WIP - to be unpacked/revised later

sul-dlss · Oct 3, 2017 · d8876a1 · d8876a1
1 parent 2a91993
commit d8876a1
Show file tree

Hide file tree

Showing 7 changed files with 117 additions and 90 deletions.
diff --git a/lib/wos_client.rb b/lib/wos_client.rb
@@ -20,7 +20,7 @@ def auth
       wsdl: AUTH_WSDL,
       headers: { 'Authorization' => "Basic #{@auth_code}", 'SOAPAction' => [''] },
       env_namespace: :soapenv,
-      log: true,
+      log: false,
       log_level: @log_level,
       pretty_print_xml: true
     )
@@ -39,7 +39,7 @@ def search
       wsdl: SEARCH_WSDL,
       headers: { 'Cookie' => "SID=\"#{session_id}\"", 'SOAPAction' => '' },
       env_namespace: :soapenv,
-      log: true,
+      log: false,
       log_level: @log_level,
       pretty_print_xml: true
     )

diff --git a/lib/wos_queries.rb b/lib/wos_queries.rb
@@ -106,9 +106,10 @@ def retrieve_additional_records(response, response_type, retrieve_operation = :r
         iterations = record_total / MAX_RECORDS
         iterations -= 1 if (record_total % MAX_RECORDS).zero?
         [*1..iterations].each do |i|
+          first_record = (MAX_RECORDS * i) + 1
           message = {
             queryId: query_id,
-            retrieveParameters: retrieve_parameters(first_record: MAX_RECORDS * i)
+            retrieveParameters: retrieve_parameters(first_record: first_record)
           }
           response_i = wos_client.search.call(retrieve_operation, message: message)
           records_i = records(response_i, "#{retrieve_operation}_response".to_sym)

diff --git a/lib/wos_records.rb b/lib/wos_records.rb
@@ -3,28 +3,25 @@
 
 # Utilities for working with Web of Science records
 class WosRecords
-  include Enumerable
   extend Forwardable
+  include Enumerable
 
+  # @return [Integer] WOS record count
   def_delegators :rec_nodes, :count
 
   # @return xml [String] WOS records in XML
   def_delegators :doc, :to_xml
 
+  # @return doc [Nokogiri::XML::Document] WOS records document
+  attr_reader :doc
+
   # @param records [String] records in XML
   # @param encoded_records [String] records in HTML encoding
   def initialize(records: nil, encoded_records: nil)
     @records = records
     @encoded_records = encoded_records
     @records = decode_records if records.nil? && !encoded_records.nil?
-  end
-
-  # # @return [Integer]
-  # alias count rec_nodes.count
-
-  # @return doc [Nokogiri::XML::Document] WOS records document
-  def doc
-    @doc ||= Nokogiri::XML(records) { |config| config.strict.noblanks }
+    @doc = Nokogiri::XML(@records) { |config| config.strict.noblanks }
   end
 
   def each
@@ -56,10 +53,16 @@ def duplicate_records(record_setB)
   # @param record_setB [WosRecords]
   # @return records [WosRecords] merged set of records
   def merge_records(record_setB)
-    rec_nodesB = record_setB.rec_nodes - duplicate_records(record_setB)
     # create a new set of merged records, use a philosophy of immutability
+
+    # binding.pry
+    # TODO: why is this changing the record_setB????
+
+    dup_uids = duplicate_uids(record_setB)
+    merge_nodes = record_setB.rec_nodes.dup
+    merge_nodes.reject { |rec| dup_uids.include? record_uid(rec) } unless dup_uids.empty?
     merge_doc = doc.dup
-    merge_doc.at('records').add_child(rec_nodesB)
+    merge_doc.at('records').add_child(merge_nodes)
     WosRecords.new(records: merge_doc.to_xml)
   end
 

diff --git a/...xtures/wos_client/wos_encoded_records.txt → ...tures/wos_client/wos_encoded_records.html b/...xtures/wos_client/wos_encoded_records.txt → ...tures/wos_client/wos_encoded_records.html
diff --git a/spec/fixtures/wos_client/wos_merge_records.html b/spec/fixtures/wos_client/wos_merge_records.html
@@ -0,0 +1,4 @@
+&lt;records xmlns="http://scientific.thomsonreuters.com/schema/wok5.4/public/FullRecord"&gt;
+&lt;REC r_id_disclaimer="ResearcherID data provided by Clarivate Analytics"&gt;&lt;UID&gt;WOS:A1972N549400003&lt;/UID&gt;&lt;static_data&gt;&lt;summary&gt;&lt;EWUID&gt;&lt;WUID coll_id="WOS"&gt;&lt;/WUID&gt;&lt;edition value="WOS.SSCI"&gt;&lt;/edition&gt;&lt;edition value="WOS.SCI"&gt;&lt;/edition&gt;&lt;/EWUID&gt;&lt;pub_info issue="5" pubtype="Journal" sortdate="1972-01-01" has_abstract="N" coverdate="1972" vol="33" pubyear="1972"&gt;&lt;page end="413" page_count="1" begin="413"&gt;413-413&lt;/page&gt;&lt;/pub_info&gt;&lt;titles count="6"&gt;&lt;title type="source"&gt;COLLEGE &amp;amp; RESEARCH LIBRARIES&lt;/title&gt;&lt;title type="source_abbrev"&gt;COLL RES LIBR&lt;/title&gt;&lt;title type="abbrev_iso"&gt;Coll. Res. Libr.&lt;/title&gt;&lt;title type="abbrev_11"&gt;COLL RES LI&lt;/title&gt;&lt;title type="abbrev_29"&gt;COLL RES LIBR&lt;/title&gt;&lt;title type="item"&gt;LIBRARY MANAGEMENT - BEHAVIOR-BASED PERSONNEL SYSTEMS (BBPS) - FRAMEWORK FOR ANALYSIS - KEMPER,RE&lt;/title&gt;&lt;/titles&gt;&lt;names count="1"&gt;&lt;name daisng_id="19669717" seq_no="1" role="author"&gt;&lt;display_name&gt;WEBER, DC&lt;/display_name&gt;&lt;full_name&gt;WEBER, DC&lt;/full_name&gt;&lt;wos_standard&gt;WEBER, DC&lt;/wos_standard&gt;&lt;first_name&gt;DC&lt;/first_name&gt;&lt;last_name&gt;WEBER&lt;/last_name&gt;&lt;/name&gt;&lt;/names&gt;&lt;doctypes count="1"&gt;&lt;doctype&gt;Book Review&lt;/doctype&gt;&lt;/doctypes&gt;&lt;publishers&gt;&lt;publisher&gt;&lt;address_spec addr_no="1"&gt;&lt;full_address&gt;50 E HURON ST, CHICAGO, IL 60611&lt;/full_address&gt;&lt;city&gt;CHICAGO&lt;/city&gt;&lt;/address_spec&gt;&lt;names count="1"&gt;&lt;name addr_no="1" role="publisher" seq_no="1"&gt;&lt;display_name&gt;ASSOC COLL RESEARCH LIBRARIES&lt;/display_name&gt;&lt;full_name&gt;ASSOC COLL RESEARCH LIBRARIES&lt;/full_name&gt;&lt;/name&gt;&lt;/names&gt;&lt;/publisher&gt;&lt;/publishers&gt;&lt;/summary&gt;&lt;fullrecord_metadata&gt;&lt;languages count="1"&gt;&lt;language type="primary"&gt;English&lt;/language&gt;&lt;/languages&gt;&lt;normalized_languages count="1"&gt;&lt;language type="primary"&gt;English&lt;/language&gt;&lt;/normalized_languages&gt;&lt;normalized_doctypes count="1"&gt;&lt;doctype&gt;Review&lt;/doctype&gt;&lt;/normalized_doctypes&gt;&lt;refs count="1"&gt;&lt;/refs&gt;&lt;addresses count="1"&gt;&lt;address_name&gt;&lt;address_spec addr_no="1"&gt;&lt;full_address&gt;STANFORD