publish_my_data 0.0.3 → 0.0.4

data/README.md

@@ -8,8 +8,7 @@ The PublishMyData Community Edition is a Rails Engine that adds Linked Data func
 - A SPARQL Endpoint
 - Filterable lists of datasets and resources
 
-This is the same code that powers the enterprise, hosted version of PublishMyData. For more details see the [PublishMyData](http://publishmydata.com) website.
-
+This is the same core code that powers the enterprise, hosted version of PublishMyData. For more details see the [PublishMyData](http://publishmydata.com) website.
 
 ## How to use
 
@@ -36,7 +35,7 @@ This is the same code that powers the enterprise, hosted version of PublishMyDat
 - See the Rails guides for [more details on Rails Engines](http://guides.rubyonrails.org/engines.html).
 - PublishMyData doesn't supply a database. You need to install one. I recommend [Fuseki](http://jena.apache.org/documentation/serving_data/index.html), which runs on port 3030 by default.
 - The views currently supplied by this Rails engine are very rudimentary. Some nicer default views coming soon, but for now you'll probably just want to override them all in your app.
-- Warning: This gem is usable now, but the API is under constant development and flux at the moment
+- Warning: This gem is usable now, but the API is under constant development and flux at the moment, and tbh it's not very well documented! :)
 
 ## Licence
 

data/app/controllers/publish_my_data/application_controller.rb

@@ -1,48 +1,37 @@
 module PublishMyData
   class ApplicationController < ActionController::Base
 
-    rescue_from Tripod::Errors::ResourceNotFound, :with => :resource_not_found
-
-    # TODO: handle:
-    # 500s, timeouts (503) etc.
+    # Note: this order matters. Perversely, need to put more general errors first.
+    rescue_from Exception, :with => :handle_uncaught_error
+    rescue_from Tripod::Errors::ResourceNotFound, :with => :handle_resource_not_found
+    rescue_from RestClient::RequestTimeout, :with => :handle_timeout
 
     private
 
-    def resource_not_found(e)
+    # TODO: deal with javaascript errors - respond with 200
+
+    def handle_uncaught_error(e)
+      @e = e
+      # TODO: notify error handling service.
       respond_to do |format|
-        format.html { render(:template => "publish_my_data/errors/not_found", :layout => 'publish_my_data/error', :status => 404) and return false }
-        #TODO: ? format.js { render(:template => "publish_my_data/errors/not_found", :status => 200) and return false } # need to return success or the ajax request fails
-        format.any { head(:status => 404, :content_type => 'text/plain') and return false }
+        format.html { render(:template => "publish_my_data/errors/uncaught", :layout => 'publish_my_data/error', :status => 500) and return false }
+        format.any{ head(:status => 500, :content_type => 'text/plain') and return false }
       end
-    end
-
-    # from the criteria passed in, sets an instance var for @count and return
-    # a Kaminari::PaginatableArray, or Array (as appropriate to the format)
-    def paginate_resources(criteria)
 
-      get_pagination_params unless @got_pagination_params
-
-      @count = criteria.count #this has to happen first, before we modify the criteria with limit/offset
-      resources = criteria.limit(@limit).offset(@offset).resources
+    end
 
-      if request.format.html?
-        Kaminari.paginate_array(resources.to_a, total_count: @count).page(@page).per(@limit)
-      else
-        resources # non html versions just need the raw array
+    def handle_timeout(e)
+      respond_to do |format|
+        format.html { render(:template => "publish_my_data/errors/timeout", :layout => 'publish_my_data/error', :status => 503) and return false }
+        format.any { head(:status => 503, :content_type => 'text/plain') and return false }
       end
     end
 
-    def get_pagination_params
-      default_page_size = 20
-
-      @per_page = (params[:per_page] || default_page_size).to_i
-      @per_page = 10000 if @per_page > 10000
-      @page = (params[:page] || 1).to_i
-
-      @limit = @per_page
-      @offset = @limit.to_i * (@page.to_i-1)
-
-      @got_pagination_params = true
+    def handle_resource_not_found(e)
+      respond_to do |format|
+        format.html { render(:template => "publish_my_data/errors/not_found", :layout => 'publish_my_data/error', :status => 404) and return false }
+        format.any { head(:status => 404, :content_type => 'text/plain') and return false }
+      end
     end
 
   end

data/app/controllers/publish_my_data/datasets_controller.rb

@@ -7,10 +7,14 @@ module PublishMyData
 
     # /datasets/:id (where :id is the dataset 'slug')
     def show
+
       @dataset = Dataset.find_by_slug(params[:id])
+
+      @dataset.eager_load_object_triples! # for the owner URI label
+
       @types = RdfType.where('?s a ?uri').graph(@dataset.data_graph_uri).resources
 
