om 1.6.0 → 1.6.1

data/COMMON_OM_PATTERNS.textile

@@ -347,4 +347,16 @@ oxns
 document namespaces & node namespaces
 _no namespace_ (suppressing oxns in xpath queries)
 
-h3. :ref and :proxy Terms
+h3. :ref and :proxy Terms
+
+If needed (as a differentiator) you can use the root element as a member of the proxy address:
+<pre>
+    t.root(:path=>"mods")
+    t.titleInfo {
+      t.title
+    } 
+    This produces a relative xpath: (e.g. //titleInfo/title)
+    t.title(:proxy=>[:titleInfo, :title])
+    This produces an absolute query (e.g. /mods/titleInfo/title)
+    t.title(:proxy=>[:mods, :titleInfo, :title])
+</pre>

data/Gemfile

@@ -2,11 +2,14 @@ source "http://rubygems.org"
 
 gemspec
 
-gem 'ruby-debug', :platform => :ruby_18
-gem 'ruby-debug19', :platform => :ruby_19
-
 group :development, :test do
-  gem "rcov"
+  gem 'ruby-debug', :platform => :ruby_18
+  gem "debugger", :platform => :ruby_19
+
+  gem 'rcov', :platform => :mri_18
+  gem 'simplecov', :platform => :mri_19
+  gem 'simplecov-rcov', :platform => :mri_19
+
   gem "yard"
   gem "RedCloth", "~> 4.2.9"
 end

data/Gemfile.lock

@@ -9,20 +9,26 @@ GEM
   remote: http://rubygems.org/
   specs:
     RedCloth (4.2.9)
-    archive-tar-minitar (0.5.2)
     columnize (0.3.5)
+    debugger (1.1.3)
+      columnize (>= 0.3.1)
+      debugger-linecache (~> 1.1.1)
+      debugger-ruby_core_source (~> 1.1.2)
+    debugger-linecache (1.1.1)
+      debugger-ruby_core_source (>= 1.1.1)
+    debugger-ruby_core_source (1.1.3)
     diff-lcs (1.1.3)
     equivalent-xml (0.2.8)
       nokogiri (>= 1.4.3)
     linecache (0.46)
       rbx-require-relative (> 0.0.4)
-    linecache19 (0.5.12)
-      ruby_core_source (>= 0.1.4)
     mediashelf-loggable (0.4.9)
     metaclass (0.0.1)
     mocha (0.10.0)
       metaclass (~> 0.0.1)
+    multi_json (1.3.6)
     nokogiri (1.5.0)
+    rake (0.9.2.2)
     rbx-require-relative (0.0.5)
     rcov (0.9.11)
     rspec (2.7.0)
@@ -38,16 +44,12 @@ GEM
       ruby-debug-base (~> 0.10.4.0)
     ruby-debug-base (0.10.4)
       linecache (>= 0.3)
-    ruby-debug-base19 (0.11.25)
-      columnize (>= 0.3.1)
-      linecache19 (>= 0.5.11)
-      ruby_core_source (>= 0.1.4)
-    ruby-debug19 (0.11.6)
-      columnize (>= 0.3.1)
-      linecache19 (>= 0.5.11)
-      ruby-debug-base19 (>= 0.11.19)
-    ruby_core_source (0.1.5)
-      archive-tar-minitar (>= 0.5.2)
+    simplecov (0.6.4)
+      multi_json (~> 1.0)
+      simplecov-html (~> 0.5.3)
+    simplecov-html (0.5.3)
+    simplecov-rcov (0.2.3)
+      simplecov (>= 0.4.1)
     yard (0.7.4)
 
 PLATFORMS
@@ -55,11 +57,14 @@ PLATFORMS
 
 DEPENDENCIES
   RedCloth (~> 4.2.9)
+  debugger
   equivalent-xml (>= 0.2.4)
   mocha (>= 0.9.8)
   om!
+  rake
   rcov
   rspec (~> 2.0)
   ruby-debug
-  ruby-debug19
+  simplecov
+  simplecov-rcov
   yard

data/History.textile

@@ -1,3 +1,7 @@
+h3. 1.6.1
+  Integration spec to illustrate selective querying.
+  Add #use_terminology and #extend_terminology methods to extend existing OM terminologies
+
 h3. 1.6.0
   Delegate all methods on the dynamic node to the found values
   Allow arrays to be set on dynamic nodes

data/lib/om/version.rb

@@ -1,3 +1,3 @@
 module Om
-  VERSION = "1.6.0"
+  VERSION = "1.6.1"
 end

data/lib/om/xml/document.rb

@@ -5,12 +5,26 @@ module OM::XML::Document
 
   module ClassMethods
 
-    attr_accessor :terminology, :template_registry
+    attr_accessor :terminology, :terminology_builder, :template_registry
 
