puree 0.20.0 → 1.0.0

data/lib/puree/api/request.rb

@@ -0,0 +1,116 @@
+module Puree
+
+  module API
+
+    # Handles requests to Pure.
+    #
+    class Request
+
+      def initialize(url:)
+        @url = url
+        @api_map = Puree::API::Map.new
+        @headers = {}
+      end
+
+      # Provide credentials if necessary
+      #
+      # @param username [String]
+      # @param password [String]
+      def basic_auth(username:, password:)
+        auth = Base64::strict_encode64("#{username}:#{password}")
+        @headers['Authorization'] = 'Basic ' + auth
+      end
+
+      # Perform a GET request to Pure
+      #
+      # @param uuid [String]
+      # @param id [String]
+      # @param rendering [String]
+      # @param latest_api [Boolean]
+      # @param resource_type [String]
+      # @param limit [Fixnum]
+      # @param offset [Fixnum]
+      # @param created_start [String]
+      # @param created_end [String]
+      # @param modified_start [String]
+      # @param modified_end [String]
+      # @param content_type [String]
+      # @return [HTTP::Response]
+      def get(uuid: nil,
+              id: nil,
+              rendering: :xml_long,
+              latest_api: true,
+              resource_type:,
+              limit: 20,
+              offset: 0,
+              created_start: nil,
+              created_end: nil,
+              modified_start: nil,
+              modified_end: nil,
+              content_type: nil)
+            @latest_api =    latest_api
+            @resource_type =  resource_type.to_sym
+            @rendering =      rendering
+            @uuid =           uuid
+            @id =             id
+            @limit =          limit
+            @offset =         offset
+            @created_start =  created_start
+            @created_end =    created_end
+            @modified_start = modified_start
+            @modified_end =   modified_end
+            @content_type =   content_type
+
+        # strip any trailing slash
+        @url = @url.sub(/(\/)+$/, '')
+        @headers['Accept'] = 'application/xml'
+        @req = HTTP.headers accept: @headers['Accept']
+        if @headers['Authorization']
+          @req = @req.auth @headers['Authorization']
+        end
+        @req.get(build_url, params: params)
+      end
+
+      private
+
+      def params
+        query = {}
+        if @uuid
+          query['uuids.uuid'] = @uuid
+        else
+          if @id
+            query['pureInternalIds.id'] = @id
+          end
+        end
+        query['rendering'] = @rendering
+        if @resource_type != :server
+          query['window.size'] = @limit
+          query['window.offset'] = @offset if @limit > 0
+        end
+
+        # Pure does allow blank value
+        query['contentType'] = @content_type if @content_type
+
+        # Pure does not allow blanks for these
+        query['createdDate.toDate'] = @created_start if @created_start
+        query['createdDate.fromDate'] = @created_end if @created_end
+        query['modifiedDate.toDate'] = @modified_start if @modified_start
+        query['modifiedDate.fromDate'] = @modified_end if @modified_end
+        query
+      end
+
+      def build_url
+        service = @api_map.service_name(@resource_type)
+        if @latest_api === false
+          service_api_mode = service
+        else
+          service_api_mode = service + '.current'
+        end
+        @url + '/' + service_api_mode
+      end
+
+    end
+
+  end
+
+end

