relaton-bib 1.13.7 → 1.13.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 493716815449edfed921ea2239b0c066b614211c2722f261bf4f59d5aa094e54
-  data.tar.gz: 661ad28bf88fcb3e5b27888885696c14385fbb9bfc891f4b20ea91d7e6dba1f1
+  metadata.gz: 50d370dad5e6a20ef5eaaed6633e0d62d80c0087f8a24d205dcb02723c025b35
+  data.tar.gz: 7ef8d3f6530096ed1e15bc66f148a3972490e39ba3c51bfc894a31cab8195d4f
 SHA512:
-  metadata.gz: 2e5ef5a5bd17eafad776a09434fb8e7df82ec2ca3b4fc1e59a6e7d722a72e1c1f57ec75a80dfb1c651a7cea71160c04fb1083517169a68ba341bc3f607c8219c
-  data.tar.gz: 7b556666124417d87dc58709325b4be7970fb4af443686e332376b739edaa8ad0041e0a26145954577a2b1f8674b078a31405ac2447868caa8df06b3eb25158c
+  metadata.gz: 305a89bc3529accd1087f9e7928172bacf6acb4672fa0215ee1b27406cca08cda484470227625c94928f5be4ac0766cd0b13b43059c0328b3f9d0ba03f6ad0af
+  data.tar.gz: 44e3711a83e329cba17baabbaacecc03736eef8c20857aac71c760bcfd7289071b1d2a60b70d0d9710d132c02c0607c0ad234522ee280b6867f1bf7fba69bed1

data/README.adoc

@@ -103,7 +103,7 @@ item.to_xml
     </bibitem>"
 ----
 
-Default root element is `bibitem`. With argument `bibdata: true` the XML wrapped with `bibdata` element.
+The default root element is `bibitem`. With argument `bibdata: true` the XML wrapped with `bibdata` element.
 
 [source,ruby]
 ----
@@ -117,7 +117,7 @@ item.to_xml bibdata: true
 
 ==== Date format
 
-By default date elements are formatted as year (yyyy). Option `:date_format` allows to output date elements in `:short` (yyyy-mm) and `:full` (yyyy-mm-dd) additiona formats.
+By default date elements are formatted as a year (yyyy). Option `:date_format` allows to output date elements in `:short` (yyyy-mm) and `:full` (yyyy-mm-dd) additional formats.
 
 [source,ruby]
 ----
@@ -249,7 +249,7 @@ id:: ISOTC211
 fetched:: 2020-08-19
 title::
 title.type:: title-main
-title.conten:: Geographic information
+title.content:: Geographic information
 title.format:: text/plain
 ...
 ----

data/grammars/biblio.rng

@@ -854,6 +854,12 @@
     <optional>
       <attribute name="type"/>
     </optional>
+    <optional>
+      <attribute name="language"/>
+    </optional>
+    <optional>
+      <attribute name="script"/>
+    </optional>
     <data type="anyURI"/>
   </define>
   <define name="DateType">
@@ -882,6 +888,7 @@
       <value>vote-started</value>
       <value>vote-ended</value>
       <value>announced</value>
+      <value>stable-until</value>
     </choice>
   </define>
   <define name="bdate">

data/lib/relaton_bib/bibxml_parser.rb

@@ -224,6 +224,13 @@ module RelatonBib
       end.compact
     end
 
+    #
+    # Parse organization from author element
+    #
+    # @param [Nokogiri::XML::Element] contrib author element
+    #
+    # @return [Hash] contribution info
+    #
     def full_name_org(contrib)
       return unless FULLNAMEORG.include? contrib[:fullname]
 

data/lib/relaton_bib/formatted_string.rb

@@ -21,8 +21,8 @@ module RelatonBib
       #   raise ArgumentError, %{Format "#{format}" is invalid.}
       # end
 
-      super(content, language, script)
       @format = format
+      super(content, language, script)
     end
 
     # @param builder [Nokogiri::XML::Builder]
@@ -107,5 +107,22 @@ module RelatonBib
       out += "#{pref}format:: #{format}\n" if format
       out
     end
+
+    #
+    # Remove HTML tags except <em>, <strong>, <stem>, <sup>, <sub>, <tt>, <br>, <p>.
+    # Replace <i> with <em>, <b> with <strong>.
+    #
+    # @param [String] str content
+    #
+    # @return [String] cleaned content
+    #
+    def cleanup(str)
+      return str unless format == "text/html"
+
+      str.gsub(/<i>/, "<em>").gsub(/<\/i>/, "</em>")
+        .gsub(/<b>/, "<strong>").gsub(/<\/b>/, "</strong>")
+        .gsub(/<(?!\/?(em|strong|stem|sup|sub|tt|br\s?\/|p))[^\s!]\/?.*?>/, "")
+        # .gsub(/<\w+\s.+?>/, "")
+    end
   end
 end

data/lib/relaton_bib/localized_string.rb

@@ -33,7 +33,7 @@ module RelatonBib
                      else c
                      end
                    end
-                 else content
+                 else cleanup content
                  end
     end
 
