cul-fedora 0.8.0 → 0.8.1

data/VERSION

@@ -1 +1 @@
-0.8.0
+0.8.1

data/cul-fedora.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = %q{cul-fedora}
-  s.version = "0.8.0"
+  s.version = "0.8.1"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["James Stuart"]
-  s.date = %q{2011-05-02}
+  s.date = %q{2011-05-26}
   s.description = %q{Columbia-specific Fedora libraries}
   s.email = %q{tastyhat@jamesstuart.org}
   s.extra_rdoc_files = [

data/lib/cul-fedora/item.rb

@@ -29,6 +29,15 @@ module Cul
         self.pid == other.pid
       end
 
+      def exists?
+        begin
+          request
+          return true
+        rescue Exception => e # we should really do some better checking of error type etc here
+          return false
+        end
+      end
+
       def request(options = {})
         @server.request(options.merge(:pid => @pid))
       end

data/lib/cul-fedora/server.rb

@@ -23,9 +23,8 @@ module Cul
       def request_path(options = {})
         sdef = options.delete(:sdef).to_s
         pid = options.delete(:pid).to_s
-        request = options.delete(:request).to_s
+        request = (options.delete(:request) || "").to_s
         method = (options.delete(:method) || "/get").to_s
-        raise(ArgumentError, "request necessary") if request.empty?
 
         sdef = "/" + sdef unless sdef.empty?
         pid = "/" + pid unless pid.empty?

data/lib/cul-fedora/solr.rb

@@ -20,8 +20,33 @@ module Cul
         rsolr.commit
       end
       
+      def delete_removed(fedora_server)
+        
+        start = 0
+        rows = 500
+        results = rsolr.select({:q => "", :fl => "id", :start => start, :rows => rows})
+        $stdout.puts "Deleting items removed from Fedora..."
+        while(!results["response"]["docs"].empty?)
+          
+          results["response"]["docs"].each do |doc|
+            if(!fedora_server.item(doc["id"]).exists?)
+              $stdout.puts "Deleting " + doc["id"] + "..."
+              rsolr.delete_by_query("id:" + doc["id"].to_s.gsub(/:/,'\\:'))
+            end
+          end
+          
+          start = start + rows
+          results = rsolr.select({:q => "", :fl => "id", :start => start, :rows => rows})
+        end
+        
+        rsolr.commit
+        
+      end
+      
       def ingest(options = {})
+        
         format = options.delete(:format) || raise(ArgumentError, "needs format")
+        fedora_server = options.delete(:fedora_server) || raise(ArgumentError, "needs fedora server")
 
         items = options.delete(:items) || []
         items = [items] unless items.kind_of?(Array)
@@ -30,10 +55,14 @@ module Cul
         ignore = options.delete(:ignore) || []
         ignore = [ignore] unless ignore.kind_of?(Array)
 
+        delete = options.delete(:delete_removed) || false
         overwrite = options.delete(:overwrite) || false
         process = options.delete(:process) || nil
         skip = options.delete(:skip) || nil
 
+        if delete
+          delete_removed(fedora_server)
+        end
         
         collections.each do |collection|
           items |= collection.listMembers
@@ -85,9 +114,19 @@ module Cul
             break if process <= 0
           end
 
+          if to_add.length >= 500
+            $stdout.puts "Adding Batch..."
+            rsolr.add(to_add)
+            to_add.clear
+          end
+
+        end
+        
+        if to_add.length > 0
+          $stdout.puts "Adding Batch..."
+          rsolr.add(to_add)
+          to_add.clear
         end
-          
-        rsolr.add(to_add)
         rsolr.commit
 
         return {:results => results, :errors => errors}

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: cul-fedora
 version: !ruby/object:Gem::Version 
-  hash: 63
+  hash: 61
   prerelease: false
   segments: 
   - 0
   - 8
-  - 0
-  version: 0.8.0
+  - 1
+  version: 0.8.1
 platform: ruby
 authors: 
 - James Stuart
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-05-02 00:00:00 -04:00
+date: 2011-05-26 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency