mead 0.0.5

data/test/helper.rb

@@ -0,0 +1,38 @@
+require 'rubygems'
+require 'bundler'
+require 'ruby-debug'
+begin
+  Bundler.setup(:default, :development)
+rescue Bundler::BundlerError => e
+  $stderr.puts e.message
+  $stderr.puts "Run `bundle install` to install missing gems"
+  exit e.status_code
+end
+require 'test/unit'
+require 'shoulda'
+require 'fakeweb'
+#require 'test_benchmark' # ruby 1.8.7 only?
+$LOAD_PATH.unshift(File.join(File.dirname(__FILE__), '..', 'lib'))
+$LOAD_PATH.unshift(File.dirname(__FILE__))
+require 'mead'
+
+# disallow connections and then register all URIs used
+FakeWeb.allow_net_connect = false
+FakeWeb.register_uri(:get, 'http://www.lib.ncsu.edu/findingaids/ua023_031.xml',
+  :response => File.join('test', 'fixtures', 'ua023_031.xml'))
+FakeWeb.register_uri(:get, 'http://www.lib.ncsu.edu/findingaids/mc00310.xml',
+  :response => File.join('test', 'fixtures', 'mc00310.xml'))
+
+class Test::Unit::TestCase
+end
+
+# test cases should use stricter NCSU specific format validation
+module Mead
+  class Identifier
+    def self.format_regexp
+      container_codes = Mead::CONTAINER_MAPPING.keys.join('|')
+      /^(ua\d{3}|mc\d{5})(_\d{3})?-\d{3}-(#{container_codes})\d{4}-\d{3}([A-Z])?-\d{3}(_\d{4})?$/
+    end
+  end
+end
+

data/test/test_barcode.rb

@@ -0,0 +1,25 @@
+require 'helper'
+
+class TestBarcode < Test::Unit::TestCase
+  context 'barcode creator' do
+    setup do
+      mead_file = File.open('test/ead/ua023_031.xml')
+      @mead1 = Mead::Identifier.new('ua023_031-001-cb0003-005-001', mead_file).extract
+      @mead2 = Mead::Identifier.new('ua023_031-001-cb0013-019A-001', mead_file).extract
+      @barcode1 = Mead::Barcode.new(@mead1)
+    end
+    should 'hold an instance of a mead' do
+      assert_equal Mead::Identifier, @barcode1.mead.class
+    end
+    should 'create a barcode file' do
+      png_file_path = @barcode1.output_barcode(File.expand_path(File.join(File.dirname(__FILE__), '..')))
+      assert File.exists?(png_file_path)
+      File.delete(png_file_path)
+    end
+    should 'create a text file' do
+      text_file_path = @barcode1.output_label(File.expand_path(File.join(File.dirname(__FILE__), '..')))
+      assert File.exists?(text_file_path)
+      File.delete(text_file_path)
+    end
+  end
+end

