ruby-rdf/rdf-sesame

RDF.rb plugin for accessing Sesame 2.0-compatible repositories.

Sesame 2.0 Adapter for RDF.rb

This is an RDF.rb plugin that adds support for Sesame 2.0-compatible RDF repositories accessed using Sesame's HTTP API.

• https://github.com/ruby-rdf/rdf-sesame

Documentation

• {RDF::Sesame}
  • {RDF::Sesame::Connection}
  • {RDF::Sesame::Server}
  • {RDF::Sesame::Repository}

This adapter implements the RDF::Repository interface; refer to the relevant RDF.rb API documentation for further usage instructions.

Limitations

• This adapter does not contain any SPARQL support. To use a Sesame endpoint with SPARQL, see the SPARQL::Client gem.
• This adapter is not yet optimized for RDF.rb 0.2.x's bulk-operation APIs, meaning that statement insertions and deletions are currently performed one by one; this may affect the performance of loading large datasets into Sesame. This will be addressed in an upcoming version.
• This adapter is not yet optimized for RDF.rb 0.2.x's enhanced query APIs; this may adversely affect triple pattern query performance. This will be addressed in an upcoming version.

Caveats

• Sesame rewrites blank node identifiers on inserted statements. For example, if you supply a _:foobar identifier, it becomes something like _:node156oo6equx12769 as soon as you insert it into a Sesame repository. This means that you can't construct and insert a statement containing a blank node term and then expect to successfully be able check for its existence using the Repository#has_statement? method. This is also the reason that 7 specs for RDF::Repository currently fail with this adapter.

Dependencies

• RDF.rb (~> 1.1)
• JSON (~> 1.8)

Installation

The recommended installation method is via RubyGems. To install the latest official release of the RDF::Sesame gem, do:

% [sudo] gem install rdf-sesame

Download

To get a local working copy of the development repository, do:

% git clone git://github.com/ruby-rdf/rdf-sesame.git

Alternatively, download the latest development version as a tarball as follows:

% wget https://github.com/ruby-rdf/rdf-sesame/tarball/master

Tests

In order to run test, you should use dotenv (which is set as a development dependency) and have a .env file with the following environment variables:

SESAME_URL=http://localhost:8080/openrdf-sesame # has to be an accessible openrdf-sesame server
SESAME_REPOSITORY=integration_test # has to be the name of an existing repository on which we'll be doing the integration tests.

You can then run them with:

% bundle exec dotenv rake spec

Mailing List

• http://lists.w3.org/Archives/Public/public-rdf-ruby/

Author

• Arto Bendiken - http://ar.to/
• Aymeric Brisse

Contributors

• Slava Kravchenko

License

This is free and unencumbered public domain software. For more information, see http://unlicense.org/ or the accompanying {file:UNLICENSE} file.

rdf-sesame's Issues

Optimize triple pattern queries

This adapter is not yet optimized for RDF.rb 0.2.x's enhanced query APIs; this may adversely affect triple pattern query performance.

This should be addressed in an upcoming version.

Bug inserting RDF-XML based graphs or reader into repositories

Bug inserting RDF-XML based graphs into repositories.

I am able to insert NTriple based readers into a Sesame repository. However, RDFXML ones cannot be inserted. Here's what I've been trying:
require 'rdf'
require 'rdf/sesame'
require 'rdf/rdfxml'
repository = RDF::Sesame::Repository.new('http://localhost:8080/openrdf-sesame/repositories/test')

Successfully returns count of 47

reader = RDF::NTriples::Reader.new(File.read(RAILS_ROOT + '/public/doap.nt'))
repository.clear
repository.insert(reader)
repository.count

Only returns count of 3

reader = RDF::RDFXML::Reader.new(File.read(RAILS_ROOT + '/public/doap.xml'))
repository = RDF::Sesame::Repository.new('http://localhost:8080/openrdf-sesame/repositories/test')
repository.clear
repository.insert(reader)
repository.count

