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

.travis.yml

@@ -1,5 +1,5 @@
 rvm:
-  - 1.9.3-p448
+  - 1.9.3-p484
   - 2.0.0-p353
 before_install:
   - sudo apt-get update

lib/dwc-archive.rb

@@ -8,6 +8,7 @@
 require 'digest'
 require 'csv'
 require 'logger'
+require 'nokogiri'
 require_relative 'dwc-archive/xml_reader'
 require_relative 'dwc-archive/ingester'
 require_relative 'dwc-archive/errors'

lib/dwc-archive/archive.rb

@@ -1,7 +1,7 @@
-require 'nokogiri'
 class DarwinCore
   class Archive
     attr_reader :meta, :eml
+
     def initialize(archive_path, tmp_dir)
       @archive_path = archive_path
       @tmp_dir = tmp_dir

lib/dwc-archive/classification_normalizer.rb

@@ -72,21 +72,11 @@ def add_vernacular_name_string(name_string)
     end
 
     def name_strings(opts = {})
-      opts = { with_hash: false }.merge(opts)
-      if !!opts[:with_hash]
-        @name_strings
-      else
-        @name_strings.keys
-      end
+      process_strings(@name_strings, opts)
     end
 
     def vernacular_name_strings(opts = {})
-      opts = { with_hash: false }.merge(opts)
-      if !!opts[:with_hash]
-        @vernacular_name_strings
-      else
-        @vernacular_name_strings.keys
-      end
+      process_strings(@vernacular_name_strings, opts)
     end
 
     def normalize(opts = {})
@@ -110,6 +100,15 @@ def normalize(opts = {})
 
   private
 
+    def process_strings(strings, opts)
+      opts = { with_hash: false }.merge(opts)
+      if !!opts[:with_hash]
+        strings
+      else
+        strings.keys
+      end
+    end
+
     def get_canonical_name(a_scientific_name)
       if @with_canonical_names
         canonical_name = @parser.parse(a_scientific_name,

lib/dwc-archive/core.rb

@@ -2,16 +2,18 @@ class DarwinCore
   class Core
     include DarwinCore::Ingester
     attr_reader :id 
+
     def initialize(dwc)
       @dwc = dwc
       @archive = @dwc.archive
       @path = @archive.files_path
       root_key = @archive.meta.keys[0]
       @data = @archive.meta[root_key][:core]
       raise DarwinCore::CoreFileError.
-        new("Cannot find core in meta.xml, is meta.xml valid?") unless @data
+        new('Cannot find core in meta.xml, is meta.xml valid?') unless @data
       @id = @data[:id][:attributes] 
       get_attributes(DarwinCore::CoreFileError)
     end
+
   end
 end

lib/dwc-archive/expander.rb

@@ -1,5 +1,6 @@
 class DarwinCore
   class Expander
+
     def initialize(archive_path, tmp_dir)
       @archive_path = archive_path
       @tmp_dir = tmp_dir

lib/dwc-archive/generator.rb

@@ -21,12 +21,7 @@ def clean
     def add_core(data, file_name, keep_headers = true)
       c = CSV.open(File.join(@path,file_name), @write)
       header = data.shift
-      fields = header.map do |f|
-        f.strip!
-        err = 'No header in core data, or header fields are not urls'
-        raise DarwinCore::GeneratorError.new(err) unless f.match(/^http:\/\//)
-        f.split('/')[-1]
-      end
+      fields = get_fields(header, 'core') 
       data.unshift(fields) if keep_headers
       ignore_header_lines = keep_headers ? 1 : 0
       @meta_xml_data[:core] = { fields: header, 
@@ -41,12 +36,7 @@ def add_extension(data, file_name,
                       row_type = 'http://rs.tdwg.org/dwc/terms/Taxon')
       c = CSV.open(File.join(@path,file_name), @write)
       header = data.shift
-      fields = header.map do |f|
-        f.strip!
-        err = 'No header in core data, or header fields are not urls'
-        raise DarwinCore::GeneratorError.new(err) unless f.match(/^http:\/\//)
-        f.split('/')[-1]
-      end
+      fields = get_fields(header, 'extension')
       data.unshift(fields) if keep_headers
       ignore_header_lines = keep_headers ? 1 : 0
       @meta_xml_data[:extensions] << { fields: header, 
@@ -81,5 +71,16 @@ def pack
       a = "cd #{@path}; tar -zcf #{@dwc_path} *"
       system(a)
     end
+
+    private
+
+    def get_fields(header, file_type)
+      header.map do |f|
+        f.strip!
+        err = "No header in %s data, or header fields are not urls" % file_type
+        raise DarwinCore::GeneratorError.new(err) unless f.match(/^http:\/\//)
+        f.split('/')[-1]
+      end
+    end
   end
 end

lib/dwc-archive/generator_eml_xml.rb

@@ -22,51 +22,9 @@ def create
             :'xmlns:res' =>        'eml://ecoinformatics.org/resource-2.1.1',
             :'xmlns:dc' =>         'http://purl.org/dc/terms/',
             :'xmlns:xsi' =>        'http://www.w3.org/2001/XMLSchema-instance',
-            :'xsi:schemaLocation' => 'eml_uri')  do
-            xml.dataset(id: @data[:id]) do
-              xml.title(@data[:title])
-              xml.license(@data[:license])
-              contacts = []
-              @data[:authors].each_with_index do |a, i|
-                creator_id = i + 1
-                contacts << creator_id
-                xml.creator(id: creator_id, scope: 'document') do
-                  xml.individualName do
-                    xml.givenName(a[:first_name])
-                    xml.surName(a[:last_name])
-                  end
-                  xml.organizationName(a[:organization]) if a[:organization]
-                  xml.positionName(a[:position]) if a[:position]
-                  xml.onlineUrl(a[:url]) if a[:url]
-                  xml.electronicMailAddress(a[:email])
-                end
-              end
-              @data[:metadata_providers].each_with_index do |a, i|
-                xml.metadataProvider do
-                  xml.individualName do
-                    xml.givenName(a[:first_name])
-                    xml.surName(a[:last_name])
-                  end
-                  xml.organizationName(a[:organization]) if a[:organization]
-                  xml.positionName(a[:position]) if a[:position]
-                  xml.onlineUrl(a[:url]) if a[:url]
-                  xml.electronicMailAddress(a[:email])
-                end
-              end if @data[:metadata_providers]
-              xml.pubDate(Time.now.to_s)
-              xml.abstract() do
-                xml.para(@data[:abstract])
-              end
-              contacts.each do |contact|
-                xml.contact { xml.references(contact) }
-              end
-            end
-            xml.additionalMetadata do
-              xml.metadata do
-                xml.citation(@data[:citation])
-                xml.resourceLogoUrl(@data[:logo_url]) if @data[:logo_url]
-              end
-            end
+            :'xsi:schemaLocation' => 'eml_uri') do
+            build_dataset(xml)
+            build_additional_metadata(xml)
             xml.parent.namespace = xml.parent.namespace_definitions.first
           end
         end
@@ -77,6 +35,71 @@ def create
       end
 
       private
+
+      def build_dataset(xml)
+        xml.dataset(id: @data[:id]) do
+          xml.title(@data[:title])
+          xml.license(@data[:license])
+          contacts = []
+          build_authors(xml, contacts)
+          build_metadata_providers(xml)
+          xml.pubDate(Time.now.to_s)
+          build_abstract(xml)
+          build_contacts(xml, contacts)
+        end
+      end
+
+      def build_abstract(xml)
+        xml.abstract() do
+          xml.para(@data[:abstract])
+        end
+      end
+
+      def build_contacts(xml, contacts)
+        contacts.each do |contact|
+          xml.contact { xml.references(contact) }
+        end
+      end
+
+      def build_metadata_providers(xml)
+        @data[:metadata_providers].each_with_index do |a, i|
+          xml.metadataProvider do
+            build_person(xml, a)
+          end
+        end if @data[:metadata_providers]
+      end
+
+      def build_authors(xml, contacts)
+        @data[:authors].each_with_index do |a, i|
+          creator_id = i + 1
+          contacts << creator_id
+          xml.creator(id: creator_id, scope: 'document') do
+            build_person(xml, a)
+          end
+        end
+      end
+
+      def build_additional_metadata(xml)
+        xml.additionalMetadata do
+          xml.metadata do
+            xml.citation(@data[:citation])
+            xml.resourceLogoUrl(@data[:logo_url]) if @data[:logo_url]
+          end
+        end
+      end
+
+      def build_person(xml, data)
+        a = data
+        xml.individualName do
+          xml.givenName(a[:first_name])
+          xml.surName(a[:last_name])
+        end
+        xml.organizationName(a[:organization]) if a[:organization]
+        xml.positionName(a[:position]) if a[:position]
+        xml.onlineUrl(a[:url]) if a[:url]
+        xml.electronicMailAddress(a[:email])
+      end
+
       def timestamp
         t = Time.now.getutc.to_a[0..5].reverse
         t[0..2].join('-') + '::' + t[-3..-1].join(':')

lib/dwc-archive/generator_meta_xml.rb

@@ -16,26 +16,7 @@ def create
                    fieldsEnclosedBy: '"', 
                    linesTerminatedBy: "\n", 
                    rowType: 'http://rs.tdwg.org/dwc/terms/Taxon' }
-          xml.archive(xmlns: 'http://rs.tdwg.org/dwc/text/',
-            :'xmlns:xsi' => 'http://www.w3.org/2001/XMLSchema-instance',
-            :'xsi:schemaLocation' => schema_uri) do
-            xml.core(opts.merge(ignoreHeaderLines: 
-                                  @data[:core][:ignoreHeaderLines])) do
-              xml.files { xml.location(@data[:core][:location]) }
-              taxon_id, fields = find_taxon_id(@data[:core][:fields])
-              xml.id_(index: taxon_id[1])
-              fields.each { |f| xml.field(term: f[0], index: f[1]) }
-            end
-            @data[:extensions].each do |e|
-              xml.extension(opts.merge(ignoreHeaderLines: e[:ignoreHeaderLines], 
-                                       rowType: e[:rowType])) do
-                xml.files { xml.location(e[:location]) }
-                taxon_id, fields = find_taxon_id(e[:fields])
-                xml.coreid(index: taxon_id[1])
-                fields.each { |f| xml.field(term: f[0], index: f[1]) }
-              end
-            end
-          end
+          build_archive(xml, opts, schema_uri)
         end
         meta_xml_data = builder.to_xml
         meta_file = open(File.join(@path, 'meta.xml'), @write)
@@ -44,6 +25,38 @@ def create
       end
 
       private
+
+      def build_archive(xml, opts, schema_uri)
+        xml.archive(xmlns: 'http://rs.tdwg.org/dwc/text/',
+          :'xmlns:xsi' => 'http://www.w3.org/2001/XMLSchema-instance',
+          :'xsi:schemaLocation' => schema_uri) do
+          build_core(xml, opts)
+          build_extensions(xml, opts)
+        end
+      end
+
+      def build_core(xml, opts)
+        xml.core(opts.merge(ignoreHeaderLines: 
+                              @data[:core][:ignoreHeaderLines])) do
+          xml.files { xml.location(@data[:core][:location]) }
+          taxon_id, fields = find_taxon_id(@data[:core][:fields])
+          xml.id_(index: taxon_id[1])
+          fields.each { |f| xml.field(term: f[0], index: f[1]) }
+        end
+      end
+
+      def build_extensions(xml, opts)
+        @data[:extensions].each do |e|
+          xml.extension(opts.merge(ignoreHeaderLines: e[:ignoreHeaderLines], 
+                                   rowType: e[:rowType])) do
+            xml.files { xml.location(e[:location]) }
+            taxon_id, fields = find_taxon_id(e[:fields])
+            xml.coreid(index: taxon_id[1])
+            fields.each { |f| xml.field(term: f[0], index: f[1]) }
+          end
+        end
+      end
+
       def find_taxon_id(data)
         fields = []
         data.each_with_index { |f, i| fields << [f.strip, i] }