data/test/test_ead.rb

@@ -0,0 +1,193 @@
+require 'helper'
+
+class TestMead < Test::Unit::TestCase
+  context "Given an eadid mc00240" do
+
+    setup do
+      opts = {:eadid => 'mc00240', :file => File.open('test/ead/mc00240.xml')}
+      @ead = Mead::Ead.new(opts)
+      @containers = @ead.containers
+    end
+
+    should "create information for a stub record for the first container" do
+      expected = {:mead => 'mc00240-001-ff0181-000-001',
+        :title => 'Adams, J. H. - Barn, 1929 (1075)',
+        :series => 1,
+        :containers => ['flatfolder 181']
+      }
+      assert_equal expected, @containers[0]
+    end
+
+    should "create information for a stub record for the last container" do
+      expected = {:mead => 'mc00240-003-bx0069-000-001',
+        :title => 'Personnel Ledger, Pt. 2, 1956',
+        :series => 3,
+        :containers=>["Box 69"]
+      }
+      assert_equal expected, @containers.last
+    end
+
+    should "determine this to be an invalid ead not suitable for large scale digitization yet" do
+      assert_equal false, @ead.valid?
+    end
+    should "provide information on the containers with invalid (duplicate) mead identifiers" do
+      assert_equal 47, @ead.invalid.length
+      assert_equal 20, @ead.dups.length
+      expected = {:series=>1, :title=>"Breach, William- Residence", 
+        :mead=>"mc00240-001-ff0298-000-001", :containers=>["flatfolder 298"]}
+      assert_equal expected, @ead.invalid.first
+    end
+
+    context "converted to csv" do
+      setup do
+        @csv = @ead.to_csv
+        @csv_lines = @csv.split("\n")
+      end
+      should "be able to create a csv file from the parsed ead" do
+        assert_equal 'mead,title,series,containers', @csv_lines[0]
+      end
+
+      should "be able to create good csv data from the parsed ead for the first container" do
+        expected = 'mc00240-001-ff0181-000-001,"Adams, J. H. - Barn, 1929 (1075)",1,flatfolder 181'
+        assert_equal expected, @csv_lines[1]
+      end
+
+      should "be able to create good csv data from the parsed ead for the last container" do
+        expected = 'mc00240-003-bx0069-000-001,"Personnel Ledger, Pt. 2, 1956",3,Box 69'
+        assert_equal expected, @csv_lines.last
+      end
+    end
+
+  end
+
+  context "Given an eadid of ua023_031" do
+    setup do
+      opts = {:eadid => 'ua023_031',:file => File.open('test/ead/ua023_031.xml')}
+      @ead = Mead::Ead.new(opts)
+      @containers = @ead.containers
+    end
+    should 'create information for a stub record for the first container' do
+      expected = {:title=>"Sules V-B on Apple [3] - Grape Study - Set #17",
+        :series=>1,
+        :mead=>"ua023_031-001-cb0006-031-001",
+        :containers => ["cardbox 6", "Envelope 31"]
+        }
+      assert_equal expected, @containers[0]
+    end
+
+    should 'create information for a stub record for the second container' do
+      expected = {
+        :containers=>["cardbox 10", "Envelope 10"],
+        :mead=>"ua023_031-001-cb0010-010-001",
+        :title=>
+          "(no. 13) Eastern Carolina, North Carolina - Diec - cabbage and treated - Hand Colored Slides, Numbered Series",
+        :series=>1}
+      assert_equal expected, @containers[1]
+    end
+
+    should 'create information for a stub record for the last container' do
+      expected = {:mead => 'ua023_031-010-cb0019-006-001',
