isodoc 1.3.1 → 1.4.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 29fb94d37c013e802491e2c1266e8b39d607ea3050f272f1a7bcd398e57c0283
-  data.tar.gz: 85ea530071d54477f462d34a5b18dbf9ba500b71a4d5ea858b62530757a6b795
+  metadata.gz: bf37a37eb402d844a36b33e3911c67d3ebcbef7bf7f864dfc0e9f3a8aca7603c
+  data.tar.gz: cc3645b92e85d43e2c6e49cfce67cc13d0f656cf1c16d67247cb7053eceba394
 SHA512:
-  metadata.gz: 9539cdc3d0fac50a04aa3cdf93581016306b660f1d0681a8a1e06255ed839ac9362c887b9a6f0ef7480711122c5275e0936fab640efff76c84aaa5ad84831f84
-  data.tar.gz: b05012f1f7189becfa14871de424ae5713a8791a9c0b520cdcca591d3f733d996c776d2f37d2c1ff23fdff6b2c146b6e09d2f9c5b97dcbeec43edafd9785dfcb
+  metadata.gz: 83bfcc73e86c73c9373bf9b42170ebc58c4ae50e4a0789cacddefa05c5b80fe1c59559e3abde2071e79aa3fd839d7844b3c0145ca89ed09b9d6d7e09b1e03de9
+  data.tar.gz: 11fbb6a5a6a19eadae64d22dd032185eaed46a1dd007f188e26cc73153e75ceb16ec3c64c8613c8dc05e0263fafab98655287e23993c34ebcf815096dca50bda

data/lib/isodoc-yaml/i18n-en.yaml

@@ -97,3 +97,58 @@ locality: {
                       note: Note,
                       formula: Formula
                     }
+inflection:
+  Clause:
+    sg: Clause
+    pl: Clauses
+  Annex:
+    sg: Annex
+    pl: Annexes
+  Appendix:
+    sg: Appendix
+    pl: Appendixes
+  Note:
+    sg: Note
+    pl: Notes
+  "Note % to entry":
+    sg: Note % to entry
+    pl: Notes % to entry
+  List:
+    sg: List
+    pl: Lists
+  Figure:
+    sg: Figure
+    pl: Figures
+  Formula:
+    sg: Formula
+    pl: Formulas
+  Table:
+    sg: Table
+    pl: Tables
+  Requirement:
+    sg: Requirement
+    pl: Requirements
+  Recommendation:
+    sg: Recommendation
+    pl: Recommendations
+  Permission:
+    sg: Permission
+    pl: Permissions
+  Example:
+    sg: Example
+    pl: Examples
+  Part:
+    sg: Part
+    pl: Parts
+  Section:
+    sg: Section
+    pl: Sections
+  Paragraph:
+    sg: Paragraph
+    pl: Paragraphs
+  Chapter:
+    sg: Chapter
+    pl: Chapters
+  Page:
+    sg: Page
+    pl: Pages

data/lib/isodoc-yaml/i18n-fr.yaml

@@ -90,3 +90,59 @@ locality: {
                       note: Note,
                       formula: Formule
                     }
+inflection:
+  Clause:
+    sg: Article
+    pl: Articles
+  Annex:
+    sg: Annexe
+    pl: Annexes
+  Appendix:
+    sg: Appendice
+    pl: Appendices
+  Note:
+    sg: Note
+    pl: Notes
+  "Note % à l’article":
+    sg: Note % à l’article
+    pl: Notes % à l’article
+  List:
+    sg: Liste
+    pl: Listes
+  Figure:
+    sg: Figure
+    pl: Figures
+  Formula:
+    sg: Formule
+    pl: Formules
+  Table:
+    sg: Tableau
+    pl: Tableaux
+  Requirement:
+    sg: Exigence
+    pl: Exigences
+  Recommendation:
+    sg: Recommandation
+    pl: Recommandations
+  Permission:
+    sg: Autorisation
+    pl: Autorisations
+  Example:
+    sg: Exemple
+    pl: Exemples
+  Part:
+    sg: Partie
+    pl: Parties
+  Section:
+    sg: Section
+    pl: Sections
+  Paragraph:
+    sg: Alinéa
+    pl: Alinéas
+  Chapter:
+    sg: Chapitre
+    pl: Chapitres
+  Page:
+    sg: Page
+    pl: Pages
+

data/lib/isodoc/convert.rb

@@ -157,7 +157,7 @@ module IsoDoc
       FileUtils.rm_rf dir
     end
 
-    def middle_clause
+    def middle_clause(docxml = nil)
       "//clause[parent::sections][not(@type = 'scope')]"\
         '[not(descendant::terms)]'
     end