     # Sets the OM::XML::Terminology for the Document
     # Expects +&block+ that will be passed into OM::XML::Terminology::Builder.new
     def set_terminology &block
-      @terminology = OM::XML::Terminology::Builder.new( &block ).build
+      @terminology_builder = OM::XML::Terminology::Builder.new( &block )
+      
+      @terminology = @terminology_builder.build
+    end
+
+    # Update the OM::XML::Terminology with additional terms
+    def extend_terminology &block
+      @terminology_builder.extend_terminology(&block)
+      @terminology = @terminology_builder.build
+    end
+
+    # (Explicitly) inherit terminology from upstream classes
+    def use_terminology klass
+      @terminology_builder = klass.terminology_builder.dup
+      @terminology = @terminology_builder.build
     end
 
     # Define a new node template with the OM::XML::TemplateRegistry.

data/lib/om/xml/terminology.rb

@@ -102,6 +102,11 @@ class OM::XML::Terminology
       return current_term
     end
 
+    # Add additional terms into this terminology
+    def extend_terminology &block
+      yield self if block_given?
+    end
+
     def build
       terminology = OM::XML::Terminology.new(:schema=>@schema, :namespaces=>@namespaces)
       root_term_builders.each do |root_term_builder|

data/lib/tasks/om.rake

@@ -1,22 +1,28 @@
 desc "Task to execute builds on a Hudson Continuous Integration Server."
 task :hudson do
   Rake::Task["om:doc"].invoke
-  Rake::Task["om:rcov"].invoke
-  Rake::Task["om:rspec"].invoke
+  Rake::Task["coverage"].invoke
+end
+
+
+desc "Execute specs with coverage"
+task :coverage do 
+  # Put spec opts in a file named .rspec in root
+  ruby_engine = defined?(RUBY_ENGINE) ? RUBY_ENGINE : "ruby"
+  ENV['COVERAGE'] = 'true' unless ruby_engine == 'jruby'
+
+
+  Rake::Task['om:rspec'].invoke
 end
 
 namespace :om do    
 
   require 'rspec/core/rake_task'
   RSpec::Core::RakeTask.new(:rspec) do |spec|
-      spec.pattern = FileList['spec/**/*_spec.rb']
-        spec.pattern += FileList['spec/*_spec.rb']
-  end
-
-  RSpec::Core::RakeTask.new(:rcov) do |spec|
-     spec.pattern = FileList['spec/**/*_spec.rb']
-     spec.pattern += FileList['spec/*_spec.rb']
-     spec.rcov = true
+    if ENV['COVERAGE'] and RUBY_VERSION =~ /^1.8/
+      spec.rcov = true
+      spec.rcov_opts = %w{-I../../app -I../../lib --exclude spec\/*,gems\/*,ruby\/* --aggregate coverage.data}
+    end
   end
 
   # Use yard to build docs

data/om.gemspec

@@ -15,6 +15,7 @@ Gem::Specification.new do |s|
   s.add_dependency('nokogiri', ">= 1.4.2")
   s.add_dependency('mediashelf-loggable')
   s.add_development_dependency "rspec", "~> 2.0"
+  s.add_development_dependency "rake"
   s.add_development_dependency "mocha", ">= 0.9.8"
   s.add_development_dependency "equivalent-xml", ">= 0.2.4"
 

data/spec/fixtures/mods_articles/hydrangea_article1.xml

@@ -37,6 +37,13 @@
         </role>
     </name>
 
+    <name type="corporate">
+        <namePart>NSF</namePart>
+        <role> 
+            <roleTerm authority="marcrelator" type="text">funder</roleTerm>
+        </role>
+    </name>
+
     <typeOfResource>text</typeOfResource>
     <genre authority="local">journal article</genre>
 

