patentscope 0.0.1

data/lib/patentscope/pct_doc_number.rb

@@ -0,0 +1,45 @@
+module Patentscope
+
+  class PctAppNumber < String
+
+    def initialize(number = "")
+      raise NoNumberError,
+        "Patent application number was not entered" if number.nil?
+      super(number.strip)
+    end
+
+    def valid?
+      self.match(/\A(([Pp][Cc][Tt]\/?)?[a-zA-Z]{2}\d{4}\/?\d{6})\Z/)
+    end
+
+    def validate
+      raise WrongNumberFormatError,
+        "PCT application number is not in correct format (PCT/CCYYYY/NNNNNN)" unless valid?
+    end
+
+    def to_ia_number
+      self.upcase!
+      self.gsub!('/', '')
+      self.gsub!('PCT', '')
+      self
+    end
+  end
+
+  class PctPubNumber < String
+
+    def initialize(number = "")
+      raise NoNumberError,
+        "Patent publication number was not entered" if number.nil?
+      super(number.strip)
+    end
+
+    def valid?
+      self.match(/\A(([Ww][Oo]\/?)?\s?\d{4}\/?\s?\d{6})\Z/)
+    end
+
+    def validate
+      raise WrongNumberFormatError,
+        "PCT publication number is not in correct format (WO/YYYY/NNNNNN)" unless valid?
+    end
+  end
+end

data/lib/patentscope/version.rb

@@ -0,0 +1,5 @@
+module Patentscope
+
+  VERSION = "0.0.1"
+
+end

data/lib/patentscope/webservice.rb

@@ -0,0 +1,74 @@
+module Patentscope
+
+  class Webservice
+
+    PATENTSCOPE_WEBSERVICE_LOCATION = "http://www.wipo.int/patentscope-webservice/servicesPatentScope"
+
+    def wsdl
+      send_wsdl_request
+    end
+
+    def get_available_documents(args = {})
+      ia_number = PctAppNumber.new(args[:ia_number]).to_ia_number
+      perform_operation(:getAvailableDocuments, { iaNumber: ia_number })
+    end
+
+    def get_document_content(args = {})
+      doc_id = args[:doc_id]
+      perform_operation(:getDocumentContent, { docId: doc_id })
+    end
+
+    def get_document_ocr_content(args = {})
+      doc_id = args[:doc_id]
+      perform_operation(:getDocumentOcrContent, { docId: doc_id })
+    end
+
+    def get_iasr(args = {})
+      ia_number = PctAppNumber.new(args[:ia_number]).to_ia_number
+      perform_operation(:getIASR, { iaNumber: ia_number })
+    end
+
+    def get_document_table_of_contents(args = {})
+      doc_id = args[:doc_id]
+      perform_operation(:getDocumentTableOfContents, { docId: doc_id })
+    end
+
+    def get_document_content_page(args = {})
+      doc_id  = args[:doc_id]
+      page_id = args[:page_id]
+      perform_operation(:getDocumentContentPage, { docId: doc_id, pageId: page_id })
+    end
+
+    private
+
+    def perform_operation(operation, options_hash)
+      soap_envelope = soapbuilder.build_envelope(operation, options_hash)
+      response = send_soap_request(soap_envelope)
+      if response.include?('This request requires HTTP authentication')
+        raise WrongCredentialsError
+      elsif response.include?('Business error during the execution of service')
+        raise BusinessError
+      else
+        soapbuilder.strip_envelope(response, operation)
+      end
+    end
+
+    def send_soap_request(soap_envelope_xml)
+      client.post_url(PATENTSCOPE_WEBSERVICE_LOCATION, "text/xml", soap_envelope_xml)
+    end
+
+    def send_wsdl_request
+      client.get_url(PATENTSCOPE_WEBSERVICE_LOCATION + '?wsdl')
+    end
+
+    def client
+      raise NoCredentialsError unless Patentscope.configured?
+      Client.new(username: Patentscope.configuration.username,
+                 password: Patentscope.configuration.password)
+    end
+
+    def soapbuilder
+      WebserviceSoapBuilder.new
+    end
+  end
+end

data/lib/patentscope/webservice_soap_builder.rb

@@ -0,0 +1,43 @@
+module Patentscope
+
+  class WebserviceSoapBuilder
+
+    def build_envelope(operation, options_hash)
+      builder = Nokogiri::XML::Builder.new do |xml|
+        xml.Envelope do
+          ns = xml.doc.root.add_namespace_definition('S', 'http://schemas.xmlsoap.org/soap/envelope/')
+          xml.doc.root.namespace = ns
+          xml.Body do
+            xml.send(operation, :'xmlns' => 'http://www.wipo.org/wsdl/ps') do
+              options_hash.each do |key, value|
+                xml.send(key, value)
+              end
+            end
+          end
+        end
+      end.to_xml
+    end
+
+    def strip_envelope(response, operation)
+      case operation
+      when :getAvailableDocuments
+        result_tag = 'doc'
+      when :getDocumentContent
+        result_tag = 'documentContent'
+      when :getDocumentOcrContent
+        result_tag = 'documentContent'
+      when :getIASR
+        result_tag = 'wo-international-application-status'
+      when :getDocumentTableOfContents
+        result_tag = 'content'
+      when :getDocumentContentPage
+        result_tag = 'pageContent'
+      end
+      doc = Nokogiri::XML(response)
+      stripped_response = doc.xpath("//iasr:#{result_tag}", 'iasr' => "http://www.wipo.org/wsdl/ps").to_xml
+      # this seems to be the only way to add back the XML declaration to the XML!
+      stripped_response_with_declaration = Nokogiri::XML(stripped_response).to_s
+      stripped_response_with_declaration
+    end
+  end
+end

data/patentscope.gemspec

@@ -0,0 +1,27 @@
+# -*- encoding: utf-8 -*-
+lib = File.expand_path('../lib', __FILE__)
+$LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
+require 'patentscope/version'
+
+Gem::Specification.new do |gem|
+  gem.name          = "patentscope"
+  gem.version       = Patentscope::VERSION
+  gem.authors       = ["Chong-Yee Khoo"]
+  gem.email         = ["mail@cykhoo.com"]
+  gem.description   = %q{Ruby interface to the Patentscope Webservice provided by the World Intellectual Property Organisation. Requires a subscription to the Patentscope Web Service}
+  gem.summary       = %q{Ruby interface with WIPO Patentscope Webservice}
+  gem.homepage      = "http://www.cantab-ip.com"
+
+  gem.files         = `git ls-files`.split($/)
+  gem.executables   = gem.files.grep(%r{^bin/}).map{ |f| File.basename(f) }
+  gem.test_files    = gem.files.grep(%r{^(test|spec|features)/})
+  gem.require_paths = ["lib"]
+
+  gem.add_development_dependency 'rake'
+  gem.add_development_dependency 'rspec', '~> 3.0.0.beta1'
+  gem.add_development_dependency 'vcr'
+  gem.add_development_dependency 'webmock'
+
+  gem.add_runtime_dependency 'nokogiri'
+  gem.add_runtime_dependency 'unicode_titlecase'
+end

data/spec/client_spec.rb

@@ -0,0 +1,75 @@
+require 'spec_helper'
+
+module Patentscope
+
+  describe Client, :core, :vcr do
+
+    before { Patentscope.configure_from_env}
+
+    let(:client) { Client.new(username: Patentscope.configuration.username, password: Patentscope.configuration.password) }
+
+    it "exists" do
+      expect(client).to_not be_nil
+    end
+
+    describe "methods" do
+
+      it "has its own attribute reader methods" do
+        expect(client).to respond_to(:username)
+        expect(client).to respond_to(:password)
+      end
+
+      it "has other methods" do
+        expect(client).to respond_to(:get_url)
+        expect(client).to respond_to(:post_url)
+      end
+    end
+
+    describe "attribute reader methods" do
+
+      describe "username method" do
+        it "returns the right username" do
+          expect(client.username).to eq(Patentscope.configuration.username)
+        end
+      end
+
+      describe "password method" do
+        it "returns the right password" do
+          expect(client.password).to eq(Patentscope.configuration.password)
+        end
+      end
+    end
+
+    describe "attribute writer methods" do
+
+      it "is not possible to set username" do
+        expect { client.username = 'foo' }.to raise_error()
+      end
+
+      it "is not possible to set password" do
+        expect { client.password = 'foo' }.to raise_error()
+      end
+    end
+
+    describe "get_url method" do
+      it "fetches the Cantab website by GET" do
+        site = client.get_url('http://www.cantab-ip.com')
+        expect(site).to include('<html')
+      end
+
+      it "fetches the Patentscope WSDL by GET" do
+        site = client.get_url('http://www.wipo.int/patentscope-webservice/servicesPatentScope?wsdl')
+        expect(site).to include('<?xml')
+      end
+    end
+
+    describe "post_url method" do
+      it "fetches a URL by POST" do
+        Patentscope.configure_from_env
+        soap_envelope_xml = '<?xml version="1.0"?><S:Envelope xmlns:S="http://schemas.xmlsoap.org/soap/envelope/"><S:Body><getIASR xmlns="http://www.wipo.org/wsdl/ps"><iaNumber>SG2009000062</iaNumber></getIASR></S:Body></S:Envelope>'
+        site = client.post_url('http://www.wipo.int/patentscope-webservice/servicesPatentScope', 'text/xml', soap_envelope_xml)
+        expect(site).to include('<?xml')
+      end
+    end
+  end
+end

data/spec/configuration_spec.rb

@@ -0,0 +1,138 @@
+require 'spec_helper'
+
+module Patentscope
+
+  describe Patentscope do
+
+    it 'has a configure class method' do
+      expect(Patentscope).to respond_to(:configure)
+    end
+
+    it 'has a configure_from_env class method' do
+      expect(Patentscope).to respond_to(:configure_from_env)
+    end
+
+    it 'has a configured? class method' do
+      expect(Patentscope).to respond_to(:configured?)
+    end
+
+    it 'has a reset_configuration class method' do
+      expect(Patentscope).to respond_to(:reset_configuration)
+    end
+
+    it 'has a configuration class method' do
+      expect(Patentscope).to respond_to(:configuration)
+    end
+  end
+
+  describe 'Patentscope::Configuration class' do
+
+    it 'has a username instance method' do
+      expect(Patentscope::Configuration.new).to respond_to(:username)
+    end
+
+    it 'has a password instance method' do
+      expect(Patentscope::Configuration.new).to respond_to(:password)
+    end
+  end
+
+  describe 'Patentscope before configuration' do
+
+    before { Patentscope.reset_configuration }
+
+    it 'configuration object is nil if configure not called' do
+      expect(Patentscope.configuration).to be_nil
+    end
+
+    it 'is not configured' do
+      expect(Patentscope.configured?).to be_falsey
+    end
+
+    it 'raises an error during client operations' do
+      expect do
+        Webservice.new.get_iasr(ia_number: 'SG2009000062')
+      end.to raise_error(Patentscope::NoCredentialsError)
+    end
+  end
+
+  describe 'Patentscope with configuration but no specific credentials given' do
+
+    before { Patentscope.reset_configuration }
+
+    it 'username defaults to null string if not explicitly set in the configure block' do
+      Patentscope.configure
+      expect(Patentscope.configuration.username).to eq('')
+    end
+
+    it 'password defaults to null string if not explicitly set in the configure block' do
+      Patentscope.configure
+      expect(Patentscope.configuration.password).to eq('')
+    end
+
+    it 'raises an error during client operations' do
+      expect do
+        Webservice.new.get_iasr(ia_number: 'SG2009000062')
+      end.to raise_error(Patentscope::NoCredentialsError)
+    end
+  end
+
+  describe 'Patentscope with configuration using configuration block' do
+
+    before do
+      Patentscope.configure do |config|
+        config.username     = 'joe_bloggs'
+        config.password     = 'b10ggsy'
+      end
+    end
+
+    after(:each) do
+      Patentscope.reset_configuration
+    end
+
+    it 'configures the username and password in a block in the configure method' do
+      expect(Patentscope.configuration.username).to eq('joe_bloggs')
+      expect(Patentscope.configuration.password).to eq('b10ggsy')
+    end
+  end
+
+  describe "Patentscope with configuration loading from environment variables" do
+
+    after(:each) do
+      Patentscope.reset_configuration
+    end
+
+    describe 'where a configuration has already been set' do
+      before do
+        Patentscope.configure do |config|
+          config.username     = 'joe_bloggs'
+          config.password     = 'b10ggsy'
+        end
+      end
+
+      it 'loading operation fails' do
+        expect(Patentscope.configure_from_env).to be false
+      end
+
+      it 'does not overwrite existing configuration' do
+        expect(Patentscope.configuration.username).to eq('joe_bloggs')
+        expect(Patentscope.configuration.password).to eq('b10ggsy')
+
+        expect(Patentscope.configuration.username).to_not eq(ENV['PATENTSCOPE_WEBSERVICE_USERNAME'])
+        expect(Patentscope.configuration.password).to_not eq(ENV['PATENTSCOPE_WEBSERVICE_PASSWORD'])
+      end
+    end
+
+    describe 'where there is no existing configuration' do
+
+      it 'loading operation succeeds' do
+        expect(Patentscope.configure_from_env).to be true
+      end
+
+      it 'assigns the configuration from environment variables' do
+        Patentscope.configure_from_env
+        expect(Patentscope.configuration.username).to eq(ENV['PATENTSCOPE_WEBSERVICE_USERNAME'])
+        expect(Patentscope.configuration.password).to eq(ENV['PATENTSCOPE_WEBSERVICE_PASSWORD'])
+      end
+    end
+  end
+end

data/spec/patentscope_spec.rb

@@ -0,0 +1,91 @@
+require 'spec_helper'
+
+module Patentscope
+
+  describe Patentscope, :core, :vcr do
+
+    before { Patentscope.configure_from_env }
+
+    let(:patentscope) { Patentscope }
+
+    it "exists" do
+      expect(patentscope).to_not be_nil
+    end
+
+    it "has the right methods" do
+      expect(patentscope).to respond_to(:wsdl)
+      expect(patentscope).to respond_to(:get_available_documents)
+      expect(patentscope).to respond_to(:get_document_content)
+      expect(patentscope).to respond_to(:get_document_ocr_content)
+      expect(patentscope).to respond_to(:get_iasr)
+      expect(patentscope).to respond_to(:get_document_table_of_contents)
+      expect(patentscope).to respond_to(:get_document_content_page)
+    end
+
+    describe "wsdl method" do
+      it "returns a wsdl document" do
+        response = patentscope.wsdl
+        expect(response).to include('<?xml')
+        expect(response).to include('<wsdl:definitions')
+      end
+    end
+
+    describe "get_available_documents method" do
+      it 'returns an appropriate XML document for the get_available_documents operation' do
+        response = patentscope.get_available_documents('SG2009000062')
+        expect(response).to include('<?xml version="1.0"?>')
+        expect(response).to include('<doc ocrPresence="no" docType="RO101" docId="id00000008679651"/>')
+        expect(response).to_not include('<getAvailableDocumentsResponse xmlns="http://www.wipo.org/wsdl/ps">')
+      end
+    end
+
+    describe "get_document_content method" do
+      it 'returns an appropriate XML document for the get_document_content operation' do
+        response = patentscope.get_document_content('090063618004ca88')
+        expect(response).to include('<?xml version="1.0"?>')
+        expect(response).to include('<documentContent>')
+        expect(response).to include('nRpZsy7ezxU2/8/fk5JM6HIXReMWymXUCmhYcRgUIjjNk2pDAkdlxox7xiSLm')
+        expect(response).to_not include('<getDocumentContentResponse xmlns="http://www.wipo.org/wsdl/ps">')
+      end
+    end
+
+    describe "get_document_ocr_content method" do
+      it 'returns an appropriate XML document for the get_document_ocr_content operation' do
+        response = patentscope.get_document_ocr_content('id00000015801579')
+        expect(response).to include('<?xml version="1.0"?>')
+        expect(response).to include('<documentContent>')
+        expect(response).to include('XdDb9Ain4kev61wgZc36X022QPCEZZASS2Rwpcy4Hx7I5GYHhriRwpsDwoX9tgjgZwcEGGEksgthsHsNtkFmyGZYQIGGCCX3dhggRDTgEEDNgVgkvuw2ECDDSYMEF')
+        expect(response).to_not include('<getDocumentOcrContentResponse xmlns="http://www.wipo.org/wsdl/ps">')
+      end
+    end
+
+    describe "get_iasr method" do
+      it 'returns an appropriate XML document for the get_iasr operation' do
+        response = patentscope.get_iasr('SG2009000062')
+        expect(response).to include('<?xml version="1.0"?>')
+        expect(response).to include('<wo-international-application-status>')
+        expect(response).to include('MESENCHYMAL STEM CELL PARTICLES')
+        expect(response).to_not include('<getIASRResponse xmlns="http://www.wipo.org/wsdl/ps">')
+      end
+    end
+
+    describe "get_document_table_of_contents method" do
+      it 'returns an appropriate XML document for the get_document_table_of_contents operation' do
+        response = patentscope.get_document_table_of_contents('090063618004ca88')
+        expect(response).to include('<?xml version="1.0"?>')
+        expect(response).to include('<content>')
+        expect(response).to include('<content>000001.tif</content>')
+        expect(response).to_not include('<getDocumentTableOfContentsResponse xmlns="http://www.wipo.org/wsdl/ps">')
+      end
+    end
+
+    describe "get_document_content_page method" do
+      it 'returns an appropriate XML document for the get_document_content_page operation' do
+        response = patentscope.get_document_content_page('090063618004ca88', '000001.tif')
+        expect(response).to include('<?xml version="1.0"?>')
+        expect(response).to include('+GP0kv9dhgiY7Rb5h2q4RN6Jj9NpDCJjuMImO0l0TfLe7QRO2yFceTvvTu6C6qTH')
+        expect(response).to_not include('<getDocumentContentPageResponse xmlns="http://www.wipo.org/wsdl/ps">')
+      end
+    end
+  end
+end