UNIV,STANFORD,CA 94305&lt;/full_address&gt;&lt;organizations count="2"&gt;&lt;organization&gt;STANFORD UNIV&lt;/organization&gt;&lt;organization pref="Y"&gt;Stanford University&lt;/organization&gt;&lt;/organizations&gt;&lt;city&gt;STANFORD&lt;/city&gt;&lt;state&gt;CA&lt;/state&gt;&lt;country&gt;USA&lt;/country&gt;&lt;zip location="AP"&gt;94305&lt;/zip&gt;&lt;/address_spec&gt;&lt;/address_name&gt;&lt;/addresses&gt;&lt;category_info&gt;&lt;headings count="1"&gt;&lt;heading&gt;Science &amp;amp; Technology&lt;/heading&gt;&lt;/headings&gt;&lt;subheadings count="1"&gt;&lt;subheading&gt;Technology&lt;/subheading&gt;&lt;/subheadings&gt;&lt;subjects count="3"&gt;&lt;subject ascatype="traditional" code="NU"&gt;Information Science &amp;amp; Library Science&lt;/subject&gt;&lt;subject ascatype="extended"&gt;Information Science &amp;amp; Library Science&lt;/subject&gt;&lt;subject ascatype="traditional" code="NU"&gt;INFORMATION SCIENCE &amp;amp; LIBRARY SCIENCE&lt;/subject&gt;&lt;/subjects&gt;&lt;/category_info&gt;&lt;/fullrecord_metadata&gt;&lt;item xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:type="itemType_wos"&gt;&lt;ids avail="Y"&gt;N5494&lt;/ids&gt;&lt;bib_id&gt;33 (5): 413-413 1972&lt;/bib_id&gt;&lt;/item&gt;&lt;/static_data&gt;&lt;dynamic_data&gt;&lt;citation_related&gt;&lt;tc_list&gt;&lt;silo_tc coll_id="WOS" local_count="0"&gt;&lt;/silo_tc&gt;&lt;/tc_list&gt;&lt;/citation_related&gt;&lt;cluster_related&gt;&lt;identifiers&gt;&lt;identifier value="0010-0870" type="issn"&gt;&lt;/identifier&gt;&lt;/identifiers&gt;&lt;/cluster_related&gt;&lt;/dynamic_data&gt;&lt;/REC&gt;
+&lt;REC r_id_disclaimer="ResearcherID data provided by Clarivate Analytics"&gt;&lt;UID&gt;WOS:A1976BW18000001&lt;/UID&gt;&lt;static_data&gt;&lt;summary&gt;&lt;EWUID&gt;&lt;WUID coll_id="WOS"&gt;&lt;/WUID&gt;&lt;edition value="WOS.SSCI"&gt;&lt;/edition&gt;&lt;edition value="WOS.SCI"&gt;&lt;/edition&gt;&lt;/EWUID&gt;&lt;pub_info coverdate="1976" has_abstract="N" issue="3" pubtype="Journal" pubyear="1976" sortdate="1976-01-01" vol="37"&gt;&lt;page begin="205" end="221" page_count="17"&gt;205-221&lt;/page&gt;&lt;/pub_info&gt;&lt;titles count="6"&gt;&lt;title type="source"&gt;COLLEGE &amp;amp; RESEARCH LIBRARIES&lt;/title&gt;&lt;title type="source_abbrev"&gt;COLL RES LIBR&lt;/title&gt;&lt;title type="abbrev_iso"&gt;Coll. Res. Libr.&lt;/title&gt;&lt;title type="abbrev_11"&gt;COLL RES LI&lt;/title&gt;&lt;title type="abbrev_29"&gt;COLL RES LIBR&lt;/title&gt;&lt;title type="item"&gt;CENTURY OF COOPERATIVE PROGRAMS AMONG ACADEMIC-LIBRARIES&lt;/title&gt;&lt;/titles&gt;&lt;names count="1"&gt;&lt;name daisng_id="19670132" seq_no="1" role="author"&gt;&lt;display_name&gt;WEBER, DC&lt;/display_name&gt;&lt;full_name&gt;WEBER, DC&lt;/full_name&gt;&lt;wos_standard&gt;WEBER, DC&lt;/wos_standard&gt;&lt;first_name&gt;DC&lt;/first_name&gt;&lt;last_name&gt;WEBER&lt;/last_name&gt;&lt;/name&gt;&lt;/names&gt;&lt;doctypes