_500
+
+  def insert_many( sql, values, options = {}, *args ) # :nodoc:
     number_of_inserts = 1
+    returned_values = []
+    ids = []
+    results = []
 
-    base_sql,post_sql = if sql.is_a?( String )
-      [ sql, '' ]
+    base_sql, post_sql = if sql.is_a?( String )
+      [sql, '']
     elsif sql.is_a?( Array )
-      [ sql.shift, sql.join( ' ' ) ]
+      [sql.shift, sql.join( ' ' )]
     end
 
     sql2insert = base_sql + values.join( ',' ) + post_sql
-    ids = select_values( sql2insert, *args )
 
-    [number_of_inserts,ids]
+    columns = returning_columns(options)
+    if columns.blank? || (options[:no_returning] && !options[:recursive])
+      insert( sql2insert, *args )
+    else
+      returned_values = if columns.size > 1
+        # Select composite columns
+        select_rows( sql2insert, *args )
+      else
+        select_values( sql2insert, *args )
+      end
+      clear_query_cache if query_cache_enabled
+    end
+
+    if options[:returning].blank?
+      ids = returned_values
+    elsif options[:primary_key].blank?
+      results = returned_values
+    else
+      # split primary key and returning columns
+      ids, results = split_ids_and_results(returned_values, columns, options)
+    end
+
+    ActiveRecord::Import::Result.new([], number_of_inserts, ids, results)
+  end
+
+  def split_ids_and_results(values, columns, options)
+    ids = []
+    results = []
+    id_indexes = Array(options[:primary_key]).map { |key| columns.index(key) }
+    returning_indexes = Array(options[:returning]).map { |key| columns.index(key) }
+
+    values.each do |value|
+      value_array = Array(value)
+      ids << id_indexes.map { |i| value_array[i] }
+      results << returning_indexes.map { |i| value_array[i] }
+    end
+
+    ids.map!(&:first) if id_indexes.size == 1
+    results.map!(&:first) if returning_indexes.size == 1
+
+    [ids, results]
   end
 
   def next_value_for_sequence(sequence_name)
@@ -21,14 +66,146 @@ module ActiveRecord::Import::PostgreSQLAdapter
   end
 
   def post_sql_statements( table_name, options ) # :nodoc:
-    unless options[:primary_key].blank?
-      super(table_name, options) << (" RETURNING #{options[:primary_key]}")
+    sql = []
+
+    if supports_on_duplicate_key_update?
+      # Options :recursive and :on_duplicate_key_ignore are mutually exclusive
+      if (options[:ignore] || options[:on_duplicate_key_ignore]) && !options[:on_duplicate_key_update] && !options[:recursive]
+        sql << sql_for_on_duplicate_key_ignore( table_name, options[:on_duplicate_key_ignore] )
+      end
+    elsif logger && options[:on_duplicate_key_ignore] && !options[:on_duplicate_key_update]
+      logger.warn "Ignoring on_duplicate_key_ignore because it is not supported by the database."
+    end
+
+    sql += super(table_name, options)
+
+    columns = returning_columns(options)
+    unless columns.blank? || (options[:no_returning] && !options[:recursive])
+      sql << " RETURNING \"#{columns.join('", "')}\""
+    end
+
+    sql
+  end
+
+  def returning_columns(options)
+    columns = []
+    columns += Array(options[:primary_key]) if options[:primary_key].present?
+    columns |= Array(options[:returning]) if options[:returning].present?
+    columns
+  end
+
+  # Add a column to be updated on duplicate key update
+  def add_column_for_on_duplicate_key_update( column, options = {} ) # :nodoc:
+    arg = options[:on_duplicate_key_update]
+    if arg.is_a?( Hash )
+      columns = arg.fetch( :columns ) { arg[:columns] = [] }
+      case columns
+      when Array then columns << column.to_sym unless columns.include?( column.to_sym )
+      when Hash then columns[column.to_sym] = column.to_sym
+      end
+    elsif arg.is_a?( Array )
+      arg << column.to_sym unless arg.include?( column.to_sym )
+    end
+  end
+
+  # Returns a generated ON CONFLICT DO NOTHING statement given the passed
+  # in +args+.
+  def sql_for_on_duplicate_key_ignore( table_name, *args ) # :nodoc:
+    arg = args.first
+    conflict_target = sql_for_conflict_target( arg ) if arg.is_a?( Hash )
+    " ON CONFLICT #{conflict_target}DO NOTHING"
+  end
+
+  # Returns a generated ON CONFLICT DO UPDATE statement given the passed
+  # in +args+.
+  def sql_for_on_duplicate_key_update( table_name, *args ) # :nodoc:
+    arg, primary_key, locking_column = args
+    arg = { columns: arg } if arg.is_a?( Array ) || arg.is_a?( String )
+    return unless arg.is_a?( Hash )
+
+    sql = ' ON CONFLICT '
+    conflict_target = sql_for_conflict_target( arg )
+
+    columns = arg.fetch( :columns, [] )
+    condition = arg[:condition]
+    if columns.respond_to?( :empty? ) && columns.empty?
+      return sql << "#{conflict_target}DO NOTHING"
+    end
+
+    conflict_target ||= sql_for_default_conflict_target( table_name, primary_key )
+    unless conflict_target
+      raise ArgumentError, 'Expected :conflict_target or :constraint_name to be specified'
+    end
+
+    sql << "#{conflict_target}DO UPDATE SET "
+    if columns.is_a?( Array )
+      sql << sql_for_on_duplicate_key_update_as_array( table_name, locking_column, columns )
+    elsif columns.is_a?( Hash )
+      sql << sql_for_on_duplicate_key_update_as_hash( table_name, locking_column, columns )
+    elsif columns.is_a?( String )
+      sql << columns
     else
-      super(table_name, options)
+      raise ArgumentError, 'Expected :columns to be an Array or Hash'
     end
+
+    sql << " WHERE #{condition}" if condition.present?
+
+    sql
   end
 
-  def support_setting_primary_key_of_imported_objects?
+  def sql_for_on_duplicate_key_update_as_array( table_name, locking_column, arr ) # :nodoc:
+    results = arr.map do |column|
+      qc = quote_column_name( column )
+      "#{qc}=EXCLUDED.#{qc}"
+    end
+    increment_locking_column!(table_name, results, locking_column)
+    results.join( ',' )
+  end
+
+  def sql_for_on_duplicate_key_update_as_hash( table_name, locking_column, hsh ) # :nodoc:
+    results = hsh.map do |column1, column2|
+      qc1 = quote_column_name( column1 )
+      qc2 = quote_column_name( column2 )
+      "#{qc1}=EXCLUDED.#{qc2}"
+    end
+    increment_locking_column!(table_name, results, locking_column)
+    results.join( ',' )
+  end
+
+  def sql_for_conflict_target( args = {} )
+    constraint_name = args[:constraint_name]
+    conflict_target = args[:conflict_target]
+    index_predicate = args[:index_predicate]
+    if constraint_name.present?
+      "ON CONSTRAINT #{constraint_name} "
+    elsif conflict_target.present?
+      '(' << Array( conflict_target ).reject( &:blank? ).join( ', ' ) << ') '.tap do |sql|
+        sql << "WHERE #{index_predicate} " if index_predicate
+      end
+    end
+  end
+
+  def sql_for_default_conflict_target( table_name, primary_key )
+    conflict_target = Array(primary_key).join(', ')
+    "(#{conflict_target}) " if conflict_target.present?
+  end
+
+  # Return true if the statement is a duplicate key record error
+  def duplicate_key_update_error?(exception) # :nodoc:
+    exception.is_a?(ActiveRecord::StatementInvalid) && exception.to_s.include?('duplicate key')
+  end
+
+  def supports_on_duplicate_key_update?
+    database_version >= MIN_VERSION_FOR_UPSERT
+  end
+
+  def supports_setting_primary_key_of_imported_objects?
     true
   end
+
+  private
+
+  def database_version
+    defined?(postgresql_version) ? postgresql_version : super
+  end
 end

data/lib/activerecord-import/adapters/sqlite3_adapter.rb

@@ -1,43 +1,174 @@
 module ActiveRecord::Import::SQLite3Adapter
   include ActiveRecord::Import::ImportSupport
+  include ActiveRecord::Import::OnDuplicateKeyUpdateSupport
 
-  MIN_VERSION_FOR_IMPORT = "3.7.11"
+  MIN_VERSION_FOR_IMPORT = "3.7.11".freeze
+  MIN_VERSION_FOR_UPSERT = "3.24.0".freeze
   SQLITE_LIMIT_COMPOUND_SELECT = 500
 