data/lib/isodoc/function/section.rb

@@ -56,7 +56,7 @@ module IsoDoc::Function
     end
 
     def clause(isoxml, out)
-      isoxml.xpath(ns(middle_clause)).each do |c|
+      isoxml.xpath(ns(middle_clause(isoxml))).each do |c|
         out.div **attr_code(clause_attrs(c)) do |s|
           clause_name(nil, c&.at(ns("./title")), s, nil)
           c.elements.reject { |c1| c1.name == "title" }.each do |c1|

data/lib/isodoc/html_function/mathvariant_to_plain.rb

@@ -0,0 +1,82 @@
+module IsoDoc
+  module HtmlFunction
+    # Class for recursively converting mathvariant text into plain text symbols
+    class MathvariantToPlain
+      MATHML = { "m" => "http://www.w3.org/1998/Math/MathML" }.freeze
+      MATHVARIANT_SPECIAL_CASE_MAPPINGS_1 = %w[bold italic sans-serif]
+        .permutation
+        .each_with_object(:sansbolditalic)
+        .map { |n, y| [n, y] }
+        .to_h
+        .freeze
+      MATHVARIANT_SPECIAL_CASE_MAPPINGS_2 = {
+        %w[bold fraktur] => :frakturbold,
+        %w[bold script] => :scriptbold,
+        %w[sans-serif bold] => :sansbold,
+        %w[sans-serif italic] => :sansitalic,
+        %w[sans-serif bold-italic] => :sansbolditalic,
+        %w[bold-sans-serif italic] => :sansbolditalic,
+        %w[sans-serif-italic bold] => :sansbolditalic,
+      }.freeze
+      MATHVARIANT_TO_PLANE_MAPPINGS = {
+        %w[double-struck] => :doublestruck,
+        %w[bold-fraktur] => :frakturbold,
+        %w[script] => :script,
+        %w[bold-script] => :scriptbold,
+        %w[fraktur] => :fraktur,
+        %w[sans-serif] => :sans,
+        %w[bold-sans-serif] => :sansbold,
+        %w[sans-serif-italic] => :sansitalic,
+        %w[sans-serif-bold-italic] => :sansbolditalic,
+        %w[monospace] => :monospace,
+      }.freeze
+
+      attr_reader :docxml
+
+      # @param [Nokogiri::Document] docxml
+      def initialize(docxml)
+        @docxml = docxml
+      end
+
+      def convert
+        docxml.xpath("//m:math", MATHML).each do |elem|
+          next if nothing_to_style(elem)
+          mathml1(elem)
+        end
+        docxml
+      end
+
+      private
+
+      def nothing_to_style(elem)
+        !elem.at("./*[@mathvariant][not(@mathvariant = 'normal')][not(@mathvariant = 'italic')]")
+      end
+
+      def mathml1(base_elem)
+        MATHVARIANT_SPECIAL_CASE_MAPPINGS_1
+          .merge(MATHVARIANT_SPECIAL_CASE_MAPPINGS_2)
+          .merge(MATHVARIANT_TO_PLANE_MAPPINGS)
+          .each_pair do |mathvariant_list, plain_font|
+            base_elem.xpath(mathvariant_xpath(mathvariant_list)).each do |elem|
+              toPlane(elem, plain_font)
+            end
+          end
+      end
+
+      def mathvariant_xpath(list)
+        list
+          .map { |variant| "//*[@mathvariant = '#{variant}']" }
+          .join
+      end
+
+      def toPlane(elem, font)
+        elem.traverse do |n|
+          next unless n.text?
+
+          n.replace(Plane1Converter.conv(HTMLEntities.new.decode(n.text), font))
+        end
+        elem
+      end
+    end
+  end
+end

data/lib/isodoc/html_function/postprocess.rb

@@ -1,35 +1,47 @@
+require "isodoc/html_function/mathvariant_to_plain"
+
 module IsoDoc::HtmlFunction
   module Html
-    def postprocess(result, filename, dir)
+    def postprocess(result, filename, _dir)
       result = from_xhtml(cleanup(to_xhtml(textcleanup(result))))
       toHTML(result, filename)
       @files_to_delete.each { |f| FileUtils.rm_rf f }
     end
 
     def script_cdata(result)
-      result.gsub(%r{<script([^>]*)>\s*<!\[CDATA\[}m, "<script\\1>").
-        gsub(%r{\]\]>\s*</script>}, "</script>").
-        gsub(%r{<!\[CDATA\[\s*<script([^>]*)>}m, "<script\\1>").
-        gsub(%r{</script>\s*\]\]>}, "</script>")
+      result.gsub(%r{<script([^>]*)>\s*<!\[CDATA\[}m, "<script\\1>")
+        .gsub(%r{\]\]>\s*</script>}, "</script>")
+        .gsub(%r{<!\[CDATA\[\s*<script([^>]*)>}m, "<script\\1>")
+        .gsub(%r{</script>\s*\]\]>}, "</script>")
     end
 
     def toHTML(result, filename)
-      result = (from_xhtml(html_cleanup(to_xhtml(result))))
-      #result = populate_template(result, :html)
+      result = from_xhtml(html_cleanup(to_xhtml(result)))
+      # result = populate_template(result, :html)
       result = from_xhtml(move_images(to_xhtml(result)))
       result = html5(script_cdata(inject_script(result)))
       File.open(filename, "w:UTF-8") { |f| f.write(result) }
     end
 
     def html5(doc)
-      doc.sub(%r{<!DOCTYPE html [^>]+>}, "<!DOCTYPE html>").
-        sub(%r{<\?xml[^>]+>}, "")
+      doc.sub(%r{<!DOCTYPE html [^>]+>}, "<!DOCTYPE html>")
+        .sub(%r{<\?xml[^>]+>}, "")
     end
 
     def html_cleanup(x)
-      footnote_format(footnote_backlinks(html_toc(
-        term_header((html_footnote_filter(html_preface(htmlstyle(x))))))
-                        ))
+      mathml(
+        footnote_format(
+          footnote_backlinks(
+            html_toc(
+              term_header(html_footnote_filter(html_preface(htmlstyle(x))))
+            )
+          )
+        )
+      )
+    end
+
+    def mathml(docxml)
+      IsoDoc::HtmlFunction::MathvariantToPlain.new(docxml).convert
     end
 
     def htmlstylesheet
@@ -76,14 +88,14 @@ module IsoDoc::HtmlFunction
     def html_cover(docxml)
       doc = to_xhtml_fragment(File.read(@htmlcoverpage, encoding: "UTF-8"))
       d = docxml.at('//div[@class="title-section"]')
-      #d.children.first.add_previous_sibling doc.to_xml(encoding: "US-ASCII")
+      # d.children.first.add_previous_sibling doc.to_xml(encoding: "US-ASCII")
       d.children.first.add_previous_sibling populate_template(doc.to_xml(encoding: "US-ASCII"), :html)
     end
 
     def html_intro(docxml)
       doc = to_xhtml_fragment(File.read(@htmlintropage, encoding: "UTF-8"))
       d = docxml.at('//div[@class="prefatory-section"]')
-      #d.children.first.add_previous_sibling doc.to_xml(encoding: "US-ASCII")
+      # d.children.first.add_previous_sibling doc.to_xml(encoding: "US-ASCII")
       d.children.first.add_previous_sibling populate_template(doc.to_xml(encoding: "US-ASCII"), :html)
     end
 
@@ -93,19 +105,19 @@ module IsoDoc::HtmlFunction
     end
 
     def toclevel_classes
-      (1..@htmlToClevels).inject([]) { |m, i| m << "h#{i}" }
+      (1..@htmlToClevels).reduce([]) { |m, i| m << "h#{i}" }
     end
 
     def toclevel
       ret = toclevel_classes.map { |l| "#{l}:not(:empty):not(.TermNum):not(.noTOC)" }
       <<~HEAD.freeze
-    function toclevel() { return "#{ret.join(',')}";}
+        function toclevel() { return "#{ret.join(',')}";}
       HEAD
     end
 
     # needs to be same output as toclevel
     def html_toc(docxml)
-      idx = docxml.at("//div[@id = 'toc']") or return docxml
+      (idx = docxml.at("//div[@id = 'toc']")) or (return docxml)
       toc = "<ul>"
       path = toclevel_classes.map do |l|
         "//main//#{l}[not(@class = 'TermNum')][not(@class = 'noTOC')][text()]"

data/lib/isodoc/metadata_contributor.rb

@@ -17,9 +17,10 @@ module IsoDoc
     def extract_person_affiliations(authors)
       authors.reduce([]) do |m, a|
         name = a&.at(ns('./affiliation/organization/name'))&.text
-        location = a&.at(ns('./affiliation/organization/address/'\
-                            'formattedAddress'))&.text
-        m << (!name.nil? && !location.nil? ? "#{name}, #{location}" :
+        subdivs = a&.xpath(ns('./affiliation/organization/subdivision'))&.map(&:text)&.join(", ")
+        name and subdivs and !subdivs.empty? and name = l10n("#{name}, #{subdivs}", @lang, @script)
+        location = a&.at(ns('./affiliation/organization/address/formattedAddress'))&.text
+        m << (!name.nil? && !location.nil? ? l10n("#{name}, #{location}", @lang, @script) :
           (name || location || ''))
         m
       end

data/lib/isodoc/presentation_function/section.rb

@@ -43,6 +43,9 @@ module IsoDoc
       prefix_name(f, "", "#{lbl}#{clausedelim}", "name")
     end
 
