Merge branch 'master' of github.com:GlobalNamesArchitecture/dwc-archive

CHANGELOG

@@ -1,3 +1,5 @@
+0.9.7 Refactoring and tests improvements
+
 0.9.6 Added support for GNUB DwCA files
 
 0.9.4 Gem dependencies updated, added travis support

lib/dwc-archive.rb

@@ -26,12 +26,12 @@
 class DarwinCore
 
   VERSION = DarwinCore::VERSION
+  DEFAULT_TMP_DIR = "/tmp"
 
   attr_reader :archive, :core, :metadata, :extensions, 
     :classification_normalizer
   alias :eml :metadata
 
-  DEFAULT_TMP_DIR = "/tmp"
 
   def self.nil_field?(field)
     return true if [nil, '', '/N'].include?(field)

lib/dwc-archive/version.rb

@@ -1,3 +1,3 @@
 class DarwinCore
-  VERSION = "0.9.6"
+  VERSION = "0.9.10"
 end

spec/lib/generator_eml_xml_spec.rb

@@ -0,0 +1,21 @@
+require_relative '../spec_helper'
+
+describe DarwinCore::Generator::EmlXml do
+  subject(:eml) { DarwinCore::Generator::EmlXml.new(data, path) }
+  let(:data) { EML_DATA }
+  let(:path) { DarwinCore::DEFAULT_TMP_DIR }
+
+  describe '.new' do
+    it 'initializes generator' do
+      expect(eml).to be_kind_of DarwinCore::Generator::EmlXml
+    end
+  end
+
+  describe '#create' do
+    it 'should create eml xml' do
+      eml.create
+      eml_xml = File.read(File.join(path, 'eml.xml'))
+      expect(eml_xml).to match /Test Classification/
+    end
+  end
+end

spec/lib/generator_meta_xml_spec.rb

@@ -0,0 +1,21 @@
+require_relative '../spec_helper'
+
+describe DarwinCore::Generator::MetaXml do
+  subject(:meta) { DarwinCore::Generator::MetaXml.new(data, path) }
+  let(:data) { META_DATA }
+  let(:path) { DarwinCore::DEFAULT_TMP_DIR }
+
+  describe '.new' do
+    it 'initializes' do
+      expect(meta).to be_kind_of DarwinCore::Generator::MetaXml
+    end
+  end
+
+  describe '#create' do
+    it 'creates metadata file' do
+      meta.create
+      meta = File.read(File.join(path, 'meta.xml'))
+      expect(meta).to match %r|<location>core.csv</location>|  
+    end
+  end
+end

spec/lib/generator_spec.rb

@@ -0,0 +1,116 @@
+# encoding: utf-8
+require_relative '../spec_helper'
+
+describe DarwinCore::Generator do
+  subject(:gen) { DarwinCore::Generator.new(dwc_path, tmp_dir) }
+  let(:tmp_dir) { DarwinCore::DEFAULT_TMP_DIR }
+  let(:dwc_path) { File.join(tmp_dir, 'spec_dwca.tar.gz') } 
+
+  describe '.new' do
+    it 'initializes empty DwCA' do
+      expect(gen).to be_kind_of DarwinCore::Generator      
+    end
+  end
+
+  describe '#add_core' do
+    it 'adds core to DwCA instance' do
+      gen.add_core(CORE_DATA.dup, 'core.csv', true)
+      core = File.read(File.join(gen.path, 'core.csv'))
+      expect(core).to match /taxonID,parentNameUsageID,scientificName/
+    end
+
+    context 'urls are not given in header' do
+      it 'raises error' do
+        data = CORE_DATA.dup
+        data[0] = data[0].map { |f| f.split('/')[-1] }
+        expect { gen.add_core(data, 'core.csv', true) }.
+                              to raise_error DarwinCore::GeneratorError
+      end
+    end
+  end
+
+  describe '#add_extension' do
+    it 'adds extension to DwCA instance' do
+      gen.add_extension(EXTENSION_DATA.dup, 
+                        'vern.csv', 
+                        true, 
+                        'http://rs.gbif.org/terms/1.0/VernacularName')
+      extension = File.read(File.join(gen.path, 'vern.csv'))
+
+      expect(extension).to match /Береза/
+    end
+  end
+
+  describe '#add_meta_xml' do
+    it 'creates metadata for DwCA' do
+      gen.add_core(CORE_DATA.dup, 'core.csv', true)
+      gen.add_extension(EXTENSION_DATA.dup, 
+                        'vern.csv', 
+                        true, 
+                        'http://rs.gbif.org/terms/1.0/VernacularName')
+
+      gen.add_meta_xml
+      meta = File.read(File.join(gen.path, 'meta.xml'))
+      expect(meta).to match %r|<location>core.csv</location>|
+    end
+  end
+
+  describe '#add_eml_data' do
+    it 'adds eml data' do
+      gen.add_eml_xml(EML_DATA)
+      eml = File.read(File.join(gen.path, 'eml.xml'))
+      expect(eml).to match /jdoe@example.com/
+    end
+  end
+
+  describe '#path' do
+    it 'returns temporary path for assembling DwCA' do
+      expect(gen.path).to match /dwc_\d+$/
+    end
+  end
+
+  describe '#files' do
+    it 'returns created files' do
+      gen.add_core(CORE_DATA.dup, 'core.csv', true)
+      gen.add_extension(EXTENSION_DATA.dup, 
+                        'vern.csv', 
+                        true, 
+                        'http://rs.gbif.org/terms/1.0/VernacularName')
+
+      gen.add_meta_xml
+      expect(gen.files).to match_array ['core.csv', 'meta.xml', 'vern.csv']  
+    end
+  end
+
+  describe '#pack' do
+    it 'creates final DwCA file' do
+      FileUtils.rm dwc_path if File.exists?(dwc_path)
+      gen.add_core(CORE_DATA.dup, 'core.csv', true)
+      gen.add_extension(EXTENSION_DATA.dup, 
+                        'vern.csv', 
+                        true, 
+                        'http://rs.gbif.org/terms/1.0/VernacularName')
+
+      gen.add_meta_xml
+      gen.add_eml_xml(EML_DATA)
+      gen.pack
+      expect(File.exists?(dwc_path)).to be_true 
+    end
+  end
+
+  describe '#clean' do
+    it 'removes temporary directory for DwCA' do
+      gen.add_eml_xml(EML_DATA)
+      expect(File.exists?(gen.path)).to be true
+      gen.clean
+      expect(File.exists?(gen.path)).to be false
+    end
+  end
+
+  describe '#eml_xml_data' do
+    it 'returns current eml data' do
+      expect(gen.eml_xml_data).to be_kind_of Hash
+    end
+  end
+
+end