+        :title => 'Examples of original storage envelopes',
+        :series => 10,
+        :containers=>["cardbox 19", "Envelope 6"]
+      }
+      assert_equal expected, @containers.last
+    end
+
+  end
+
+  context "Given an eadid of mc00310" do
+    setup do
+
+      opts = {:eadid => 'mc00310', :url => 'http://www.lib.ncsu.edu/findingaids/mc00310.xml'}
+      @ead = Mead::Ead.new( opts)
+      @containers = @ead.containers
+    end
+    should 'determine it to be a valid ead for large scale digitization' do
+      assert @ead.valid?
+    end
+  end
+
+  context "Given a baseurl and no filehandle for an Ead" do
+    setup do
+      opts = {:eadid => 'ua023_031', :baseurl => 'http://www.lib.ncsu.edu/findingaids'}
+      @ead = Mead::Ead.new(opts)
+    end
+    should 'save the baseurl as an instance variable' do
+      assert_equal 'http://www.lib.ncsu.edu/findingaids', @ead.baseurl
+    end
+    should 'create information for a stub record for the last container' do
+      containers = @ead.containers
+      expected = {:title=>"Sules V-B on Apple [3] - Grape Study - Set #17", :series=>1, 
+        :mead=>"ua023_031-001-cb0006-031-001",
+          :containers => ["cardbox 6", "Envelope 31"]
+        }
+      assert_equal expected, containers.first
+    end
+  end
+
+  context "Given a url and no file or baseurl for an Ead" do
+    setup do
+      opts = {:eadid => 'ua023_031', :url => 'http://www.lib.ncsu.edu/findingaids/ua023_031.xml'}
+      @ead = Mead::Ead.new(opts)
+    end
+    should 'save the url to an instance variable' do
+      assert_equal 'http://www.lib.ncsu.edu/findingaids/ua023_031.xml', @ead.url
+    end
+    should 'create information for a stub record for the last container' do
+      containers = @ead.containers
+      expected = {:title=>"Sules V-B on Apple [3] - Grape Study - Set #17", 
+      :series=>1, :mead=>"ua023_031-001-cb0006-031-001",
+          :containers => ["cardbox 6", "Envelope 31"]}
+      assert_equal expected, containers.first
+    end
+  end
+
+  context "Given a baseurl and no eadid" do
+    should 'raise an exception' do
+      opts = {:baseurl => 'http://www.lib.ncsu.edu/findingaids'}
+      assert_raise RuntimeError do
+        Mead::Ead.new(opts)
+      end
+    end
+  end
+
+
+  context 'missing an eadid' do
+    setup do
+      @expected = {:title=>"Sules V-B on Apple [3] - Grape Study - Set #17",
+        :series=>1,
+        :mead=>"ua023_031-001-cb0006-031-001",
+        :containers=>["cardbox 6", "Envelope 31"]
+        }
+    end
+    context 'Given a file' do
+      setup do
+        opts = {:file => File.open('test/ead/ua023_031.xml')}
+        @ead = Mead::Ead.new(opts)
+      end
+      should 'try to find the eadid in the EAD XML' do
+        assert_equal 'ua023_031', @ead.eadid
+      end
+      should 'get information for the first container' do
+        assert_equal @expected, @ead.containers.first
+      end
+    end
+
+    context 'Given a full URL' do
+      setup do
+        opts = {:url => 'http://www.lib.ncsu.edu/findingaids/ua023_031.xml'}
+        @ead = Mead::Ead.new(opts)
+      end
+      should 'try to find the eadid in the EAD XML' do
+        assert_equal 'ua023_031', @ead.eadid
+      end
+      should 'get information for the first container' do
+        assert_equal @expected, @ead.containers.first
+      end
+    end
+  end
+
+
+end
+

data/test/test_ead_validator.rb

@@ -0,0 +1,18 @@
+require 'helper'
+
+class TestEadValidator < Test::Unit::TestCase
+  
+  context 'Given a directory of EAD XML validate them all' do
+    setup do
+      @eadv = Mead::EadValidator.new('test/ead/')
+    end
+    should 'store the directory path' do
+      assert_equal 'test/ead/', @eadv.directory
+    end
+    should 'return results' do
+      expected = {:valid=>["ua021_428", "ua023_006"], 
+      :invalid=>["mc00145","mc00240", "ua015_010", "ua023_031", "ua110_041"]}
+      assert_equal expected, @eadv.validate!
+    end
+  end
+end

data/test/test_extractor.rb

@@ -0,0 +1,83 @@
+require 'helper'
+
+class TestMeadExtractor < Test::Unit::TestCase
+  context "the mead ua023_031-008-cb0013-001-001" do
+    setup do
+
+      @identifier = 'ua023_031-008-cb0013-001-001'
+      @expected_1 = [
+        {:level=>"file", 
+        :unitdate=>nil, 
+        :unitid=>nil, 
+        :unittitle=>
+            "Horticulture - students in greenhouse - for photograph see - Agriculture school - Horticulture - Hand Colored Slides, agriculture school",
+        :item_location => 'cardbox 13, Envelope 1',
+        :containers => [
+          Mead::Container.new(:type => 'cardbox', :label => "Mixed materials", :text => '13'), 
+          Mead::Container.new(:type => 'Envelope', :text => '1')]
+        },
+        {:level=>"subseries", :unitdate=>nil, :unitid=>nil, :unittitle=>"Students"},
+        {:level=>"series", :unitdate=>nil, :series_number=>8, :unitid=>"Series 8", :unittitle=>"People"}]
+    end
+
+    should "produce a good extraction with a filehandle location for the Ead" do
+      fh   = File.open('test/ead/ua023_031.xml')
+      mead = Mead::Identifier.new(@identifier, fh)
+      assert_equal @expected_1, Mead::Extractor.new(mead).extract
+    end
+
+    should "producde a good extraction with a remote location Ead containing no eadid" do
+      assert_equal @expected_1, Mead::Extractor.new(Mead::Identifier.new(@identifier, 'http://www.lib.ncsu.edu/findingaids')).extract
+    end
+    should 'produce a good extraction with a remote location Ead with a full URL' do
+      assert_equal @expected_1, Mead::Extractor.new(Mead::Identifier.new(@identifier, 'http://www.lib.ncsu.edu/findingaids/ua023_031.xml')).extract
+    end
+
+
+    context 'mc00240-001-ff0052-000-001' do
+      setup do
+        @expected_mc00240 = [{:unittitle=>"Friends Church",
+                    :item_location=>"flatfolder 52",
+                    :unitdate=>"1927",
+                    :level=>"file",
+                    :unitid=>"903",
+                    :containers => 
+                    [Mead::Container.new(:type => 'flatfolder', :label => 'Mixed materials', :text => '52')]},
+                   {:series_number=>1,
+                    :unittitle=>"Drawings",
+                    :unitdate=>"1917-1980",
+                    :level=>"series",
+                    :unitid=>"MC 240 Series 1"}]
+      end
+      should 'handle empty folder properly' do
+        mead = Mead::Identifier.new('mc00240-001-ff0052-000-001', File.open('test/ead/mc00240.xml'))
+        assert_equal @expected_mc00240, Mead::Extractor.new(mead).extract
+      end
+
+      should 'cache the extraction within the Mead::Identifier' do
+        mead = Mead::Identifier.new('mc00240-001-ff0052-000-001', File.open('test/ead/mc00240.xml')).extract
+        assert_equal @expected_mc00240, mead.metadata
+      end
+    end
+
+  end
+
+  should "raise an exception if the extractor is given something other than a Mead::Identifier" do
+    assert_raise RuntimeError do
+      Mead::Extractor.new({})
+    end
+  end
+  should "raise an exception if the extractor tries to extract from a Mead::Identifier without an Ead location set" do
+    assert_raise RuntimeError do
+      Mead::Extractor.new(Mead::Identifier.new('mc00240-001-ff0052-000-001')).extract
+    end
+  end
+  should "raise an exception when a duplicate Mead::Identifier is found by matching too many nodes" do
+    assert_raise RuntimeError do
+      Mead::Extractor.new(Mead::Identifier.new('mc00240-003-bx0069-000-001', File.open('test/ead/mc00240.xml'))).extract
+    end
+  end
+
+
+end
+

data/test/test_mc00145.rb

@@ -0,0 +1,19 @@
+require 'helper'
+
+class TestMc00145 < Test::Unit::TestCase
+  context "the ead mc00145" do
+    setup do
+      opts = {:eadid => 'mc00145', :file => File.open('test/ead/mc00145.xml')}
+      @ead = Mead::Ead.new(opts)
+      @containers = @ead.containers
+    end
+    
+    should "create short meads for component parts with more than 2 containers" do
+      assert !@containers.empty?
+    end
+    
+    should "create meads for EAD with no series" do
+      assert_equal 'mc00145-001-te0001-000-001', @containers.last[:mead]
+    end
+  end
+end

