cul-fedora 0.7.0 → 0.8.0

data/VERSION

@@ -1 +1 @@
-0.7.0
+0.8.0

data/cul-fedora.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = %q{cul-fedora}
-  s.version = "0.7.0"
+  s.version = "0.8.0"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["James Stuart"]
-  s.date = %q{2011-04-22}
+  s.date = %q{2011-05-02}
   s.description = %q{Columbia-specific Fedora libraries}
   s.email = %q{tastyhat@jamesstuart.org}
   s.extra_rdoc_files = [

data/lib/cul-fedora/item.rb

@@ -103,9 +103,12 @@ module Cul
 
         get_fullname = lambda { |node| node.nil? ? nil : (node.css("namePart[@type='family']").collect(&:content) | node.css("namePart[@type='given']").collect(&:content)).join(", ") }
 
-        roles = ["Author","author","Creator","Thesis Advisor","Collector","Owner","Speaker","Seminar Chairman","Secretary","Rapporteur","Committee Member","Degree Grantor","Moderator","Editor","Interviewee","Interviewer","Organizer of Meeting","Originator","Teacher"]
-
+        roles = ["Author","Creator","Thesis Advisor","Collector","Owner","Speaker","Seminar Chairman","Secretary","Rapporteur","Committee Member","Degree Grantor","Moderator","Editor","Interviewee","Interviewer","Organizer of Meeting","Originator","Teacher"]
+        roles = roles.map { |role| role.downcase }
 
+        organizations = []
+        departments = []
+          
         begin
           collections = self.belongsTo
           meta = describedBy.first
@@ -124,29 +127,57 @@ module Cul
 
 
 
-            title = normalize_space.call(mods.css("titleInfo>nonSort,title").collect(&:content).join(" "))
+            title = mods.css("titleInfo>title").first.text
+            title_search = normalize_space.call(mods.css("titleInfo>nonSort,title").collect(&:content).join(" "))
+            record_creation_date = mods.at_css("recordInfo>recordCreationDate")
+            if(record_creation_date.nil?)
+              record_creation_date = mods.at_css("recordInfo>recordChangeDate")
+            end
+            if(!record_creation_date.nil? || !record_creation_date.empty?)
+              record_creation_date = DateTime.parse(record_creation_date.text.gsub("UTC", "").strip)
+              add_field.call("record_creation_date", record_creation_date.strftime("%Y-%m-%dT%H:%M:%SZ"))
+            end
             add_field.call("title_display", title)
-            add_field.call("title_search", title)
+            add_field.call("title_search", title_search)
 
             all_names = []
             mods.css("name[@type='personal']").each do |name_node|
-              if name_node.css("role>roleTerm[@type='text']").collect(&:content).any? { |role| roles.include?(role) }
+              if name_node.css("role>roleTerm").collect(&:content).any? { |role| roles.include?(role) }
 
                 fullname = get_fullname.call(name_node)
 
                 all_names << fullname
-                add_field.call("author_id_uni", name_node.at_css("authorID[@type='institution']"))
-                add_field.call("author_id_repository", name_node.at_css("authorID[@type='repository']"))
-                add_field.call("author_id_naf", name_node.at_css("authorID[@type='naf']"))
+                if(!name_node["ID"].nil?)
+                  add_field.call("author_id_uni", name_node["ID"])
+                end
                 add_field.call("author_search", fullname.downcase)
                 add_field.call("author_facet", fullname)
+                
+                name_node.css("affiliation").each do |affiliation_node|
+                  affiliation_text = affiliation_node.text
+                  if(affiliation_text.include?(". "))
+                    affiliation_split = affiliation_text.split(". ")
+                    organizations.push(affiliation_split[0].strip)
+                    departments.push(affiliation_split[1].strip)
+                  end
+                end
 