spec/spec_helper.rb

@@ -1,3 +1,4 @@
+# encoding: utf-8
 # HACK: to suppress warnings
 $VERBOSE = nil
 
@@ -12,3 +13,66 @@
 RSpec.configure do |config|
 end
 
+unless defined?(SPEC_CONSTANTS)
+  EML_DATA = {
+      id: '1234',
+      license: 'http://creativecommons.org/licenses/by-sa/3.0/',
+      title: 'Test Classification',
+      authors: [
+        { first_name: 'John',
+          last_name: 'Doe',
+          email: 'jdoe@example.com',
+          organization: 'Example',
+          position: 'Assistant Professor',
+          url: 'http://example.org' },
+          { first_name: 'Jane',
+            last_name: 'Doe',
+            email: 'jane@example.com' }
+    ],
+      metadata_providers: [
+        { first_name: 'Jim',
+          last_name: 'Doe',
+          email: 'jimdoe@example.com',
+          url: 'http://aggregator.example.org' }],
+      abstract: 'test classification',
+      citation:
+        'Test classification: Doe John, Doe Jane, Taxnonmy, 10, 1, 2010',
+      url: 'http://example.com'
+    }
+  META_DATA = {
+    extensions: [
+      {fields: [
+        'http://rs.tdwg.org/dwc/terms/TaxonID', 
+        'http://rs.tdwg.org/dwc/terms/vernacularName'], 
+        ignoreHeaderLines: 1, 
+        location: 'vern.csv', 
+        rowType: 'http://rs.gbif.org/terms/1.0/VernacularName'}], 
+        core: {
+          fields: [
+            'http://rs.tdwg.org/dwc/terms/taxonID', 
+            'http://rs.tdwg.org/dwc/terms/parentNameUsageID', 
+            'http://rs.tdwg.org/dwc/terms/scientificName', 
+            'http://rs.tdwg.org/dwc/terms/taxonRank'], 
+            ignoreHeaderLines: 1, 
+            location: 'core.csv'}}
+  CORE_DATA = [
+    ["http://rs.tdwg.org/dwc/terms/taxonID",
+     "http://rs.tdwg.org/dwc/terms/parentNameUsageID",
+     "http://rs.tdwg.org/dwc/terms/scientificName",
+     "http://rs.tdwg.org/dwc/terms/taxonRank"],
+    [1, 0, "Plantae", "kingdom"],
+    [2, 1, "Betula", "genus"],
+    [3, 2, "Betula verucosa", "species"]
+  ] 
+  EXTENSION_DATA = [
+    ["http://rs.tdwg.org/dwc/terms/TaxonID",
+    "http://rs.tdwg.org/dwc/terms/vernacularName"],
+    [1, "Plants"],
+    [1, "Растения"],
+    [2, "Birch"],
+    [2, "Береза"],
+    [3, "Wheeping Birch"],
+    [3, "Береза плакучая"]
+  ]
+  SPEC_CONSTANTS = true
+end