   # Override our conformance to ActiveRecord::Import::ImportSupport interface
   # to ensure that we only support import in supported version of SQLite.
   # Which INSERT statements with multiple value sets was introduced in 3.7.11.
-  def supports_import?(current_version=self.sqlite_version)
-    if current_version >= MIN_VERSION_FOR_IMPORT
-      true
-    else
-      false
-    end
+  def supports_import?
+    database_version >= MIN_VERSION_FOR_IMPORT
+  end
+
+  def supports_on_duplicate_key_update?
+    database_version >= MIN_VERSION_FOR_UPSERT
   end
 
   # +sql+ can be a single string or an array. If it is an array all
   # elements that are in position >= 1 will be appended to the final SQL.
-  def insert_many(sql, values, *args) # :nodoc:
+  def insert_many( sql, values, _options = {}, *args ) # :nodoc:
     number_of_inserts = 0
-    base_sql,post_sql = if sql.is_a?( String )
-      [ sql, '' ]
+
+    base_sql, post_sql = if sql.is_a?( String )
+      [sql, '']
     elsif sql.is_a?( Array )
-      [ sql.shift, sql.join( ' ' ) ]
+      [sql.shift, sql.join( ' ' )]
     end
 
     value_sets = ::ActiveRecord::Import::ValueSetsRecordsParser.parse(values,
-      :max_records => SQLITE_LIMIT_COMPOUND_SELECT)
+      max_records: SQLITE_LIMIT_COMPOUND_SELECT)
 
-    value_sets.each do |values|
-      number_of_inserts += 1
-      sql2insert = base_sql + values.join( ',' ) + post_sql
-      insert( sql2insert, *args )
+    transaction(requires_new: true) do
+      value_sets.each do |value_set|
+        number_of_inserts += 1
+        sql2insert = base_sql + value_set.join( ',' ) + post_sql
+        insert( sql2insert, *args )
+      end
     end
 
-    [number_of_inserts,[]]
+    ActiveRecord::Import::Result.new([], number_of_inserts, [], [])
+  end
+
+  def pre_sql_statements( options )
+    sql = []
+    # Options :recursive and :on_duplicate_key_ignore are mutually exclusive
+    if !supports_on_duplicate_key_update? && (options[:ignore] || options[:on_duplicate_key_ignore])
+      sql << "OR IGNORE"
+    end
+    sql + super
+  end
+
+  def post_sql_statements( table_name, options ) # :nodoc:
+    sql = []
+
+    if supports_on_duplicate_key_update?
+      # Options :recursive and :on_duplicate_key_ignore are mutually exclusive
+      if (options[:ignore] || options[:on_duplicate_key_ignore]) && !options[:on_duplicate_key_update]
+        sql << sql_for_on_duplicate_key_ignore( options[:on_duplicate_key_ignore] )
+      end
+    end
+
+    sql + super
   end
 