data/lib/puree/extractor/collection.rb

@@ -0,0 +1,131 @@
+module Puree
+
+  module Extractor
+
+    # A collection extractor can retrieve any number of resources of the same type.
+    #
+    class Collection
+      include Puree::API::Authentication
+
+      # @option (see Puree::API::Authentication#configure_api)
+      # @param resource [Symbol]
+      def initialize(config:, resource:)
+        @resource_type = resource
+        @api_map = Puree::API::Map.new.get
+        configure_api config
+      end
+
+      # Gets an array of objects of resource type specified in constructor.
+      #
+      # @param limit [Fixnum]
+      # @param offset [Fixnum]
+      # @param created_start [String]
+      # @param created_end [String]
+      # @param modified_start [String]
+      # @param modified_end [String]
+      # @return [Array<Puree::Model::Resource subclass>] Resource metadata e.g. Puree::Model::Dataset.
+      def get(
+              limit:            0,
+              offset:           0,
+              created_start:    nil,
+              created_end:      nil,
+              modified_start:   nil,
+              modified_end:     nil
+      )
+        @response = @request.get rendering:        :system,
+                                 limit:            limit,
+                                 offset:           offset,
+                                 resource_type:    @resource_type,
+                                 created_start:    created_start,
+                                 created_end:      created_end,
+                                 modified_start:   modified_start,
+                                 modified_end:     modified_end
+        set_content @response.body
+      end
+
+      # Gets a random resource of type specified in constructor.
+      #
+      # @return [Puree::Model::Resource subclass, nil] Resource metadata e.g. Puree::Model::Dataset.
+      def random_resource
+        @response = @request.get rendering:        :system,
+                                 limit:            1,
+                                 offset:           rand(0..count-1),
+                                 resource_type:    @resource_type
+        content = set_content @response.body
+        content[0] if content
+      end
+
+
+      # Count of records available for a resource type.
+      #
+      # @return [Fixnum]
+      def count
+        get_count
+      end
+
+      private
+
+      # Array of UUIDs (from system response).
+      #
+      # @return [Array<String>]
+      def uuids
+        @extractor.uuids
+      end
+
+      def combine_metadata
+        collect_resources
+      end
+
+      def get_count
+        @response = @request.get resource_type: @resource_type,
+                                 rendering: :system,
+                                 limit: 0
+        make_xml_extractor
+        @extractor.count
+      end
+
+      def collect_resources
+        data = []
+        resource_class = "Puree::Extractor::#{@resource_type.to_s.capitalize}"
+
+        # whitelist symbol
+        if @api_map[:resource_type].has_key?(@resource_type)
+          config = {
+            url: @config.url,
+            username: @config.username,
+            password: @config.password
+          }
+
+          uuids.each do |u|
+            r = Object.const_get(resource_class).new config
+            record = r.find uuid: u
+            data << record
+          end
+          data
+        else
+          raise 'Invalid resource class'
+        end
+      end
+
+      # Set content from XML. In order for metadata extraction to work, the XML must have
+      # been retrieved using the .current version of the Pure API endpoints
+      #
+      # @param xml [String]
+      def set_content(xml)
+        if xml
+          make_xml_extractor
+          @extractor.get_data? ? combine_metadata : nil
+        end
+      end
+
+      def make_xml_extractor
+        @extractor = Puree::XMLExtractor::Collection.new xml: @response.body
+      end
+
+      alias :find :get
+
+    end
+
+  end
+
+end

data/lib/puree/extractor/dataset.rb

@@ -0,0 +1,48 @@
+module Puree
+
+  module Extractor
+
+    # Dataset extractor.
+    #
+    class Dataset < Puree::Extractor::Resource
+
+      # @option (see Puree::Extractor::Resource#initialize)
+      def initialize(config)
+        super
+        setup :dataset
+      end
+
+      private
+
+      def combine_metadata
+        super
+        @model.access = @extractor.access
+        @model.associated = @extractor.associated
+        @model.available = @extractor.available
+        @model.description = @extractor.description
+        @model.doi = @extractor.doi
+        @model.files = @extractor.files
+        @model.keywords = @extractor.keywords
+        @model.links = @extractor.links
+        @model.legal_conditions = @extractor.legal_conditions
+        @model.organisations = @extractor.organisations
+        @model.owner = @extractor.owner
+        @model.persons_internal = @extractor.persons_internal
+        @model.persons_external = @extractor.persons_external
+        @model.persons_other = @extractor.persons_other
+        @model.projects = @extractor.projects
+        @model.production = @extractor.production
+        @model.publications = @extractor.publications
+        @model.publisher = @extractor.publisher
+        @model.spatial_places = @extractor.spatial_places
+        @model.spatial_point = @extractor.spatial_point
+        @model.temporal = @extractor.temporal
+        @model.title = @extractor.title
+        @model
+      end
+
+    end
+
+  end
+
+end

data/lib/puree/extractor/download.rb

@@ -0,0 +1,71 @@
+module Puree
+
+  module Extractor
+
+    # Download extractor.
+    #
+    class Download
+      include Puree::API::Authentication
+
+      # @option (see Puree::API::Authentication#configure_api)
+      def initialize(config)
+        @resource_type = :download
+        @api_map = Puree::API::Map.new.get # Workararound to provide access to service_family
+        configure_api config
+      end
+
+      # Get download statistics. Only for Datasets.
+      #
+      # @param limit [Fixnum]
+      # @param offset [Fixnum]
+      # @param resource [Symbol] The resource being reported
+      # @return [Array<Puree::Model::DownloadHeader>]
+      def get(limit: 0,
+              offset: 0,
+              resource:)
+        raise 'Cannot perform a request without a configuration' if @config.nil?
+        @response = @request.get rendering:      :system,
+                                 limit:          limit,
+                                 offset:         offset,
+                                 resource_type:  @resource_type,
+                                 content_type:   service_family(resource)
+        set_content @response.body
+      end
+
+      private
+
+      def combine_metadata
+        @extractor.statistics
+      end
+
+      # Set content from XML. In order for metadata extraction to work, the XML must have
+      # been retrieved using the .current version of the Pure API endpoints
+      #
+      # @param xml [String]
+      def set_content(xml)
+        if xml
+          make_extractor
+          @extractor.get_data? ? combine_metadata : []
+        end
+      end
+
+      def make_extractor
+        @extractor = Puree::XMLExtractor::Download.new xml: @response.body
+      end
+
+      def service_family(resource_type)
+        if @api_map[:resource_type].has_key? resource_type
+          # Family data is only populated for datasets (required for download)
+          @api_map[:resource_type][resource_type][:family]
+        else
+          raise "#{resource_type} is an unrecognised resource type"
+        end
+      end
+
+      alias :find :get
+
+    end
+
+  end
+
+end

data/lib/puree/extractor/event.rb

@@ -0,0 +1,33 @@
+module Puree
+
+  module Extractor
+
+    # Event extractor.
+    #
+    class Event < Puree::Extractor::Resource
+
+      # @option (see Puree::Extractor::Resource#initialize)
+      def initialize(config)
+        super
+        setup :event
+      end      
+ 
+      private
+
+      def combine_metadata
+        super
+        @model.city = @extractor.city
+        @model.country = @extractor.country
+        @model.date = @extractor.date
+        @model.description = @extractor.description
+        @model.location = @extractor.location
+        @model.title = @extractor.title
+        @model.type = @extractor.type
+        @model
+      end
+
+    end
+
+  end
+
+end

data/lib/puree/extractor/extractor.rb

@@ -0,0 +1,10 @@
+module Puree
+
+  # An Extractor manages HTTP requests to Pure and the subsequent
+  # extraction of metadata from an XML response into a Ruby data model.
+  #
+  module Extractor
+
+  end
+
+end

data/lib/puree/extractor/journal.rb

@@ -0,0 +1,29 @@
+module Puree
+
+  module Extractor
+
+    # Journal extractor.
+    #
+    class Journal < Puree::Extractor::Resource
+
+      # @option (see Puree::Extractor::Resource#initialize)
+      def initialize(config)
+        super
+        setup :journal
+      end
+
+      private
+
+      def combine_metadata
+        super
+        @model.issn = @extractor.issn
+        @model.publisher = @extractor.publisher
+        @model.title = @extractor.title
+        @model
+      end
+
+    end
+
+  end
+
+end