               end
-
+            end
+            
+            mods.css("name[@type='corporate']").each do |corp_name_node|
+              if(!corp_name_node["ID"].nil? && corp_name_node["ID"].include?("originator"))
+                name_part = corp_name_node.at_css("namePart").text
+                if(name_part.include?(". "))
+                  name_part_split = name_part.split(". ")
+                  organizations.push(name_part_split[0].strip)
+                  departments.push(name_part_split[1].strip)
+                end
+              end
             end
 
             add_field.call("authors_display",all_names.join("; "))
-            add_field.call("date", mods.at_css("*[@keyDate='yes']"))
+            add_field.call("pub_date", mods.at_css("*[@keyDate='yes']"))
 
             mods.css("genre").each do |genre_node|
               add_field.call("genre_facet", genre_node)
@@ -158,14 +189,14 @@ module Cul
             add_field.call("abstract", mods.at_css("abstract"))
             add_field.call("handle", mods.at_css("identifier[@type='hdl']"))
 
-            mods.css("subject:not([@authority='local'])>topic").each do |topic_node|
-              add_field.call("keyword_search", topic_node.content.downcase)
-              add_field.call("keyword_facet", topic_node)
-            end
-
-            mods.css("subject[@authority='local']>topic").each do |topic_node|
-              add_field.call("subject", topic_node)
-              add_field.call("subject_search", topic_node)
+            mods.css("subject").each do |subject_node|
+              if(subject_node.attributes.count == 0)
+                subject_node.css("topic").each do |topic_node|
+                  add_field.call("keyword_search", topic_node.content.downcase)
+                  add_field.call("subject", topic_node)
+                  add_field.call("subject_search", topic_node)
+                end
+              end
             end
 
 
@@ -203,7 +234,22 @@ module Cul
               add_field.call("geographic_area_search", geo)
             end
 
-            add_field.call("export_as_mla_citation_txt","test")
+            add_field.call("export_as_mla_citation_txt","")
+            
+            if(organizations.count > 0)
+              organizations = organizations.uniq
+              organizations.each do |organization|
+                add_field.call("affiliation_organization", organization)
+              end
+            end
+            
+            if(departments.count > 0)
+              departments = departments.uniq
+              departments.each do |department|
+                add_field.call("affiliation_department", department.to_s.sub(", Department of", "").strip)
+              end
+            end
+            
           end
 
 

data/lib/cul-fedora/solr.rb

@@ -15,6 +15,11 @@ module Cul
         @rsolr ||= RSolr.connect(:url => @url)
       end
 
+      def delete_index
+        rsolr.delete_by_query("*:*")
+        rsolr.commit
+      end
+      
       def ingest(options = {})
         format = options.delete(:format) || raise(ArgumentError, "needs format")
 
@@ -22,13 +27,14 @@ module Cul
         items = [items] unless items.kind_of?(Array)
         collections = options.delete(:collections) || []
         collections = [collections] unless collections.kind_of?(Array)
+        ignore = options.delete(:ignore) || []
+        ignore = [ignore] unless ignore.kind_of?(Array)
 
         overwrite = options.delete(:overwrite) || false
         process = options.delete(:process) || nil
         skip = options.delete(:skip) || nil
 
-
-
+        
         collections.each do |collection|
           items |= collection.listMembers
         end
@@ -40,6 +46,12 @@ module Cul
         errors = {}
       
         items.each do |i|
+          
+          if(ignore.index(i.pid).nil? == false)
+            $stdout.puts "Ignoring " + i.pid + "..."
+            next
+          end
+          
           if process && skip && skip > 0
             skip -= 1
             next
@@ -55,7 +67,7 @@ module Cul
           end    
           
 
-
+          $stdout.puts "Indexing " + i.pid + "..."
 
           result_hash = i.send("index_for_#{format}", options)
 

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: cul-fedora
 version: !ruby/object:Gem::Version 
-  hash: 3
+  hash: 63
   prerelease: false
   segments: 
   - 0
-  - 7
+  - 8
   - 0
-  version: 0.7.0
+  version: 0.8.0
 platform: ruby
 authors: 
 - James Stuart
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-04-22 00:00:00 -04:00
+date: 2011-05-02 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency