relaton-render 0.1.0

data/lib/parse/parse_extract.rb

@@ -0,0 +1,155 @@
+class Iso690Parse
+  def title(doc)
+    doc&.at("./title")&.text
+  end
+
+  def medium(doc, host)
+    x = doc.at("./medium") || host&.at("./medium") or return nil
+
+    %w(content genre form carrier size scale).each_with_object([]) do |i, m|
+      m << x.at("./#{i}")&.text
+    end.compact.join(", ")
+  end
+
+  def blank?(text)
+    text.nil? || text.empty?
+  end
+
+  def edition(doc, host)
+    x = doc.at("./edition") || host&.at("./edition") or return nil
+
+    x.text
+  end
+
+  def place(doc, host)
+    x = doc.at("./place") || host&.at("./place") or return nil
+
+    x.text
+  end
+
+  def publisher(doc, host)
+    x = doc.at("./contributor[role/@type = 'publisher']/organization/name") ||
+      host&.at("./contributor[role/@type = 'publisher']/organization/name") or
+      return nil
+    x.text
+  end
+
+  def distributor(doc, host)
+    x = doc.at("./contributor[role/@type = 'distributor']/organization/name") ||
+      host&.at("./contributor[role/@type = 'distributor']/organization/name") or
+      return nil
+    x.text
+  end
+
+  def series_title(doc)
+    doc&.at("./title")&.text || doc&.at("./formattedref")&.text
+  end
+
+  def series_abbr(doc)
+    doc&.at("./abbreviation")&.text
+  end
+
+  def series_num(doc)
+    doc&.at("./number")&.text
+  end
+
+  def series_partnumber(doc)
+    doc&.at("./partnumber")&.text
+  end
+
+  def series_run(doc)
+    doc&.at("./run")&.text
+  end
+
+  def standardidentifier(doc)
+    doc.xpath("./docidentifier").each_with_object([]) do |id, ret|
+      ret << id.text unless %w(metanorma metanorma-ordinal).include? id["type"]
+    end
+  end
+
+  def uri(doc)
+    uri = doc.at("./uri[@type = 'doi']") || doc.at("./uri[@type = 'uri']") ||
+      doc.at("./uri[@type = 'src']") || doc.at("./uri")
+    uri&.text
+  end
+
+  def access_location(doc, host)
+    x = doc.at("./accessLocation") || host&.at("./accessLocation") or
+      return nil
+    x.text
+  end
+
+  def included(type)
+    ["article", "inbook", "incollection", "inproceedings"].include? type
+  end
+
+  def wrap(text, startdelim = " ", enddelim = ".")
+    return "" if blank?(text)
+
+    "#{startdelim}#{text}#{enddelim}"
+  end
+
+  def type(doc)
+    type = doc.at("./@type") and return type&.text
+    doc.at("./relation[@type = 'includedIn']") and return "inbook"
+    "book"
+  end
+
+  def extent1(localities)
+    localities.each_with_object({}) do |l, ret|
+      ret[(l["type"] || "page").to_sym] = {
+        from: l.at("./referenceFrom")&.text,
+        to: l.at("./referenceTo")&.text,
+      }
+    end
+  end
+
+  def extent0(elem, acc, ret1)
+    case elem.name
+    when "localityStack"
+      acc << ret1
+      ret1 = {}
+      acc << extent1(elem.elements)
+    when "locality" then ret1.merge!(extent1([elem]))
+    when "referenceFrom" then ret1.merge!(extent1([elem.parent]))
+    end
+    [acc, ret1]
+  end
+
+  def extent(doc)
+    ret1 = {}
+    ret = doc.xpath("./extent").each_with_object([]) do |e, acc|
+      e.elements.each do |l|
+        acc, ret1 = extent0(l, acc, ret1)
+        break if l.name == "referenceFrom"
+      end
+    end
+    ret << ret1
+    ret.reject(&:empty?)
+  end
+
+  def draft(doc)
+    dr = doc&.at("./status/stage")&.text
+
+    iterord = iter_ordinal(doc)
+    status = status_print(dr)
+    status = "#{iterord} #{status}" if iterord
+    status
+  end
+
+  def iter_ordinal(isoxml)
+    return nil unless isoxml.at(("./status/iteration"))
+
+    iter = isoxml.at(("./status/iteration"))&.text || "1"
+    iter.to_i.localize.to_rbnf_s("SpelloutRules",
+                                 "spellout-ordinal").capitalize
+  end
+
+  def status_print(status)
+    status
+  end
+
+  def status(doc)
+    doc&.at("./status/stage")&.text
+  end
+end