data/spec/integration/differentiated_elements_spec.rb

@@ -0,0 +1,39 @@
+require 'spec_helper'
+
+describe "use the root element as a member of the proxy address" do
+  before(:all) do
+    class BlahTerminology
+      include OM::XML::Document
+
+      set_terminology do |t|
+        t.root :path => 'root_element', :xmlns => "asdf"
+        t.relevant_container do
+          t.the_thing_we_want
+        end
+
+        t.the_thing_we_want :proxy => [:root_element, :relevant_container, :the_thing_we_want]
+      end
+    end
+  end
+
+  subject do
+    BlahTerminology.from_xml('<root_element xmlns="asdf">
+                                <arbitrary_container_element>
+                                  <relevant_container>
+                                    <the_thing_we_want but="not really">1</the_thing_we_want>
+                                  </relevant_container>
+                                </arbitrary_container_element>
+                                <relevant_container>
+                                  <the_thing_we_want>2</the_thing_we_want>
+                                </relevant_container>
+                              </root_element>')
+  end
+
+  it "should pull out all occurences of the_thing_we_want in the relevant_container" do
+    subject.relevant_container.the_thing_we_want.should == ["1", "2"]
+  end
+
+  it "should only pull out the_thing_we_want at the root level" do
+    subject.the_thing_we_want.should == ["2"]
+  end
+end

data/spec/integration/element_value_spec.rb

@@ -0,0 +1,91 @@
+require 'spec_helper'
+
+describe "element values" do
+  before(:all) do
+    class ElementValueTerminology
+      include OM::XML::Document
+
+      set_terminology do |t|
+        t.root(:path => "outer", :xmlns => nil)
+        t.elementA
+        t.elB(:path => "elementB")
+        t.elC(:path => "elementC")
+
+        t.elementC(:attributes=>{:animal=>"seagull"}, :namespace_prefix => nil)
+        t.here(:path=>"resource", :attributes=>{:type=>"ead"}, :namespace_prefix => nil)
+        t.there(:path=>"resource", :attributes=>{:type=>"nowhere"}, :namespace_prefix => nil)
+        t.elementD(:attributes=>{:animal=>:none}, :namespace_prefix => nil)
+        t.no_attrib(:path => "elementB", :attributes=>{:animal=>:none}, :namespace_prefix => nil)
+
+        t.elementB {
+          t.my_attr(:path => {:attribute=>"animal"}, :namespace_prefix => nil)
+        }
+        t.alternate(:path => "elementB/@animal", :namespace_prefix => nil)
+        t.another(:proxy=>[:elementB, :my_attr])
+        t.animal_attrib(:path => {:attribute=>"animal"}, :namespace_prefix => nil)
+      end
+    end
+  end
+
+  subject do
+    ElementValueTerminology.from_xml <<-EOF
+<outer outerId="hypatia:outer" type="outer type">
+  <elementA>valA</elementA>
+  <elementB>valB1</elementB>
+  <elementB animal="vole">valB2</elementB>
+  <elementC type="c type" animal="seagull">valC</elementC>
+  <elementD >valD1</elementC>
+  <elementD animal="seagull">valD2</elementC>
+  <resource type="ead" id="coll.ead" objectId="hypatia:ead_file_asset_fixture">
+    <file id="my_ead.xml" format="XML" mimetype="text/xml" size="47570">
+      <checksum type="md5">123</checksum>
+      <checksum type="sha1">456</checksum>
+    </file>
+  </resource>
+  <resource type="image" id="image" objectId="hypatia:coll_img_file_asset_fixture">
+    <file id="my_image.jpg" format="JPG" mimetype="image/jpeg" size="302080">
+      <checksum type="md5">789</checksum>
+      <checksum type="sha1">666</checksum>
+    </file>
+  </resource>
+</outer>
+EOF
+end
+
+  it "should handle single-element terms correctly" do
+    subject.elementA.should == ["valA"]
+  end
+
+  it "should handle term paths" do
+    subject.elC.should == ["valC"]
+  end
+
+  it "should handle multiple-element, terms with paths correctly" do
+    subject.elB.should == ["valB1", "valB2"]
+  end
+
+  it "should handle terms that require specific attributes" do
+    subject.elementC.should == ["valC"]
+  end
+
+  it "should handle" do
+    subject.here.length.should == 1
+    subject.here.first.split(/\W/).should include('123', '456')
+  end
+
+  it "should handle missing terms" do
+    subject.there.should be_empty
+  end
+
+  it "should handle element  value given the absence of a specific attribute" do
+    subject.elementD.should == ["valD1"]
+    subject.no_attrib.should == ["valB1"]
+  end
+
+  it "should handle OM terms for an attribute value" do
+    subject.elementB.my_attr.should == ["vole"]
+    subject.alternate.should == ["vole"]
+    subject.another.should == ["vole"]
+    subject.animal_attrib.should include("vole", "seagull")
+  end
+end