-      if request.format.html?
+      if request.format && request.format.html?
         @type_resource_counts = {}
         @types.each do |t|
           @type_resource_counts[t.uri.to_s] = Resource.where("?uri a <#{t.uri.to_s}>").count
@@ -21,28 +25,12 @@ module PublishMyData
     end
 
     # /datasets?page=2&per_page=10
-    # /datasets?theme=foo
     def index
       dataset_criteria = Dataset.all
-      dataset_criteria = add_theme_filter(dataset_criteria)
-      @datasets = paginate_resources(dataset_criteria)
+      @pagination_params = PaginationParams.from_request(request)
+      @datasets = Paginator.new(dataset_criteria, @pagination_params).paginate
       respond_with(@datasets)
     end
 
-    def themes
-      # TODO: decide on how the themes will be represented in the metadata.
-      # Get a list of themes.
-      @themes = ['theme1', 'theme2']
-    end
-
-    private
-
-    def add_theme_filter(criteria)
-      unless params[:theme].blank?
-        @theme = params[:theme]
-        criteria.where("?uri <#{PMD_DS.theme}> '#{@theme}'")
-      end
-      criteria
-    end
   end
 end

data/app/controllers/publish_my_data/resources_controller.rb

@@ -14,7 +14,8 @@ module PublishMyData
       resource_criteria = add_type_filter(resource_criteria)
       resource_criteria = add_dataset_filter(resource_criteria)
 
-      @resources = paginate_resources(resource_criteria)
+      @pagination_params = PaginationParams.from_request(request)
+      @resources = Paginator.new(resource_criteria, @pagination_params).paginate
       respond_with(@resources)
     end
 
@@ -22,10 +23,7 @@ module PublishMyData
     def show
       uri = params[:uri]
       begin
-        # try to look it up
-        @resource = Resource.find(uri)
-        eager_load_labels()
-        respond_with(@resource)
+        render_resource_with_uri(uri)
       rescue Tripod::Errors::ResourceNotFound
         # if it's not there
         respond_to do |format|
@@ -50,29 +48,30 @@ module PublishMyData
     # http://example.com/doc/blah
     def doc
       uri = Resource.uri_from_host_and_doc_path(request.host, params[:path], params[:format])
-      @resource = Resource.find(uri)
-      eager_load_labels() if request.format.html?
-
-      # TODO: special views like ontology, dataset, etc?
-      respond_with(@resource) do |format|
-        format.html { render :template => 'publish_my_data/resources/show' }
-      end
+      render_resource_with_uri(uri)
     end
 
     # http://example.com/def/blah
     def definition
       uri = 'http://' + request.host + '/def/' + params[:path]
-      @resource = Resource.find(uri)
-      eager_load_labels() if request.format.html?
-
-      # TODO: special views like ontology, dataset, etc?
-      respond_with(@resource) do |format|
-        format.html { render :template => 'publish_my_data/resources/show' }
-      end
+      render_resource_with_uri(uri)
     end
 
     private
 
+    def render_resource_with_uri(uri)
+      resource = Resource.find(uri)
+
+      if request.format.html?
+        resource.eager_load_predicate_triples!
+        resource.eager_load_object_triples!
+      end
+
+      respond_with(resource) do |format|
+        format.html { render resource.render_params(request) }
+      end
+    end
+
     # TODO: move the filter management into an object
     def add_type_filter(criteria)
       unless params[:type_uri].blank?
@@ -92,11 +91,6 @@ module PublishMyData
       criteria
     end
 
-    def eager_load_labels
-      @resource.eager_load_predicate_triples!
-      @resource.eager_load_object_triples!
-    end
-
   end
 
