activerdf_sparql 1.3.4 → 1.3.5

data/CHANGELOG

@@ -1,3 +1,8 @@
+== activerdf_sparql (1.3.5) Mon, 01 Oct 2007 12:03:40 +0200
+  * support for POSTing SPARQL queres (Edward Benson)
+  * added experimental sparql query cache: it never expires (Benjamin Heitmann)
+  * reverted gem dependency to json (Benjamin Heitmann)
+
 == activerdf_sparql (1.3.4) Fri, 21 Sep 2007 10:06:57 +0100
   * parse bnodes in SPARQL queries
 

data/lib/activerdf_sparql/sparql.rb

@@ -9,28 +9,38 @@ require "#{File.dirname(__FILE__)}/sparql_result_parser"
 class SparqlAdapter < ActiveRdfAdapter
 	$activerdflog.info "loading SPARQL adapter"
 	ConnectionPool.register_adapter(:sparql, self)
+
+  attr_reader :engine
+  attr_reader :caching
+
+  @@sparql_cache = {}
+
+  def SparqlAdapter.get_cache
+    return @@sparql_cache
+  end
 	
 	# Instantiate the connection with the SPARQL Endpoint.
 	# available parameters:
 	# * :url => url: endpoint location e.g. "http://m3pe.org:8080/repositories/test-people"
 	# * :results => one of :xml, :json, :sparql_xml
-  attr_reader :engine
-
+  # * :request_method => :get (default) or :post
+  # * :timeout => timeout in seconds to wait for endpoint response
 	def initialize(params = {})	
 		@reads = true
 		@writes = false
 
 		@url = params[:url] || ''
+    @caching = params[:caching] || false
+    @timeout = params[:timeout] || 50
+
 		@result_format = params[:results] || :json
+		raise ActiveRdfError, "Result format unsupported" unless [:xml, :json, :sparql_xml].include? @result_format
+
     @engine = params[:engine]
+		raise ActiveRdfError, "SPARQL engine unsupported" unless [:yars2, :sesame2, :joseki, :virtuoso].include? @engine
 
-    supported_engines = [:yars2, :sesame2, :joseki, :virtuoso]
-		raise ActiveRdfError, "SPARQL engine unsupported" unless supported_engines.include?(@engine)
-		
-		known_formats = [:xml, :json, :sparql_xml]
-		raise ActiveRdfError, "Result format unsupported" unless known_formats.include?(@result_format)
-		
-		$activerdflog.info "SPARQL adapter initialised #{inspect}"
+    @request_method = params[:request_method] || :get
+    raise ActiveRdfError, "Request method unsupported" unless [:get,:post].include? @request_method
 	end
 
 	def size
@@ -40,27 +50,49 @@ class SparqlAdapter < ActiveRdfAdapter
 	# query datastore with query string (SPARQL), returns array with query results
 	# may be called with a block
 	def query(query, &block)
-		time = Time.now
     qs = Query2SPARQL.translate(query)
-		execute_sparql_query(qs, header(query), &block)
+
+    if @caching
+       result = query_cache(qs)
+       if result.nil?
+         $activerdflog.debug "cache miss for query #{qs}"
+       else
+         $activerdflog.debug "cache hit for query #{qs}"
+         return result
+       end
+    end
+
+    result = execute_sparql_query(qs, header(query), &block)
+    add_to_cache(qs, result) if @caching
+		result = [] if result == "timeout"
+		return result
 	end
 		
 	# do the real work of executing the sparql query
 	def execute_sparql_query(qs, header=nil, &block)
     header = header(nil) if header.nil?
 
-		# encoding query string in URL
-		url = "#@url?query=#{CGI.escape(qs)}"
-    #url += "&content-type=#{CGI.escape('application/sparql-results+xml')}" if @yars2
-    url = url.gsub("DISTINCT", "") if @yars2
-		$activerdflog.debug "querying #{url}"
-
     # querying sparql endpoint
+    require 'timeout'
 		response = ''
 		begin 
-		  open(url, header) do |f|
-				response = f.read
-			end
+      case @request_method
+      when :get
+        # encoding query string in URL
+        url = "#@url?query=#{CGI.escape(qs)}"
+        $activerdflog.debug "GET #{url}"
+        timeout(@timeout) do
+          open(url, header) do |f|
+            response = f.read
+          end
+        end
+      when :post
+        $activerdflog.debug "POST #@url with #{qs}"
+        response = Net::HTTP.post_form(URI.parse(@url),{'query'=>qs}).body
+      end
+  	rescue Timeout::Error
+			raise ActiveRdfError, "timeout on SPARQL endpoint"
+  	  return "timeout"
 		rescue OpenURI::HTTPError => e
 			raise ActiveRdfError, "could not query SPARQL endpoint, server said: #{e}"
 			return []
@@ -68,7 +100,6 @@ class SparqlAdapter < ActiveRdfAdapter
 			raise ActiveRdfError, "connection refused on SPARQL endpoint #@url"
 			return []
 	 	end
-    $activerdflog.debug "response:\n#{response}"
 
     # we parse content depending on the result format
     results = case @result_format
@@ -92,6 +123,26 @@ class SparqlAdapter < ActiveRdfAdapter
 	end
 	
 	private
+  def add_to_cache(query_string, result)
+    unless result.nil? or result.empty?
+      if result == "timeout"
+        @@sparql_cache.store(query_string, [])
+      else 
+        $activerdflog.debug "adding to sparql cache - query: #{query_string}"
+        @@sparql_cache.store(query_string, result) 
+      end
+    end
+  end
+  
+  
+  def query_cache(query_string)
+    if @@sparql_cache.include?(query_string)
+      return @@sparql_cache.fetch(query_string)
+    else
+      return nil
+    end
+  end
+
 	# constructs correct HTTP header for selected query-result format
 	def header(query)
 		case @result_format
@@ -146,4 +197,7 @@ class SparqlAdapter < ActiveRdfAdapter
       value.to_s
     end
   end
+  
+  
+  
 end

data/test/test_sparql_adapter.rb

@@ -56,7 +56,6 @@ class TestSparqlAdapter < Test::Unit::TestCase
     begin
       movies = Query.new.
         select(:title).
-        where(:film, RDF.type, RDFS::Resource.new('http://dbpedia.org/class/yago/film')).
         where(:film, RDFS.label, :title).
         where(:title, RDFS::Resource.new('bif:contains'), 'kill').
         filter_regex(:title, /Kill$/).execute

metadata

@@ -3,8 +3,8 @@ rubygems_version: 0.9.4
 specification_version: 1
 name: activerdf_sparql
 version: !ruby/object:Gem::Version 
-  version: 1.3.4
-date: 2007-09-21 00:00:00 +01:00
+  version: 1.3.5
+date: 2007-10-01 00:00:00 +02:00
 summary: ActiveRDF adapter to SPARQL endpoint
 require_paths: 
 - lib