data/spec/integration/proxies_and_ref_spec.rb

@@ -0,0 +1,81 @@
+require 'spec_helper'
+
+describe "an example with :proxy and :ref" do
+  before(:all) do
+    class ExampleProxyAndRefTerminology
+      include OM::XML::Document
+
+      set_terminology do |t|
+        t.root(:path => "outer", :xmlns => '', :namespace_prefix => nil)
+    
+        t.resource(:namespace_prefix => nil) {
+          t.fedora_pid(:path=>{:attribute=>"objectId"}, :namespace_prefix => nil)
+          t.file(:ref=>[:file], :namespace_prefix => nil, :namespace_prefix => nil)
+        }
+        t.file(:namespace_prefix => nil) {
+          t.ds_label(:path=>{:attribute=>"id"}, :namespace_prefix => nil)
+          t.size(:path=>{:attribute=>"size"}, :namespace_prefix => nil)
+          t.md5(:path=>"checksum", :attributes=>{:type=>"md5"}, :namespace_prefix => nil)
+          t.sha1(:path=>"checksum", :attributes=>{:type=>"sha1"}, :namespace_prefix => nil)
+        }
+        #  really want ead where the type is ead and the file format is XML and the file mimetype is text/xml (and the file id is (?coll_ead.xml ... can be whatever the label of the DS is in the FileAsset object)) 
+        t.ead(:ref=>:resource, :attributes=>{:type=>"ead"}) 
+        t.image(:ref=>:resource, :attributes=>{:type=>"image"})
+    
+        t.ead_fedora_pid(:proxy=>[:ead, :fedora_pid])
+        t.ead_ds_label(:proxy=>[:ead, :file, :ds_label])
+        t.ead_size(:proxy=>[:ead, :file, :size])
+        t.ead_md5(:proxy=>[:ead, :file, :md5])
+        t.ead_sha1(:proxy=>[:ead, :file, :sha1])
+    
+        t.image_fedora_pid(:proxy=>[:image, :fedora_pid])
+        t.image_ds_label(:proxy=>[:image, :file, :ds_label])
+        t.image_size(:proxy=>[:image, :file, :size])
+        t.image_md5(:proxy=>[:image, :file, :md5])
+        t.image_sha1(:proxy=>[:image, :file, :sha1])
+        end
+    end
+  end
+
+  subject do
+    ExampleProxyAndRefTerminology.from_xml <<-EOF
+<outer outerId="hypatia:outer" type="outer type">
+  <resource type="ead" id="coll.ead" objectId="hypatia:ead_file_asset_fixture">
+    <file id="my_ead.xml" format="XML" mimetype="text/xml" size="47570">
+      <checksum type="md5">123</checksum>
+      <checksum type="sha1">456</checksum>
+    </file>
+  </resource>
+  <resource type="image" id="image" objectId="hypatia:coll_img_file_asset_fixture">
+    <file id="my_image.jpg" format="JPG" mimetype="image/jpeg" size="302080">
+      <checksum type="md5">789</checksum>
+      <checksum type="sha1">666</checksum>
+    </file>
+  </resource>
+</outer>
+
+EOF
+end
+
+  describe "image" do
+    it "should have the right proxy terms" do
+      subject.ead_fedora_pid.should include "hypatia:ead_file_asset_fixture"
+      subject.ead_ds_label.should include "my_ead.xml"
+      subject.ead_size.should include "47570"
+      subject.ead_md5.should include "123"
+      subject.ead_sha1.should include "456"
+    end
+  end
+
+  describe "ead" do
+    it "should have the right proxy terms" do
+      subject.image_fedora_pid.should include "hypatia:coll_img_file_asset_fixture"
+      subject.image_ds_label.should include "my_image.jpg"
+      subject.image_size.should include "302080"
+      subject.image_md5.should include "789"
+      subject.image_sha1.should include "666"
+    end
+
+  end
+
+end