Graph count returns 47, but insertion into repository fails

stream = File.open(RAILS_ROOT + '/public/doap.xml')
graph = RDF::Graph.new
RDF::RDFXML::Reader.new(stream).each { |s| graph << s }
puts graph.count

Now that we have a graph, insert into repository

This inserts 3 entries, but not the entire 47

repository.clear
repository.insert(graph.to_a)
repository.count

Same thing, returns 3

repository.clear
repository.insert(*graph.to_a)
repository.count

Same thing, returns 3

repository.clear
repository<<graph.to_rdf
repository.count

The << method in writable.rb looks like it detects a graph and then

calls insert_graph, but none of these seem to work either

repository<<graph
repository.insert(graph)
repository.insert(graph.to_ntriples)
repository.insert(_graph.to_ntriples)
graph.each { |s| repository << s }
graph.each { |s| repository.insert(_s) }
graph.each { |s| repository.insert(s) }

stream = File.open(RAILS_ROOT + '/public/doap.xml')
graph = RDF::Graph.new
RDF::RDFXML::Reader.new(stream).each { |s| graph << s }
reader = RDF::NTriples::Reader.new(graph.to_ntriples)
repository.insert(reader)

I am able to get this to work with the RDF Context gem:
require 'rdf'
require 'rdf/sesame'
require 'rdf_context'
url = "http://localhost:8080/openrdf-sesame/repositories/test"
repository = RDF::Sesame::Repository.new(url)

The doap.xml file fails but rdf.xml does get parsed correctly

f = File.open(RAILS_ROOT + '/public/rdf.xml')
graph = RdfContext::Graph.new
graph.parse(f.read, nil, :type => :rdfxml)

statements = RDF::NTriples::Reader.new(graph.to_ntriples)
repository.insert(*statements)
repository.count
repository.clear!

Deleting statements in a context

With the RDF::Repository, I can set a context with the #set_context method. After than, every statement I'm going to insert into the repository will be assigned this context.

But when I delete statements from the repository, the context that is used to delete the statements is the one already assigned to the statement (RDF::Statement#context= or #graph=). So the two operations are not symmetric at all.

I'm going to do a PR in a few to adjust that behavior.

Implement support for persistent HTTP connections

Currently, connections are internally implemented using Net::HTTP from Ruby's standard library, and connections are always transient, i.e. they do not persist from one request to the next and must always be reopened when used.

We should implement support for persistent Keep-Alive connections as a future improvement.

Implement SPARQL support

This adapter does not contain any SPARQL support at present. To use a Sesame endpoint with SPARQL, see the SPARQL::Client gem.

Once the SPARQL client reaches a beta-quality level, we should add it as a dependency and have RDF::Sesame::Repository inherit from SPARQL::Client::Repository.

Improve CONSTRUCT detection

See https://github.com/ruby-rdf/rdf-sesame/blob/develop/lib/rdf/sesame/repository.rb#L311

Today the detection of SPARQL Construct to define the default format to text/plain is quite ugly (based on regular expression /bconstruct/b). It should be removed or enhance.

Optimize bulk operations

This adapter is not yet optimized for RDF.rb 0.2.x's bulk-operation APIs, meaning that statement insertions and deletions are currently performed one by one; this may affect the performance of loading large datasets into Sesame.

This should be addressed in an upcoming version.

Infinite loop when accessing statements or graphs of a Sesame repository.

I can't get the contexts used inside a Sesame repository anymore. It loops infinitely until the stack level is too deep and it raises an error. It does the same when I try to list the statements using '#each_statement".

The code to trigger this behavior is pretty simple:

server = RDF::Sesame::Server.new(db_url)
repository = server.repository(repository_name')

# either this
repository.each_statement do |s| # <= each starts the infinite loop 
  puts s.inspect
end

# or this
puts repository.graph_names.inspect

I'm pushing a PR in a few minutes, I'll reference this issue.