   def next_value_for_sequence(sequence_name)
     %{nextval('#{sequence_name}')}
   end
+
+  # Add a column to be updated on duplicate key update
+  def add_column_for_on_duplicate_key_update( column, options = {} ) # :nodoc:
+    arg = options[:on_duplicate_key_update]
+    if arg.is_a?( Hash )
+      columns = arg.fetch( :columns ) { arg[:columns] = [] }
+      case columns
+      when Array then columns << column.to_sym unless columns.include?( column.to_sym )
+      when Hash then columns[column.to_sym] = column.to_sym
+      end
+    elsif arg.is_a?( Array )
+      arg << column.to_sym unless arg.include?( column.to_sym )
+    end
+  end
+
+  # Returns a generated ON CONFLICT DO NOTHING statement given the passed
+  # in +args+.
+  def sql_for_on_duplicate_key_ignore( *args ) # :nodoc:
+    arg = args.first
+    conflict_target = sql_for_conflict_target( arg ) if arg.is_a?( Hash )
+    " ON CONFLICT #{conflict_target}DO NOTHING"
+  end
+
+  # Returns a generated ON CONFLICT DO UPDATE statement given the passed
+  # in +args+.
+  def sql_for_on_duplicate_key_update( table_name, *args ) # :nodoc:
+    arg, primary_key, locking_column = args
+    arg = { columns: arg } if arg.is_a?( Array ) || arg.is_a?( String )
+    return unless arg.is_a?( Hash )
+
+    sql = ' ON CONFLICT '
+    conflict_target = sql_for_conflict_target( arg )
+
+    columns = arg.fetch( :columns, [] )
+    condition = arg[:condition]
+    if columns.respond_to?( :empty? ) && columns.empty?
+      return sql << "#{conflict_target}DO NOTHING"
+    end
+
+    conflict_target ||= sql_for_default_conflict_target( primary_key )
+    unless conflict_target
+      raise ArgumentError, 'Expected :conflict_target to be specified'
+    end
+
+    sql << "#{conflict_target}DO UPDATE SET "
+    if columns.is_a?( Array )
+      sql << sql_for_on_duplicate_key_update_as_array( table_name, locking_column, columns )
+    elsif columns.is_a?( Hash )
+      sql << sql_for_on_duplicate_key_update_as_hash( table_name, locking_column, columns )
+    elsif columns.is_a?( String )
+      sql << columns
+    else
+      raise ArgumentError, 'Expected :columns to be an Array or Hash'
+    end
+
+    sql << " WHERE #{condition}" if condition.present?
+
+    sql
+  end
+
+  def sql_for_on_duplicate_key_update_as_array( table_name, locking_column, arr ) # :nodoc:
+    results = arr.map do |column|
+      qc = quote_column_name( column )
+      "#{qc}=EXCLUDED.#{qc}"
+    end
+    increment_locking_column!(table_name, results, locking_column)
+    results.join( ',' )
+  end
+
+  def sql_for_on_duplicate_key_update_as_hash( table_name, locking_column, hsh ) # :nodoc:
+    results = hsh.map do |column1, column2|
+      qc1 = quote_column_name( column1 )
+      qc2 = quote_column_name( column2 )
+      "#{qc1}=EXCLUDED.#{qc2}"
+    end
+    increment_locking_column!(table_name, results, locking_column)
+    results.join( ',' )
+  end
+
+  def sql_for_conflict_target( args = {} )
+    conflict_target = args[:conflict_target]
+    index_predicate = args[:index_predicate]
+    if conflict_target.present?
+      '(' << Array( conflict_target ).reject( &:blank? ).join( ', ' ) << ') '.tap do |sql|
+        sql << "WHERE #{index_predicate} " if index_predicate
+      end
+    end
+  end
+
+  def sql_for_default_conflict_target( primary_key )
+    conflict_target = Array(primary_key).join(', ')
+    "(#{conflict_target}) " if conflict_target.present?
+  end
+
+  # Return true if the statement is a duplicate key record error
+  def duplicate_key_update_error?(exception) # :nodoc:
+    exception.is_a?(ActiveRecord::StatementInvalid) && exception.to_s.include?('duplicate key')
+  end
+
+  private
+
+  def database_version
+    defined?(sqlite_version) ? sqlite_version : super
+  end
 end

data/lib/activerecord-import/base.rb

@@ -3,31 +3,37 @@ require "active_record"
 require "active_record/version"
 
 module ActiveRecord::Import
-  AdapterPath = "activerecord-import/active_record/adapters"
+  ADAPTER_PATH = "activerecord-import/active_record/adapters".freeze
 
   def self.base_adapter(adapter)
     case adapter
-    when 'mysqlspatial' then 'mysql'
+    when 'mysql2_makara' then 'mysql2'
     when 'mysql2spatial' then 'mysql2'
     when 'spatialite' then 'sqlite3'
+    when 'postgresql_makara' then 'postgresql'
+    when 'makara_postgis' then 'postgresql'
     when 'postgis' then 'postgresql'
+    when 'cockroachdb' then 'postgresql'
     else adapter
     end
   end
 
   # Loads the import functionality for a specific database adapter
   def self.require_adapter(adapter)
-    require File.join(AdapterPath,"/abstract_adapter")
-    begin
-      require File.join(AdapterPath,"/#{base_adapter(adapter)}_adapter")
-    rescue LoadError
-      # fallback
-    end
+    require File.join(ADAPTER_PATH, "/#{base_adapter(adapter)}_adapter")
+  rescue LoadError
+    # fallback
   end
 
   # Loads the import functionality for the passed in ActiveRecord connection
   def self.load_from_connection_pool(connection_pool)
-    require_adapter connection_pool.spec.config[:adapter]
+    adapter =
+      if connection_pool.respond_to?(:db_config) # ActiveRecord >= 6.1
+        connection_pool.db_config.adapter
+      else
+        connection_pool.spec.config[:adapter]
+      end
+    require_adapter adapter
   end
 end