data/spec/integration/querying_documents_spec.rb

@@ -0,0 +1,73 @@
+require 'spec_helper'
+require "om/samples"
+
+describe "Rspec tests for QUERYING_DOCUMENTS.textile" do
+
+  before(:all) do
+    @xml_file = "mods_articles/hydrangea_article1.xml"
+    @doc      = OM::Samples::ModsArticle.from_xml(fixture @xml_file) { |conf|
+      conf.default_xml.noblanks
+    }
+    @term = OM::Samples::ModsArticle.terminology
+  end
+
+  it "xpath_for()" do
+    @term.xpath_for(:name).should == 
+      '//oxns:name'
+    @term.xpath_for(:person).should == 
+      '//oxns:name[@type="personal"]'
+    @term.xpath_for(:organization).should == 
+      '//oxns:name[@type="corporate"]'
+    @term.xpath_for(:person, :first_name).should ==
+      '//oxns:name[@type="personal"]/oxns:namePart[@type="given"]'
+    @term.xpath_for(:journal, :issue, :pages, :start).should ==
+      '//oxns:relatedItem[@type="host"]/oxns:part/oxns:extent[@unit="pages"]/oxns:start'
+  end
+
+  it "term_values()" do
+    @doc.term_values(:person, :first_name).should == ["GIVEN NAMES", "Siddartha"]
+    @doc.term_values(:person, :last_name).should == ["FAMILY NAME", "Gautama"]
+    @doc.term_values(:organization, :namePart).should == ['NSF']
+    @doc.term_values(:journal, :issue, :pages, :start).should == ['195']
+    @doc.term_values(:journal, :title_info, :main_title).should == ["TITLE OF HOST JOURNAL"]
+  end
+
+  it "xpath_for(): relative vs absolute" do
+    xp_rel = '//oxns:titleInfo/oxns:title'
+    xp_abs = '//oxns:mods/oxns:titleInfo/oxns:title'
+    @term.xpath_for(       :title_info, :main_title).should == xp_rel
+    @term.xpath_for(:mods, :title_info, :main_title).should == xp_abs
+  end
+
+  it "term_values(): relative vs absolute" do
+    exp = [
+      "ARTICLE TITLE HYDRANGEA ARTICLE 1",
+      "Artikkelin otsikko Hydrangea artiklan 1",
+      "TITLE OF HOST JOURNAL",
+    ]
+    xp_abs = '//oxns:mods/oxns:titleInfo/oxns:title'
+    @doc.term_values(       :title_info, :main_title).should == exp
+    @doc.term_values(:mods, :title_info, :main_title).should == exp[0..1]
+  end
+
+  it "find_by_terms()" do
+    exp_xml_role  = '<role><roleTerm authority="marcrelator" type="text">funder</roleTerm></role>'
+    exp_xml_start = '<start>195</start>'
+    @doc.find_by_terms(:organization, :role).class.should == Nokogiri::XML::NodeSet
+    @doc.find_by_terms(:organization, :role).to_xml.should be_equivalent_to exp_xml_role
+    @doc.find_by_terms(:journal, :issue, :pages, :start).to_xml.should == exp_xml_start
+  end
+
+  it "find_by_terms() error" do
+    exp_err = OM::XML::Terminology::BadPointerError
+    expect { @doc.find_by_terms :journal, :issue, :BLAH, :start }.to raise_error exp_err
+  end
+
+  it "proxies" do
+    @term.xpath_for(:title).should ==
+      '//oxns:titleInfo/oxns:title'
+    @term.xpath_for(:journal_title)
+      '//oxns:relatedItem[@type="host"]/oxns:titleInfo/oxns:title'
+  end
+
+end