 

data/app/controllers/publish_my_data/sparql_controller.rb

@@ -15,9 +15,12 @@ module PublishMyData
         @sparql_query = PublishMyData::SparqlQuery.new(@query_text, request.format.to_sym)
 
         if @sparql_query.allow_pagination?
-          get_pagination_params
-          @sparql_query_result = @sparql_query.paginate(@page, @per_page)
-          @more_pages = @sparql_query.as_pagination_query(@page, @per_page, 1).count > @per_page if request.format.html?
+          @pagination_params = PaginationParams.from_request(request)
+          @sparql_query_result = @sparql_query.paginate(@pagination_params.page, @pagination_params.per_page)
+          if request.format.html?
+            count = @sparql_query.as_pagination_query(@pagination_params.page, @pagination_params.per_page, 1).count
+            @more_pages = (count > @pagination_params.per_page)
+          end
         else
           @sparql_query_result = @sparql_query.execute
         end

data/app/controllers/publish_my_data/themes_controller.rb

@@ -0,0 +1,28 @@
+require_dependency "publish_my_data/application_controller"
+
+module PublishMyData
+  class ThemesController < ApplicationController
+
+    respond_to :html, :ttl, :rdf, :nt, :json
+
+    def index
+      # don't bother paginating this for now - there probably wont be that many themes
+      @themes = Theme.all.resources
+      respond_with(@themes)
+    end
+
+    def show
+      @theme = Theme.by_slug(params[:id])
+
+      if @theme
+        @pagination_params = PaginationParams.from_request(request)
+        @datasets = Paginator.new(@theme.datasets_criteria, @pagination_params).paginate
+        respond_with(@datasets)
+      else
+        raise Tripod::Errors::ResourceNotFound
+      end
+
+    end
+
+  end
+end

data/app/helpers/publish_my_data/resources_helper.rb

@@ -2,10 +2,10 @@ module PublishMyData
   module ResourcesHelper
 
     # uses eager loaded data to get the uri or label for a term
-    def resource_uri_or_label(term)
+    def resource_uri_or_label(resource, term)
 
       if term.uri?
-        res = @resource.get_related_resource(term, PublishMyData::Resource)
+        res = resource.get_related_resource(term, PublishMyData::Resource)
         if res
           link_to((res.label || res.uri.to_s), resource_path_from_uri(res.uri))
         else

data/app/models/publish_my_data/concept.rb

@@ -0,0 +1,14 @@
+module PublishMyData
+  class Concept
+    include Tripod::Resource
+    include PublishMyData::ResourceModule
+    field :label, RDF::RDFS.label
+    field :in_scheme, RDF::SKOS.inScheme
+
+    rdf_type RDF::SKOS.Concept
+
+    def concept_scheme
+      ConceptScheme.find(self.in_scheme) rescue nil if self.in_scheme
+    end
+  end
+end

data/app/models/publish_my_data/concept_scheme.rb

@@ -1,5 +1,21 @@
 module PublishMyData
   class ConceptScheme
     include Tripod::Resource
+    include PublishMyData::ResourceModule #some common methods for resources.
+
+    rdf_type RDF::SKOS.ConceptScheme
+    field :label, RDF::RDFS.label
+
+    def concepts
+      Resource.find_by_sparql(
+        "SELECT DISTINCT ?uri ?graph
+          WHERE {
+            GRAPH ?graph {
+              ?uri <#{RDF::SKOS.inScheme.to_s}> <#{self.uri}> .
+              ?uri a <#{RDF::SKOS.Concept.to_s}> .
+          }
+        }"
+      )
+    end
   end
 end

data/app/models/publish_my_data/dataset.rb

@@ -1,12 +1,38 @@
 module PublishMyData
   class Dataset
     include Tripod::Resource
+    include PublishMyData::ResourceModule #some common methods for resources.
 
+    # basics
     field :title, RDF::DC.title
-    field :description, RDF::DC.description
-    field :theme, PMD_DS.theme
+    field :comment, RDF::RDFS.comment #short desc
+    field :description, RDF::DC.description # long desc
 
-    rdf_type PMD_DS.Dataset
+    # licence, owner, contact
+    field :publisher, RDF::DC.publisher # value is a URI of a publisher
+    field :license, RDF::DC.license # value is URI of where licence is defined.
+    # NOTE: for contact, use :publisher's foaf:mbox value (in this metadata graph).
+
+    # quality, updates, maintenance
+    field :issued, RDF::DC.issued, :datatype => RDF::XSD.dateTime # value is DateTime literal
+    field :modified, RDF::DC.modified, :datatype => RDF::XSD.dateTime # value is DateTime literal
+    field :update_periodicity, RDF::DC.accrualPeriodicity # waiting for response on what the value should be.
+
+    # where to get it
+    field :data_dump, RDF::VOID.dataDump # full download URI
+
+     # what the data is about
+    field :theme, RDF::DCAT.theme
+    field :tags, RDF::DCAT.keyword, :multivalued => true # values are string literals
+
+    field :spatial_coverage, RDF::DC.spatial # value is a URI for region covered, e.g. England.
+    field :temporal_coverage, RDF::DC.temporal # value is a time interval URI
+    field :spatial_granularity, RDF::DC.spatial # value is class of the objects of refArea
+    field :temporal_granularity, RDF::DC.temporal # value is class of objects of refPeriod
+
+    field :size, RDF::VOID.triples # value is integer.
+
+    rdf_type RDF::PMD_DS.Dataset
 
     def slug
       Dataset.slug_from_uri(self.uri)
@@ -20,19 +46,25 @@ module PublishMyData
       slug
     end
 
-    class << self
+    def resources_in_dataset_criteria
+      Resource.all.graph(self.data_graph_uri)
+    end
 
-      def by_theme_criteria(theme)
-        Dataset.where("?uri <#{PMD_DS.theme}> '#{theme}'")
-      end
+    def resources_count
+      resources_in_dataset_criteria.count
+    end
 
-      def count_by_theme(theme)
-        by_theme_criteria(theme).count
-      end
+    def theme_obj
+      Theme.find(self.theme) rescue nil
+    end
 
-      def by_theme(theme)
-        by_theme_criteria(theme).resources
-      end
+    # use :publisher's foaf:mbox value (in this metadata graph).
+    def contact_email
+      publisher_obj = Resource.find(self.publisher) if publisher
+      publisher_obj.read_predicate(RDF::FOAF.mbox) if publisher_obj
+    end
+
+    class << self
 
       # this is the graph that dataset metadata goes in.
       def metadata_graph_uri(slug)
@@ -56,6 +88,9 @@ module PublishMyData
         uri.to_s.split('/').last
       end
 
+      def slug_from_data_graph_uri(data_graph_uri)
+        data_graph_uri.to_s.split("/").last
+      end
     end
   end
 end

data/app/models/publish_my_data/ontology.rb

@@ -1,5 +1,40 @@
 module PublishMyData
   class Ontology
     include Tripod::Resource
+    include PublishMyData::ResourceModule
+
+    rdf_type RDF::OWL.Ontology
+    field :label, RDF::RDFS.label
+
+    def ontology_classes
+      Resource.find_by_sparql("
+        SELECT DISTINCT ?uri ?graph
+        WHERE {
+          GRAPH ?graph {
+            {
+              ?uri <http://www.w3.org/2000/01/rdf-schema#isDefinedBy> <#{self.uri}> .
+              ?uri a <http://www.w3.org/2002/07/owl#Class> .
+            }
+            UNION
+            {
+              ?uri <http://www.w3.org/2000/01/rdf-schema#isDefinedBy> <#{self.uri}> .
+              ?uri a <http://www.w3.org/2000/01/rdf-schema#Class>
+            }
+          }
+        }"
+      )
+    end
+
+    def ontology_properties
+      Resource.find_by_sparql("
+        SELECT DISTINCT ?uri ?graph
+         WHERE {
+          GRAPH ?graph {
+            ?uri <http://www.w3.org/2000/01/rdf-schema#isDefinedBy> <#{self.uri}> .
+            ?uri a <http://www.w3.org/1999/02/22-rdf-syntax-ns#Property> .
+          }
+        }"
+      )
+    end
   end
 end

data/app/models/publish_my_data/ontology_class.rb

@@ -0,0 +1,11 @@
+module PublishMyData
+  class OntologyClass
+    include Tripod::Resource
+    include PublishMyData::ResourceModule
+    include PublishMyData::DefinedByOntology
+
+    field :label, RDF::RDFS.label
+    field :defined_by, RDF::RDFS.isDefinedBy
+
+  end
+end