data/test/test_mc00240.rb

@@ -0,0 +1,140 @@
+require 'helper'
+
+class TestMeadMC00240 < Test::Unit::TestCase
+
+  context "a legacy mead from mc00240" do
+    setup do
+      @mead_id = 'mc00240-001-ff0147-001-001_0001' #empty folder was given 001 rather than 000
+      @fh = File.open('test/ead/mc00240.xml')
+      @mead = Mead::Identifier.new(@mead_id, @fh)
+    end
+    should "retrieve the correct metadata" do
+      @mead.extract
+      expected = [{:unittitle=>"Moravian Chapel at Southside", :level=>"file",
+                    :item_location=>"flatfolder 147", :unitdate=>"1928", :unitid=>"1034",
+                      :containers => 
+                      [Mead::Container.new(:type => 'flatfolder', 
+                                            :label => "Mixed materials",
+                                            :text => '147')]
+                    }, 
+                   {:unittitle=>"Drawings", :series_number=>1,
+                    :level=>"series", :unitdate=>"1917-1980", :unitid=>"MC 240 Series 1"}]
+      assert_equal expected, @mead.metadata
+    end
+  end
+  
+  
+  context "a mead from mc00240" do
+    setup do
+      @mead_id = 'mc00240-001-ff0042-000-001'      
+      @fh = File.open('test/ead/mc00240.xml')
+      @mead    = Mead::Identifier.new(@mead_id, @fh)
+    end
+
+    context "parsing a mead from mc00240" do      
+
+      should "produce expected output of eadid" do
+        assert_equal @mead.eadid, 'mc00240'
+      end
+
+      should "produce the expected series" do
+        assert_equal @mead.series, '1'
+      end
+
+      should "produce the expected container" do
+        expected = {:type=> 'flatfolder', :number => '42'}
+        assert_equal expected, @mead.container
+      end
+
+      should "produce the expected folder" do
+        assert_equal @mead.folder, nil
+      end
+
+      should "produce the expected sequence" do
+        assert_equal @mead.sequence, '1'
+      end
+    end
+
+    context "extracting metadata from a mead" do
+      setup do
+        @extractor = Mead::Extractor.new(@mead)
+        #@file = File.open('test/ead/mc00240.xml')
+      end
+
+      should "be able to create an extractor" do
+        assert_equal @extractor.class, Mead::Extractor
+      end
+
+      should "convert a string to a mead object" do
+        assert_equal @extractor.mead.class, Mead::Identifier
+      end
+
+      context "should give back the metadata" do
+        setup do
+          @result = @extractor.extract
+        end
+        
+        should "Cache the metadata in the Mead::Identifier" do
+          mead = Mead::Identifier.new(@mead_id, @fh).extract
+          assert_equal @result, mead.metadata  
+        end
+
+        should "extract the item's unittitle" do
+          assert_equal 'Amos Hosiery Mill - Addition', @extractor.stack[0][:unittitle]
+        end
+
+        should "extract the item's unitdate" do
+          assert_equal '1953', @extractor.stack[0][:unitdate]
+        end
+        
+        should "extract the item's level" do
+          assert_equal 'file', @extractor.stack[0][:level]
+        end
+        
+        should "extract the item's unitid" do
+          assert_equal '1421', @extractor.stack[0][:unitid]
+        end
+        
+        should "extract the item's containers" do          
+          assert_equal @extractor.stack[0][:containers].first.class, Mead::Container
+        end
+
+        should "extract the parent unittitle" do
+          assert_equal "Drawings", @extractor.stack[1][:unittitle]
+        end
+
+        should "extract the parent unitdate" do
+          assert_equal "1917-1980", @extractor.stack[1][:unitdate]
+        end
+        
+        should "extract the parent level" do
+          assert_equal "series", @extractor.stack[1][:level]
+        end
+        
+        should "extract the parent unitid" do
+          assert_equal 'MC 240 Series 1', @extractor.stack[1][:unitid]
+        end
+        
+        should "extract a series' series number" do
+          assert_equal 1, @extractor.stack[1][:series_number]
+        end
+
+#        should "only extract up to the series level" do
+#          assert_equal [
+#            {:unittitle=>"Amos Hosiery Mill - Addition", :unitdate=>"1953",
+#            :level => 'file', :unitid => '1421', :item_location => 'flatfolder 42'},
+#            {:unittitle=>"Drawings", :unitdate=>"1917-1980", :level => 'series',
+#              :unitid => 'MC 240 Series 1', :series_number => 1
+#            }
+#          ], @extractor.stack
+#        end
+
+      end
+
+    end
+
+
+
+  end
+
+end

data/test/test_mead.rb

@@ -0,0 +1,5 @@
+require 'helper'
+
+class TestMead < Test::Unit::TestCase
+  
+end

