RDF::Graph should use Repository#insert_statements

[no-reply]

RDF::Graph relies on Writable#insert_statements, which always inserts statements individually.

Some (hopefully all) Repository implementations that interact with upstream data implement #insert_statements as a single request. Using Graph to interact with a repository results in sending individual requests, regardless of the Repository implementation.

e.g., the following results in some_statements.count individual requests to some_repository:

my_graph = RDF::Graph.new(RDF::URI('http://ex.org/my_graph'), data: some_repository)
my_graph << some_statements

I'm working on a fix for this.

[gkellogg]

I think the way to do this may be to redirect to Repository#insert_statements and update each statement's graph_name using map or in an enumerator, depending on the form of statements. Note that insert_statements has a special case if the first element is an Enumerable.

[no-reply]

I came up with:

    def insert_statements(statements)
      enum = statements.respond_to?(:each_statement) ? statements.each_statement : statements
      @data.insert(*enum.map { |st| statement = st.dup; statement.graph_name = graph_name; statement })
    end

But planning to dig around a little more to make sure that's the best plan.

[gkellogg]

There was a recent change to make sure that it could take an enumerator, as the *enum.map... can create too many arguments; see #214 (comment).

I was thinking of something like the following:

def insert_statements(statements)
  enum = Enumerable::Enumerator.new do |yielder|
    this.send(statements.respond_to?(:each_statement) ? :each_statement : :each) do |st|
      st = st.dup
      st.graph_name = graph_name;
      yielder << st
    end
  end
  @data.insert_statements(enum)
end

This allows @data.insert_statements to take chunks of statements to insert without evaluating everything at once.