@@ -127,5 +127,16 @@ module RelatonBib
       else count > 1 ? "#{prefix}::\n" : ""
       end
     end
+
+    #
+    # Should be implemented in subclass.
+    #
+    # @param [String] str content
+    #
+    # @return [String] cleaned content
+    #
+    def cleanup(str)
+      str
+    end
   end
 end

data/lib/relaton_bib/typed_title_string.rb

@@ -101,6 +101,7 @@ module RelatonBib
     # @param content [String]
     # @param language [String]
     # @param script [String]
+    # @param format [String]
     def initialize(**args) # rubocop:disable Metrics/MethodLength
       unless args[:title] || args[:content]
         raise ArgumentError, %{Keyword "title" or "content" should be passed.}
@@ -108,24 +109,35 @@ module RelatonBib
 
       @type = args[:type]
 
-      if args[:title]
-        @title = args[:title]
+      case args[:title]
+      when FormattedString then @title = args[:title]
+      when Hash then @title = FormattedString.new(**args[:title])
       else
         fsargs = args.select { |k, _v| ARGS.include? k }
         @title = FormattedString.new(**fsargs)
       end
     end
 
-    # @param title [String]
-    # @return [TypedTitleStringCollection]
-    def self.from_string(title, lang = nil, script = nil)
+    #
+    # Create TypedTitleStringCollection from string
+    #
+    # @param title [String] title string
+    # @param lang [String, nil] language code Iso639
+    # @param script [String, nil] script code Iso15924
+    # @param format [String] format text/html, text/plain
+    #
+    # @return [TypedTitleStringCollection] collection of TypedTitleString
+    #
+    def self.from_string(title, lang = nil, script = nil, format = "text/plain")
       types = %w[title-intro title-main title-part]
       ttls = split_title(title)
       tts = ttls.map.with_index do |p, i|
-        new type: types[i], content: p, language: lang, script: script if p
+        next unless p
+
+        new type: types[i], content: p, language: lang, script: script, format: format
       end.compact
       tts << new(type: "main", content: ttls.compact.join(" - "),
-                 language: lang, script: script)
+                 language: lang, script: script, format: format)
       TypedTitleStringCollection.new tts
     end
 
@@ -140,10 +152,10 @@ module RelatonBib
     end
 
     # @param ttls [Array<String>]
-    # @return [Array<Strin, nil>]
+    # @return [Array<String, nil>]
     def self.intro_or_part(ttls)
       if /^(Part|Partie) \d+:/.match? ttls[1]
-        [nil, ttls[0], ttls[1..-1].join(" -- ")]
+        [nil, ttls[0], ttls[1..].join(" -- ")]
       else
         parts = ttls.slice(2..-1)
         part = parts.join " -- " if parts.any?
@@ -162,13 +174,6 @@ module RelatonBib
       th = title.to_hash
       return th unless type
 
-      # hash = { "type" => type }
-      # if th.is_a? String
-      #   hash["content"] = th
-      # else
-      #   hash.merge! th
-      # end
-      # hash
       th.merge "type" => type
     end
 
@@ -176,7 +181,7 @@ module RelatonBib
     # @param count [Integer] number of titles
     # @return [String]
     def to_asciibib(prefix = "", count = 1) # rubocop:disable Metrics/AbcSize
-      pref = prefix.empty? ? prefix : prefix + "."
+      pref = prefix.empty? ? prefix : "#{prefix}."
       out = count > 1 ? "#{pref}title::\n" : ""
       out += "#{pref}title.type:: #{type}\n" if type
       out += title.to_asciibib "#{pref}title", 1, !(type.nil? || type.empty?)

data/lib/relaton_bib/typed_uri.rb

@@ -3,15 +3,19 @@ require "addressable"
 module RelatonBib
   # Typed URI
   class TypedUri
-    # @return [Symbol] :src/:obp/:rss
-    attr_reader :type
+    # @return [String, nil]
+    attr_reader :type, :language, :script
     # @retutn [Addressable::URI]
     attr_reader :content
 
-    # @param type [String, NilClass] src/obp/rss
-    # @param content [String]
-    def initialize(content:, type: nil)
+    # @param content [String] URL
+    # @param type [String, nil] src/obp/rss
+    # @param language [String, nil] language code Iso639 (optional) (default: nil)
+    # @param script [String, nil] script code Iso15924 (optional) (default: nil)
+    def initialize(content:, type: nil, language: nil, script: nil)
       @type    = type
+      @language = language
+      @script  = script
       @content = Addressable::URI.parse content if content
     end
 
@@ -24,12 +28,16 @@ module RelatonBib
     def to_xml(builder)
       doc = builder.uri content.to_s
       doc[:type] = type if type
+      doc[:language] = language if language
+      doc[:script] = script if script
     end
 
     # @return [Hash]
     def to_hash
       hash = { "content" => content.to_s }
       hash["type"] = type.to_s if type
+      hash["language"] = language if language
+      hash["script"] = script if script
       hash
     end
 
@@ -41,6 +49,8 @@ module RelatonBib
       out = count > 1 ? "#{pref}::\n" : ""
       out += "#{pref}.type:: #{type}\n" if type
       out += "#{pref}.content:: #{content}\n"
+      out += "#{pref}.language:: #{language}\n" if language
+      out += "#{pref}.script:: #{script}\n" if script
       out
     end
   end

data/lib/relaton_bib/version.rb

@@ -1,3 +1,3 @@
 module RelatonBib
-  VERSION = "1.13.7".freeze
+  VERSION = "1.13.9".freeze
 end

data/lib/relaton_bib/xml_parser.rb

@@ -10,7 +10,7 @@ module RelatonBib
         if bibitem
           bib_item item_data(bibitem)
         else
-          warn "[relaton-bib] WARNING: can't find bibitem or bibdata element "\
+          warn "[relaton-bib] WARNING: can't find bibitem or bibdata element " \
                "in the XML"
         end
       end
@@ -23,8 +23,8 @@ module RelatonBib
       def item_data(bibitem) # rubocop:disable Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
         ext = bibitem.at "//ext"
         {
-          id: bibitem[:id]&.empty? ? nil : bibitem[:id],
-          type: bibitem[:type]&.empty? ? nil : bibitem[:type],
+          id: bibitem[:id].nil? || bibitem[:id].empty? ? nil : bibitem[:id],
+          type: bibitem[:type].nil? || bibitem[:type].empty? ? nil : bibitem[:type],
           fetched: bibitem.at("./fetched")&.text,
           title: fetch_titles(bibitem),
           formattedref: fref(bibitem),
@@ -141,7 +141,7 @@ module RelatonBib
             organization: sr.at("organization")&.text,
             abbreviation: abbreviation, from: sr.at("from")&.text,
             to: sr.at("to")&.text, number: sr.at("number")&.text,
-            partnumber: sr.at("partnumber")&.text, run: sr.at("run")&.text,
+            partnumber: sr.at("partnumber")&.text, run: sr.at("run")&.text
           )
         end
       end
@@ -160,9 +160,6 @@ module RelatonBib
       def fetch_extent(item)
         item.xpath("./extent").reduce([]) do |a, ex|
           a + localities(ex)
-          # Locality.new(
-          #   ex[:type], ex.at("referenceFrom")&.text, ex.at("referenceTo")&.text
-          # )
         end
       end
 
@@ -401,7 +398,8 @@ module RelatonBib
       # @return [Arra<RelatonBib::TypedUri>]
       def fetch_link(item)
         item.xpath("./uri").map do |l|
-          TypedUri.new type: l[:type], content: l.text
+          TypedUri.new(type: l[:type], content: l.text, language: l[:language],
+                       script: l[:script])
         end
       end
 
@@ -412,7 +410,7 @@ module RelatonBib
       def fetch_relations(item, klass = DocumentRelation)
         item.xpath("./relation").map do |rel|
           klass.new(
-            type: rel[:type]&.empty? ? nil : rel[:type],
+            type: rel[:type].nil? || rel[:type].empty? ? nil : rel[:type],
             description: relation_description(rel),
             bibitem: bib_item(item_data(rel.at("./bibitem"))),
             locality: localities(rel),
@@ -453,7 +451,7 @@ module RelatonBib
       # Create Locality object from Nokogiri::XML::Element
       #
       # @param loc [Nokogiri::XML::Element]
-      # @param klass [RelatonBib::Locality.class, RelatonBib::LocalityStack.class]
+      # @param klass [RelatonBib::Locality, RelatonBib::LocalityStack]
       #
       # @return [RelatonBib::Locality]
       def locality(loc, klass = Locality)

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: relaton-bib
 version: !ruby/object:Gem::Version
-  version: 1.13.7
+  version: 1.13.9
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2022-09-20 00:00:00.000000000 Z
+date: 2022-09-24 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: byebug