From: Michael Shigorin <mike@...>
Date: 2010-03-21T07:16:09+09:00
Subject: ops.epo.org hacks (was: Wsdl -  Webservice Client)

--qMm9M+Fa2AknHoGS
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline

On Wed, Mar 17, 2010 at 12:31:57AM +0900, Axel Fuchs wrote:
> Brian, thank you very much for your help. This is really
> unnecessarily complex.  The only reason I am interested in SOAP
> is so that I can access the patent data information from the
> European Patent Office (see my other post).

I've posted some code on their forum:
http://forums.epo.org/open-patent-services-and-publication-server-web-service/topic898.html

The attached copies are slightly more updated.

The code doesn't use anything but handmade XML processing
since soap4r couldn't cope with WSDL2 or whatever new and
improved spec version EPO has migrated to (espacenet was
fine), and time was pretty short to wrestle with wso2.org
first.  I did run it last December, still worked back then.

-- 
 ---- WBR, Michael Shigorin <mike@altlinux.ru>
  ------ Linux.Kiev http://www.linux.kiev.ua/

--qMm9M+Fa2AknHoGS
Content-Type: text/plain; charset=us-ascii
Content-Disposition: attachment; filename="ops.rb"

#!/usr/bin/ruby
require 'net/http'
require 'rexml/document'

module OPS

	class SoapReq < REXML::Element
		DEBUG = false
		#DEBUG = true
		attr_reader :xml	# FIXME: should be unneeded
		@@xpath_prefix = "SOAP-ENV:Body/ops:world-patent-data"
		def initialize(action, docnumber, kind = "%%")
			@query = <<-END_QUERY
			<soapenv:Envelope
				 xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/"
				 xmlns:ops="http://ops.epo.org"
				 xmlns:exc="http://www.epo.org/exchange">
				 <soapenv:Header/>
				 <soapenv:Body>
						<ops:#{action} full-publication-cycle="true">
							 <exc:publication-reference data-format="epodoc">
									<exc:document-id>
										 <exc:country></exc:country>
										 <exc:doc-number>#{docnumber}</exc:doc-number>
										 <exc:kind>#{kind}</exc:kind>
									</exc:document-id>
							 </exc:publication-reference>
						</ops:#{action}>
				 </soapenv:Body>
			</soapenv:Envelope>
			END_QUERY
			@path = "/soap-services/#{action}"
			@headers = {
				"SOAPAction" => action,
				"Content-Type" => "text/xml; charset=utf-8"
			}
			puts "[ops.rb] SOAP request:", @query if DEBUG
			begin
				@http = Net::HTTP.new("ops.epo.org")
				@response, @data = @http.post(@path, @query, @headers)
				puts "[ops.rb] HTTP response: #{@response}" if DEBUG
			rescue => net_error
				puts "** [ops.rb] trouble while querying OPS: #{net_error}"
			end
			@xml = REXML::Document.new(@data).root
		end
	end

	class Biblio < SoapReq
		attr_reader :kind
		@@xpath1 = "#{@@xpath_prefix}/exchange-documents/exchange-document"
		@@xpath2 = "#{@@xpath_prefix}/exchange-documents/exchange-document/bibliographic-data"
		def initialize(docnumber)
			# that req handler does accept <kind> wildcards
			super("biblio-retrieval", docnumber, "%%")
			@kind = @xml.elements[@@xpath1].attributes["kind"]
			puts "[ops.rb] Biblio: #{@xml.elements[@@xpath2].inspect}" if DEBUG
			@xml = @xml.elements[@@xpath2]
		end
	end

	class Equiv < SoapReq
		@@xpath = "#{@@xpath_prefix}/ops:equivalents-inquiry"
		def initialize(docnumber, kind)
			# that req handler does NOT accept <kind> wildcards
			super("equivalents-inquiry", docnumber, kind)
			puts "[ops.rb] Equivalents: #{@xml.elements[@@xpath].inspect}" if DEBUG
			@xml = @xml.elements[@@xpath]
		end
	end

	class Family < SoapReq
		@@xpath = "#{@@xpath_prefix}/ops:patent-family"
		def initialize(docnumber)
			# that req handler only accepts empty <kind>
			super("family-retrieval", docnumber, "")
			puts "[ops.rb] Family: #{@xml.elements[@@xpath].inspect}" if DEBUG
			@xml = @xml.elements[@@xpath]
		end
	end

end

--qMm9M+Fa2AknHoGS
Content-Type: text/plain; charset=us-ascii
Content-Disposition: attachment; filename="patent.rb"

#!/usr/bin/env ruby
require 'rexml/document'
require 'ops.rb'

module OPS

class Reference
	attr_reader :docnumber	# :date is special
	def initialize(docnumber, date)
		@docnumber = docnumber
		@date = date
	end
	def date
		"#{@date[0..3]}-#{@date[4..5]}-#{@date[6..7]}"
	end
	def to_s
		"#{@docnumber} #{@date}"
	end
	alias inspect to_s
end

class PublicationReference < Reference
	attr_reader :country, :number, :kind
	@@xpath = "publication-reference[@data-format='epodoc']/document-id"
	def initialize(xml)
    e = xml.elements[@@xpath]
		@docnumber = e.elements["doc-number"].text
		@country = @docnumber[0..1]
		@number = @docnumber[2..-1]
		@date = e.elements["date"].text
		# FIXME: <kind> was _needed_ for table B.6.1, revisit this
		@kind = xml.elements["publication-reference[@data-format='docdb']/document-id/kind"].text
		self
	end
	def to_s
		"#{@docnumber} (#{kind}) -- #{self.date}"
	end
	alias inspect to_s
end

class ApplicationReference < Reference
	@@xpath = "application-reference[@data-format='epodoc']/document-id"
	def initialize(xml)
		e = xml.elements[@@xpath]
		@docnumber = e.elements["doc-number"].text
		@date = e.elements["date"].text
		self
	end
end

class PriorityClaim < Reference; end
class PriorityClaims < Array
	@@xpath = "priority-claims/priority-claim[@data-format='epodoc']/document-id"
	def initialize(xml)
		xml.elements.each(@@xpath) { |e|
			docnumber = e.elements["doc-number"].text
			date = e.elements["date"].text
			self.push PriorityClaim.new(docnumber, date)
		}
		self
	end
end

class ClassificationIPCR < String
	def initialize(text)
		self << text.delete(" ").chop.chop
	end
end

class Classifications < Array
	@@xpath = "classifications-ipcr/classification-ipcr/text"
	def initialize(xml)
		xml.elements.each(@@xpath) { |e| self.push ClassificationIPCR.new(e.text) }
		self
	end
end

class Applicants < Array
	@@xpath = "parties/applicants/applicant[@data-format='epodoc']/applicant-name/name"
	def initialize(xml)
		xml.elements.each(@@xpath) { |e| self.push e.text }
		self
	end
end

class Title	< String # lang="en"
	def initialize(xml)
		# FIXME: what if several translations exist? [probably w/o en]
		begin
			self << xml.elements["invention-title[@lang='en']"].text
		rescue NoMethodError
			self << xml.elements["invention-title"].text
		end
	end
end

class Equivalent < String		# not < Reference since only docnumber
	def initialize(xml)
		self << xml.elements["doc-number"].text
	end
end

class Equivalents < Array
	@@xpath = "ops:inquiry-result/publication-reference[@data-format='epodoc']/document-id"
	def initialize(xml)
		begin
			xml.elements.each(@@xpath) { |e|
				self.push Equivalent.new(e)
			}
		rescue NoMethodError
		end
		self
	end
end

class Patent
	attr_reader :docnumber, :pubref, :appref, :classes
	attr_reader :prioclaims, :applicants, :title, :kind
	attr_reader :equivs
	def initialize(docnumber)
		@docnumber = docnumber
		# NB: after replacing biblio with biblio+xml,
		# the Patent lost the ability to be inspected (?!?)
		@biblio = OPS::Biblio.new(@docnumber)
		@kind = @biblio.kind	# it's an attribute, not sub-element
		# FIXME: currently we don't need equivalents, maybe separate them?
		#@equivs = Equivalents.new(OPS::Equiv.new(@docnumber, @kind).xml)
		@equivs = []
		@xml = @biblio.xml
		@pubref = PublicationReference.new(@xml)
    @appref = ApplicationReference.new(@xml)
    @classes = Classifications.new(@xml)
    @prioclaims = PriorityClaims.new(@xml)
    @applicants = Applicants.new(@xml)
    @title = Title.new(@xml)
	end
	def inspect
		"No. #{@docnumber} (#{@kind}) \"#{@title}\" #{@applicants}; Appl. No #{@appref}; Pub. No #{@pubref}; Pri. No #{@prioclaims}; Cls. #{@classes}; Eq. #{@equivs}"
	end
end

end

--qMm9M+Fa2AknHoGS
Content-Type: text/plain; charset=us-ascii
Content-Disposition: attachment; filename="family.rb"

#!/usr/bin/env ruby
require 'rexml/document'
require 'patent.rb'
require 'ops.rb'

module OPS

class FamilyPublications < Array
  @@xpath = "ops:family-member"
  def initialize(xml)
    xml.elements.each(@@xpath) { |e|
      self.push PublicationReference.new(e)
    }
    self
  end
end

class PatentFamily
	attr_reader :docnumber, :fampubs
	def initialize(docnumber)
		@docnumber = docnumber
		@fampubs = FamilyPublications.new(OPS::Family.new(@docnumber).xml)
	end
	def inspect
		"No. #{@docnumber}'s family: #{@fampubs}"
	end
end

end

--qMm9M+Fa2AknHoGS--