data/test/test_ua015_010.rb

@@ -0,0 +1,111 @@
+require 'helper'
+
+class TestMeadUA015_010 < Test::Unit::TestCase
+
+  context "a mead from ua015_010" do
+    setup do
+      @mead = 'ua015_010-004-bx0039-005-002'     
+    end
+
+    context "parsing a mead from ua015_010" do
+      setup do
+        @result = Mead::Identifier.new(@mead, File.open('test/ead/ua015_010.xml')) #this is where processing
+      end
+
+      should "produce expected output of eadid" do
+        assert_equal @result.eadid, 'ua015_010'
+      end
+
+      should "produce the expected series" do
+        assert_equal @result.series, '4'
+      end
+
+      should "produce the expected container" do
+        expected = {:type=> 'box', :number => '39'}
+        assert_equal expected, @result.container
+      end
+
+      should "produce the expected folder" do
+        assert_equal @result.folder, {:type=> 'folder', :number => '5'}
+      end
+
+      should "produce the expected sequence" do
+        assert_equal @result.sequence, '2'
+      end
+    end
+
+    context "extracting metadata from a mead" do
+      setup do
+        @result = Mead::Identifier.new(@mead, File.open('test/ead/ua015_010.xml'))
+        @extractor = Mead::Extractor.new(@result)        
+      end
+
+      should "be able to create an extractor" do
+        assert_equal @extractor.class, Mead::Extractor
+      end
+
+      should "convert a string to a mead object" do
+        assert_equal @extractor.mead.class, Mead::Identifier
+      end
+
+      context "should give back the metadata" do
+        setup do          
+          @result = @extractor.extract
+        end
+
+        should "extract the item's unittitle" do
+          assert_equal 'Programs', @extractor.stack[0][:unittitle]
+        end
+
+        should "extract the item's unitdate" do
+          assert_equal '1949-1950', @extractor.stack[0][:unitdate]
+        end
+        
+        should "extract the item level" do
+          assert_equal 'file', @extractor.stack[0][:level]
+        end
+        
+        should "extract the item unitid" do
+          assert_nil @extractor.stack[0][:unitid]
+        end
+
+        should "extract the parent unittitle" do
+          assert_equal "Men's Basketball", @extractor.stack[1][:unittitle]
+        end
+
+        should "extract the parent unitdate" do
+          assert_equal "1911-2006", @extractor.stack[1][:unitdate]
+        end
+        
+        should "extract the parent level" do
+          assert_equal 'series', @extractor.stack[1][:level]
+        end
+        
+        should "extract the parent unitid" do
+          assert_equal 'Series 4', @extractor.stack[1][:unitid]
+        end
+        
+        should "extract a series' series number" do
+          assert_equal 4, @extractor.stack[1][:series_number]
+        end
+
+#        should "only extract up to the series level" do
+#          assert_equal [
+#            {:unittitle=>"Programs", :unitdate=>"1949-1950",
+#            :level => 'file', :unitid => nil, :item_location => 'Box 39, Folder 5'},
+#            {:unittitle=>"Men's Basketball", :unitdate=>"1911-2006",
+#              :level => 'series', :unitid => 'Series 4',
+#              :series_number => 4
+#            }
+#          ], @extractor.stack
+#        end
+
+      end
+
+    end
+
+    
+
+  end
+  
+end

data/test/test_ua021_428.rb

@@ -0,0 +1,26 @@
+require 'helper'
+
+class TestMeadUA021_428 < Test::Unit::TestCase
+  context "extracting metadata from a mead" do
+      setup do
+        @mead = 'ua021_428-001-cb0007-000-001'
+        @result = Mead::Identifier.new(@mead, File.open('test/ead/ua021_428.xml'))
+        @extractor = Mead::Extractor.new(@result) 
+        @extractor.extract       
+      end
+      
+      should "have metadata" do
+        assert !@extractor.stack.empty?
+      end
+      
+      should 'show ua021_428-001-cb0007-000-001 to be a valid mead' do
+        mead = Mead::Identifier.new(@mead, File.open('test/ead/ua021_428.xml')).extract
+        assert mead.valid_format?
+        assert mead.valid?
+      end
+
+      
+    end
+  
+end
+