count="1"&gt;&lt;doctype&gt;Article&lt;/doctype&gt;&lt;/doctypes&gt;&lt;publishers&gt;&lt;publisher&gt;&lt;address_spec addr_no="1"&gt;&lt;full_address&gt;50 E HURON ST, CHICAGO, IL 60611&lt;/full_address&gt;&lt;city&gt;CHICAGO&lt;/city&gt;&lt;/address_spec&gt;&lt;names count="1"&gt;&lt;name addr_no="1" role="publisher" seq_no="1"&gt;&lt;display_name&gt;ASSOC COLL RESEARCH LIBRARIES&lt;/display_name&gt;&lt;full_name&gt;ASSOC COLL RESEARCH LIBRARIES&lt;/full_name&gt;&lt;/name&gt;&lt;/names&gt;&lt;/publisher&gt;&lt;/publishers&gt;&lt;/summary&gt;&lt;fullrecord_metadata&gt;&lt;languages count="1"&gt;&lt;language type="primary"&gt;English&lt;/language&gt;&lt;/languages&gt;&lt;normalized_languages count="1"&gt;&lt;language type="primary"&gt;English&lt;/language&gt;&lt;/normalized_languages&gt;&lt;normalized_doctypes count="1"&gt;&lt;doctype&gt;Article&lt;/doctype&gt;&lt;/normalized_doctypes&gt;&lt;refs count="41"&gt;&lt;/refs&gt;&lt;addresses count="1"&gt;&lt;address_name&gt;&lt;address_spec addr_no="1"&gt;&lt;full_address&gt;STANFORD UNIV LIB,STANFORD,CA 94305&lt;/full_address&gt;&lt;organizations count="2"&gt;&lt;organization&gt;STANFORD UNIV LIB&lt;/organization&gt;&lt;organization pref="Y"&gt;Stanford University&lt;/organization&gt;&lt;/organizations&gt;&lt;city&gt;STANFORD&lt;/city&gt;&lt;state&gt;CA&lt;/state&gt;&lt;country&gt;USA&lt;/country&gt;&lt;zip location="AP"&gt;94305&lt;/zip&gt;&lt;/address_spec&gt;&lt;/address_name&gt;&lt;/addresses&gt;&lt;category_info&gt;&lt;headings count="1"&gt;&lt;heading&gt;Science &amp;amp; Technology&lt;/heading&gt;&lt;/headings&gt;&lt;subheadings count="1"&gt;&lt;subheading&gt;Technology&lt;/subheading&gt;&lt;/subheadings&gt;&lt;subjects count="3"&gt;&lt;subject ascatype="traditional" code="NU"&gt;Information Science &amp;amp; Library Science&lt;/subject&gt;&lt;subject ascatype="extended"&gt;Information Science &amp;amp; Library Science&lt;/subject&gt;&lt;subject ascatype="traditional" code="NU"&gt;INFORMATION SCIENCE &amp;amp; LIBRARY SCIENCE&lt;/subject&gt;&lt;/subjects&gt;&lt;/category_info&gt;&lt;/fullrecord_metadata&gt;&lt;item xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" coll_id="WOS" xsi:type="itemType_wos"&gt;&lt;ids avail="Y"&gt;BW180&lt;/ids&gt;&lt;bib_id&gt;37 (3): 205-221 1976&lt;/bib_id&gt;&lt;/item&gt;&lt;/static_data&gt;&lt;dynamic_data&gt;&lt;citation_related&gt;&lt;tc_list&gt;&lt;silo_tc coll_id="WOS" local_count="12"&gt;&lt;/silo_tc&gt;&lt;/tc_list&gt;&lt;/citation_related&gt;&lt;cluster_related&gt;&lt;identifiers&gt;&lt;identifier type="issn" value="0010-0870"&gt;&lt;/identifier&gt;&lt;/identifiers&gt;&lt;/cluster_related&gt;&lt;/dynamic_data&gt;&lt;/REC&gt;
+&lt;/records&gt;