solrizer.rb in solrizer-fedora-2.0.0

- old
+ new

@@ -5,13 +5,10 @@
 require 'solrizer/xml'
 require 'solrizer/html'
 
 require 'active_support/core_ext/hash'
 
-require 'fastercsv' # this is used by solrize_objects when you pass it a csv file of pids
-
-
 module Solrizer::Fedora
   class Solrizer
     ALL_FIELDS = [
       :pid, :label, :fType, :cModel, :state, :ownerId, :cDate, :mDate, :dcmDate, 
       :bMech, :title, :creator, :subject, :description, :contributor,
@@ -93,50 +90,30 @@
       # retrieve a list of all the pids in the fedora repository
       num_docs = 1000000   # modify this number to guarantee that all the objects are retrieved from the repository
       puts "WARNING: You have turned off indexing of Full Text content.  Be sure to re-run indexer with @@index_full_text set to true in main.rb" if index_full_text == false
 
       if @@index_list == false
-        
-        objects = find_objects(:limit=>num_docs)
-
-        puts "Shelving #{objects.length} Fedora objects"
-        objects.each do |object|
-          solrize( object, opts )
-        end
-       
+        solrize_from_fedora_search(opts) 
       else
-         if File.exists?(@@index_list)
-            arr_of_pids = FasterCSV.read(@@index_list, :headers=>false)
-            
-            puts "Indexing from list at #{@@index_list}"
-            puts "Shelving #{arr_of_pids.length} Fedora objects"
-            
-           arr_of_pids.each do |row|
-              pid = row[0]
-              solrize( pid )
-           end #FASTERCSV
-         else
-           puts "#{@@index_list} does not exists!"
-         end #if File.exists
-       
-      end #if Index_LISTS
-    end #solrize_objects
+        solrize_from_csv
+      end
+    end
 
-    def find_objects(*args)
-        raise ArgumentError, "Missing query string" unless args.length >= 1
-        options = args.last.is_a?(Hash) ? args.pop : {}
-
-        params = {}
-        params[:query] = ''
-        params[:maxResults] = options[:limit] if options[:limit]
-        params[:pid] = true
-        
-        pids = []
-        connections.each do |conn|
-          response = Hash.from_xml(conn.find_objects(params))
-          pids << response["result"]["resultList"]["objectFields"].map{|x| x["pid"]}
+    def solrize_from_fedora_search(opts)
+      connections.each do |conn|
+        conn.search(nil) do |object|
+          solrize( object.pid, opts )
         end
-        pids.flatten
+      end 
+    end
+
+    def solrize_from_csv
+      raise ArgumentException, "#{@@index_list} does not exists!" unless File.exists?(@@index_list)
+      puts "Indexing from list at #{@@index_list}"
+      CSV.foreach(@@index_list) do |row|
+          pid = row[0]
+          solrize( pid )
+      end 
     end
 
     private
     
     def connections