data/lib/relaton/config.yml

@@ -0,0 +1,67 @@
+nametemplate:
+  one: "{{surname[0] | upcase }}, {{given[0]}} {{middle[0}}"
+  two: "{{surname[0] | upcase}}, {{given[0]}} {{middle[0]}} {{ labels['and'] }} {{given[1]}} {{middle[1]}} {{surname[1] | upcase}}"
+  more: "{{surname[0] | upcase}}, {{given[0]}} {{middle[0}} , {{given[1]}} {{middle[1]}} {{surname[1] | upcase}} {{ labels['and'] }} {{given[2]}} {{middle[2]}} {{surname[2] | upcase}}"
+# disabled the following: they should be provided in inheriting calls
+# etal: "{{surname[0] | upcase}}, {{given[0]}} {{middle[0]}}, {{given[1]}} {{middle[1]}} {{surname[1] | upcase}} <em>et al.</em>"
+# etal_count: 5
+seriestemplate: "{% if series_abbr %}{{series_abbr}}{% else %}{{series_title}}{% endif %} ({{series_run}}) {{series_num}}|({{series_partnumber}})"
+journaltemplate: "<em>{% if series_abbr %}{{series_abbr}}{% else %}{{series_title}}{% endif %}</em> ({{series_run}}) {{ labels['volume'] }}_{{series_num}} {{ labels['part'] }}_{{series_partnumber}}"
+extenttemplate:
+  book: "{{ volume }}_{{ labels['extent']['volume'] }}, {{ page }}_{{ labels['extent']['page'] }}"
+  booklet: book
+  proceedings: book
+  journal: book
+  standard: book
+  techreport: book
+  inbook: "{{ volume }}: {{ page }}"
+  misc: "{{ labels['extent']['volume'] }}_{{ volume }}, {{ labels['extent']['page'] }}_{{ page }}"
+  # TODO: software extent
+language: en
+script: Latn
+template:
+  book: "{{ creatornames }} ({{role}}) . <em>{{ title }}</em> [{{medium}}] . {{ edition | capitalize }}. ({{ series }}.) {{place}}: {{publisher}}. {{date}}. {{ labels['updated'] | capitalize }}:_{{date_updated}}. {{ standardidentifier | first }}. {{ uri }}. {{ labels['at'] | capitalize}}:_{{ access_location }}. [{{ labels['viewed'] }}:_{{date_accessed}}]. {{extent}}."
+    # TODO: omitted: author ids, subsidiary titles, subsidiary creators, rights metadata, distributor, item attributes, relationships
+  booklet: book
+  manual: book
+  proceedings: book
+  inbook: "{{ creatornames }} ({{role}}) . {{ title }} . {{ labels['in'] | capitalize }}: {{ host_creatornames}} ({{ host_role}}) : <em>{{host_title}}</em> [{{medium}}] . {{ edition | capitalize }}. ({{ series }}.) {{place}}: {{publisher}}. {{date}}. {{extent}}. {{ labels['updated'] | capitalize }}:_{{date_updated}}. {{ standardidentifier | first }}. {{ uri }}. {{ labels['at'] | capitalize}}:_{{ access_location }}. [{{ labels['viewed'] }}:_{{date_accessed}}]."
+  inproceedings: inbook
+  incollection: inbook
+    # TODO: omitted: author ids, additional info for component part, subsidiary titles of host resource, rights metadata, distributor, relationships
+  journal: "<em>{{ title}}</em> [{{medium}}] . {{ edition | capitalize }}. {{place}}: {{publisher}}. {{date}}. {{extent}}. {{ standardidentifier | first }}. {{ uri }}. {{ labels['at'] | capitalize}}:_{{ access_location }}. [{{ labels['viewed'] }}:_{{date_accessed}}]."
+    # TODO subsidiary titles, rights metadata, item attributes, relationships
+  article: "{{ creatornames }} ({{role}}) . {{ title }}. {{ series }} [{{medium}}] . {{ edition | capitalize }}. {{ extent }}. {{place}}: {{publisher}}. {{date}}. {{ labels['updated'] | capitalize }}:_{{date_updated}}. {{ standardidentifier | first }}. {{ uri }}. {{ labels['at'] | capitalize}}:_{{ access_location }}. [{{ labels['viewed'] }}:_{{date_accessed}}]."
+    # TODO: omitted: author ids, subsidiary titles, rights metadata, distributor, item attributes, relationships; newspapers
+  software: "{{ creatornames }} ({{ role}}) . <em>{{ title }}</em> . {{ labels['version'] | capitalize }}_{{ edition_raw }}. {{medium | capitalize}}. {{place}}: {{publisher}}. {{date}}. {{ labels['updated'] | capitalize }}:_{{date_updated}}. {{ standardidentifier | first }}. {{ uri }}. {{ extent}}. [{{ labels['viewed'] }}:_{{date_accessed}}]. "
+    # TODO: omitted: author ids, subsidiary titles, subsidiary creators, copyright, license, distributor, system requirements, relationships
+  electronic resource:  software
+  standard: "{{ creatornames }} ({{ role}}) . <em>{{ title }}</em> . {{ labels['in'] | capitalize }}:_{{ series_title }}. {{ medium | capitalize }}. {{ edition | capitalize }}. {{place}}: {{publisher}}. {{date}}. {{ extent }}. {{ standardidentifier | first }}. {{ uri }}. {{ labels['at'] | capitalize}}:_{{ access_location }}. [{{ labels['viewed'] }}:_{{date_accessed}}]."
+    # TODO: omitted: author ids, subsidiary titles, rights metadata, distributor, persistent identifier, item attributes, relationships
+  techreport: standard
+  dataset: "{{ creatornames }} ({{ role }}) . <em>{{ title }}</em> . {{ labels['version'] | capitalize }}_{{ edition_raw }}. {{medium | capitalize }}. {{ labels['in'] | capitalize }}:_{{series}}. {{date}}. {{ labels['updated'] | capitalize }}:_{{date_updated}}. {{ standardidentifier | first }}. {{ uri }}. {{ extent}}. [{{ labels['viewed'] }}:_{{date_accessed}}]. "
+    # TODO: omitted: author ids, system requirements, host archive, copyright, license, data source, distributor, usage instructions, relationships, provenance
+  website: "{{ creatornames }} ({{ role }}) . <em>{{ title }}</em> . {{ labels['version'] | capitalize }}_{{ edition_raw }}. {{medium | capitalize }}. {{ place }}: {{ publisher }}. {{date}}. {{ labels['updated'] | capitalize }}:_{{date_updated}}. {{ standardidentifier | first }}. {{ uri }}. [{{ labels['viewed'] }}:_{{date_accessed}}]. "
+    # TODO: omitted: author ids, page title, system requirements, subsidiary creator, rights metadata, distributor, persistent identifier, archive location, archive date and time, relationships
+  webresource: website
+    # TODO: component part
+  unpublished: "{{ creatornames }} ({{ role }}) . <em>{{ title }}</em> . {{ medium | capitalize }}. {{ date }}. {{ labels['at'] | capitalize}}:_{{ access_location }}.  {{ standardidentifier | first }}. {{ uri }}. [{{ labels['viewed'] }}:_{{date_accessed}}]."
+    # TODO: omitted: author ids, subsidiary titles, subsidiary creators, host archive, item attributes, relationships
+  presentation: unpublished
+  thesis: "{{ creatornames }} ({{ role }}) . <em>{{ title }}</em> . {{ medium | capitalize }}. {{ place }}: {{ publisher }}. {{ date }}. {{ labels['at'] | capitalize}}:_{{ access_location }}.  {{ standardidentifier | first }}. {{ uri }}. [{{ labels['viewed'] }}:_{{date_accessed}}]."
+  # unsupported types:
+  # map: ""
+  # audiovisual: ""
+  # film: ""
+  # video: ""
+  # broadcast: ""
+  # graphic_work: ""
+  # music: ""
+  # performance: "" (not in Relaton)
+  # patent: ""
+  # archival: ""
+  # social_media: ""
+  # alert: ""
+  # message: ""
+  # conversation: ""
+  # misc: ""

data/lib/relaton/render.rb

@@ -0,0 +1,123 @@
+require_relative "render_classes"
+require_relative "render_fields"
+require "yaml"
+require "liquid"
+require_relative "../isodoc/i18n"
+require_relative "../utils/utils"
+
+module Relaton
+  module Render
+    class General
+      attr_reader :template, :journaltemplate, :seriestemplate, :nametemplate,
+                  :extenttemplate, :lang, :script, :edition, :edition_number,
+                  :date
+
+      def initialize(opt = {})
+        options = YAML.load_file(File.join(File.dirname(__FILE__),
+                                           "config.yml"))
+          .merge(Utils::string_keys(opt))
+        @type = self.class.name.downcase.sub(/relaton::render::/, "")
+        root_initalize(options)
+        render_initialize(options)
+        @parse ||= options["parse"]
+        @i18n ||= options["i18n"]
+      end
+
+      def root_initalize(opt)
+        i18n_initialize(opt)
+        @parse = Iso690Parse.new
+        @nametemplate = Iso690NameTemplate
+          .new(template: opt["nametemplate"], i18n: @i18n)
+        @seriestemplate = Iso690SeriesTemplate
+          .new(template: opt["seriestemplate"], i18n: @i18n)
+        @journaltemplate = Iso690SeriesTemplate
+          .new(template: opt["journaltemplate"], i18n: @i18n)
+        @extenttemplate = extentrenderers(opt)
+      end
+
+      def i18n_initialize(opt)
+        @lang = opt["language"]
+        @script = opt["script"]
+        @i18n = i18n(opt["language"], opt["script"])
+        @edition_number = opt["edition_number"] || @i18n.edition_number
+        @edition = opt["edition"] || @i18n.edition
+        @date = opt["date"] || @i18n.date
+      end
+
+      def render_initialize(opt)
+        case opt["template"]
+        when String
+          @template = Iso690Template.new(template: opt["template"],
+                                         i18n: @i18n)
+        when Hash
+          @render = renderers(opt)
+        end
+      end
+
+      def renderers(opt)
+        template_hash_fill(opt["template"]).each_with_object({}) do |(k, v), m|
+          @type == "general" || @type == k or next
+          m[k] = General.subclass(k)
+            .new(template: v, parse: @parse, i18n: @i18n,
+                 language: @lang, script: @script)
+        end
+      end
+
+      def extentrenderers(opt)
+        Iso690ExtentTemplate
+          .new(template: template_hash_fill(opt["extenttemplate"]), i18n: @i18n)
+      end
+
+      def default_template
+        "{{creatornames}}. {{title}}. {{date}}."
+      end
+
+      def i18n(lang = "en", script = "Latn")
+        ::IsoDoc::I18n.new(lang, script)
+      end
+
+      def render(bib, embedded: false)
+        docxml = Nokogiri::XML(bib)
+        docxml.remove_namespaces!
+        parse(docxml.root, embedded: embedded)
+      end
+
+      def parse(doc, embedded: false)
+        f = doc.at("./formattedref") and
+          return embedded ? f.children.to_xml : doc.to_xml
+
+        ret = parse1(doc)
+        embedded and return ret
+        "<formattedref>#{ret}</formattedref>"
+      end
+
+      def renderer(doc)
+        unless ret = @template || @render[doc["type"]]&.template
+          raise "No renderer defined for #{doc['type']}"
+        end
+        unless @type == "general" || @type == doc["type"]
+          raise "No renderer defined for #{doc['type']}"
+        end
+
+        ret
+      end
+
+      def parse1(doc)
+        r = renderer(doc)
+        data = @parse.extract(doc)
+        data_liquid = Fields.new(renderer: self).compound_fields_format(data)
+        @i18n.l10n(r.render(data_liquid))
+      end
+
+      private
+
+      def template_hash_fill(templates)
+        BIBTYPE.each_with_object({}) do |type, m|
+          template = templates[type] || templates["misc"] || default_template
+          BIBTYPE.include?(template) and template = templates[template]
+          m[type] = template
+        end
+      end
+    end
+  end
+end

data/lib/relaton/render_classes.rb

@@ -0,0 +1,126 @@
+module Relaton
+  module Render
+    BIBTYPE =
+      %w(article book booklet manual proceedings presentation thesis techreport
+         standard unpublished map electronic_resource audiovisual film video
+         broadcast software graphic_work music patent inbook incollection
+         inproceedings journal website webresource dataset archival social_media
+         alert message conversation misc).freeze
+
+    class General
+      singleton_class.send(:attr_reader, :descendants)
+      @descendants = {}
+
+      def self.inherited(subclass) # rubocop:disable Lint/MissingSuper
+        type = subclass.name.downcase.sub(/relaton::render::/, "")
+        General.descendants[type] = subclass
+      end
+
+      def self.subclass(type)
+        @descendants[type]
+      end
+    end
+
+    class Book < General
+    end
+
+    class Journal < General
+    end
+
+    class Booklet < General
+    end
+
+    class Manual < General
+    end
+
+    class Proceedings < General
+    end
+
+    class Presentation < General
+    end
+
+    class Thesis < General
+    end
+
+    class Techreport < General
+    end
+
+    class Standard < General
+    end
+
+    class Unpublished < General
+    end
+
+    class Map < General
+    end
+
+    class Electronic_resource < General
+    end
+
+    class Software < Electronic_resource
+    end
+
+    class Webresource < Electronic_resource
+    end
+
+    class Website < Webresource
+    end
+
+    class Dataset < Electronic_resource
+    end
+
+    class Social_media < Electronic_resource
+    end
+
+    class Alert < Social_media
+    end
+
+    class Message < Social_media
+    end
+
+    class Audiovisual < General
+    end
+
+    class Film < Audiovisual
+    end
+
+    class Video < Audiovisual
+    end
+
+    class Broadcast < Audiovisual
+    end
+
+    class Graphic_work < General
+    end
+
+    class Music < General
+    end
+
+    class Patent < General
+    end
+
+    class Hosted < General
+    end
+
+    class Article < Hosted
+    end
+
+    class InBook < Hosted
+    end
+
+    class InCollection < Hosted
+    end
+
+    class InProceedings < Hosted
+    end
+
+    class Archival < General
+    end
+
+    class Conversation < General
+    end
+
+    class Misc < General
+    end
+  end
+end

data/lib/relaton/render_fields.rb

@@ -0,0 +1,151 @@
+module Relaton
+  module Render
+    class Fields
+      def initialize(options)
+        @r = options[:renderer]
+      end
+
+      def compound_fields_format(hash)
+        name_fields_format(hash)
+        role_fields_format(hash)
+        date_fields_format(hash)
+        misc_fields_format(hash)
+      end
+
+      def name_fields_format(hash)
+        hash[:creatornames] = nameformat(hash[:creators])
+        hash[:host_creatornames] = nameformat(hash[:host_creators])
+      end
+
+      def role_fields_format(hash)
+        hash[:role] = role_inflect(hash[:creators], hash[:role_raw])
+        hash[:host_role] =
+          role_inflect(hash[:host_creators], hash[:host_role_raw])
+      end
+
+      def misc_fields_format(hash)
+        hash[:series] = seriesformat(hash)
+        hash[:edition] = editionformat(hash[:edition_raw])
+        hash[:extent] = extentformat(hash[:extent_raw], hash)
+        hash
+      end
+
+      def date_fields_format(hash)
+        hash[:date] = dateformat(hash[:date], hash)
+        hash[:date_updated] = dateformat(hash[:date_updated], hash)
+        hash[:date_accessed] = dateformat(hash[:date_accessed], hash)
+      end
+
+      def seriesformat(hash)
+        parts = %i(series_title series_abbr series_num series_partnumber
+                   series_run)
+        series_out = parts.each_with_object({}) do |i, m|
+          m[i] = hash[i]
+        end
+        t = hash[:type] == "article" ? @r.journaltemplate : @r.seriestemplate
+        t.render(series_out)
+      end
+
+      def nameformat(names)
+        return names if names.nil?
+
+        parts = %i(surname initials given middle)
+        names_out = names.each_with_object({}) do |n, m|
+          parts.each do |i|
+            m[i] ||= []
+            m[i] << n[i]
+          end
+        end
+        @r.nametemplate.render(names_out)
+      end
+
+      def role_inflect(contribs, role)
+        return nil if role.nil? || contribs.size.zero?
+
+        number = contribs.size > 1 ? "pl" : "sg"
+        @r.i18n.get[role][number] || role
+      end
+
+      def editionformat(edn)
+        return edn unless /^\d+$/.match?(edn)
+
+        num = edn.to_i.localize(@r.lang.to_sym)
+          .to_rbnf_s(*@r.edition_number)
+        @r.edition.sub(/%/, num)
+      end
+
+      def extentformat(extent, hash)
+        extent.map do |e|
+          extent_out = e.merge(type: hash[:type],
+                               host_title: hash[:host_title])
+            .transform_values do |v|
+              v.is_a?(Hash) ? range(v) : v
+            end
+          @r.extenttemplate.render(extent_out.merge(orig: e))
+        end.join("; ")
+      end
+
+      def range(hash)
+        if hash[:on] then hash[:on]
+        elsif hash.has_key?(:from) && hash[:from].nil? then nil
+        elsif hash[:from]
+          hash[:to] ? "#{hash[:from]}&#x2013;#{hash[:to]}" : hash[:from]
+        else hash
+        end
+      end
+
+      def date_range(hash)
+        if hash[:from]
+          "#{hash[:from]}&#x2013;#{hash[:to]}"
+        else range(hash)
+        end
+      end
+
+      def dateformat(date, hash)
+        return nil if date.nil?
+
+        %i(from to on).each do |k|
+          date[k] = daterender(date[k], hash)
+        end
+        date_range(date)
+      end
+
+      def daterender(date, _hash)
+        return date if date.nil? || /^\d+$/.match?(date)
+
+        daterender1(date, dategranularity(date), hash)
+      end
+
+      def daterender1(date, format, _hash)
+        datef = dateparse(date, format, @r.lang.to_sym)
+        case @r.date[format]
+        when "to_full_s", "to_long_s", "to_medium_s", "to_short_s"
+          datef.send @r.date[format]
+        else
+          datef.to_additional_s(@r.date[format])
+        end
+      end
+
+      private
+
+      def dategranularity(date)
+        case date
+        when /^\d+-\d+$/ then "month_year"
+        when /^\d+-\d+-\d+$/ then "day_month_year"
+        else "date_time"
+        end
+      end
+
+      def dateparse(date, format, lang)
+        case format
+        when "date_time" then DateTime.parse(date)
+          .localize(lang, timezone: "Zulu")
+        when "day_month_year" then DateTime.parse(date)
+          .localize(lang, timezone: "Zulu").to_date
+        when "month_year" then Date.parse(date)
+          .localize(lang, timezone: "Zulu").to_date
+        end
+      end
+    end
+  end
+end

data/lib/relaton/version.rb

@@ -0,0 +1,5 @@
+module Relaton
+  module Render
+    VERSION = "0.1.0".freeze
+  end
+end

data/lib/relaton-render.rb

@@ -0,0 +1,6 @@
+require "relaton/version"
+require "relaton/render"
+require "parse/parse"
+require "template/template"
+require "isodoc/i18n"
+