+    def references(docxml)
+    end
+
     def index(docxml)
       docxml.xpath(ns("//index | //index-xref")).each do |f|
         f.remove

data/lib/isodoc/presentation_xml_convert.rb

@@ -30,6 +30,7 @@ module IsoDoc
       clause docxml
       annex docxml
       term docxml
+      references docxml
       index docxml
     end
 

data/lib/isodoc/version.rb

@@ -1,3 +1,3 @@
 module IsoDoc
-  VERSION = "1.3.1".freeze
+  VERSION = "1.4.0".freeze
 end

data/lib/isodoc/xref/xref_sect_gen.rb

@@ -68,7 +68,7 @@ module IsoDoc::XrefGen
     end
 
     def clause_names(docxml, sect_num)
-      docxml.xpath(ns(@klass.middle_clause)).each_with_index do |c, i|
+      docxml.xpath(ns(@klass.middle_clause(docxml))).each_with_index do |c, i|
         section_names(c, (i + sect_num), 1)
       end
     end

data/spec/isodoc/blocks_spec.rb

@@ -534,7 +534,7 @@ INPUT
       <div>
         <h1 class='ForewordTitle'>Foreword</h1>
         <p id='A'>
-          ABC 
+          ABC
           <div id='B' class='Note'>
             <p>
               <span class='note_label'>NOTE 1</span>

data/spec/isodoc/metadata_spec.rb

@@ -60,6 +60,8 @@ RSpec.describe IsoDoc do
       </name>
       <affiliation>
       <organization><name>Slate Inc.</name>
+      <subdivision>Hermeneutics Unit</subdivision>
+      <subdivision>Exegesis Subunit</subdivision>
       <address>
       <formattedAddress>Bedrock</formattedAddress>
       </address>
@@ -106,7 +108,7 @@ INPUT
 :activateddate=>"2013",
 :agency=>"ISO",
 :authors=>["Barney Rubble", "Fred Flintstone"],
-:authors_affiliations=>{"Slate Inc., Bedrock"=>["Barney Rubble"], ""=>["Fred Flintstone"]},
+:authors_affiliations=>{"Slate Inc., Hermeneutics Unit, Exegesis Subunit, Bedrock"=>["Barney Rubble"], ""=>["Fred Flintstone"]},
 :circulateddate=>"2015",
 :confirmeddate=>"2017",
 :copieddate=>"2016",

data/spec/isodoc/postproc_spec.rb

@@ -617,6 +617,169 @@ TOCLEVEL
 
   end
 
+  describe "mathvariant to plain" do
+    context "when `mathvariant` attr equal to `script`" do
+      it "converts mathvariant text chars into associated plain chars" do
+        FileUtils.rm_f "test.html"
+        FileUtils.rm_rf "test_htmlimages"
+        input = <<~INPUT
+          <?xml version="1.0" encoding="UTF-8"?>
+          <iso-standard xmlns="https://www.metanorma.org/ns/iso" type="semantic" version="1.5.14">
+            <sections>
+              <clause id="_clause" inline-header="false" obligation="normative">
+                <title>Clause</title>
+                <p id="_20514f5a-9f86-454e-b6ce-927f65ba6441">
+                  <stem type="MathML"><math xmlns="http://www.w3.org/1998/Math/MathML">
+                      <mi>x</mi>
+                      <mo>=</mo>
+                      <mstyle mathvariant="script">
+                        <mi>l</mi>
+                      </mstyle>
+                      <mo>+</mo>
+                      <mn>1</mn>
+                    </math></stem>
+                </p>
+              </clause>
+            </sections>
+          </iso-standard>
+        INPUT
+        output = <<~OUTPUT
+          <main class="main-section"><button onclick="topFunction()" id="myBtn" title="Go to top">Top</button>
+            <p class="zzSTDTitle1"></p>
+            <div id="_clause">
+              <h1>Clause</h1>
+              <p id="_20514f5a-9f86-454e-b6ce-927f65ba6441">
+              <span class="stem"><math xmlns="http://www.w3.org/1998/Math/MathML">
+                  <mi>x</mi>
+                  <mo>=</mo>
+                  <mstyle mathvariant="script">
+                    <mi>&#x1D4C1;</mi>
+                  </mstyle>
+                  <mo>+</mo>
+                  <mn>1</mn>
+                </math></span>
+            </p>
+            </div>
+          </main>
+        OUTPUT
+        IsoDoc::HtmlConvert.new({}).convert("test", input, false)
+        html = File.read("test.html")
+                .sub(/^.*<main class="main-section">/m, '<main class="main-section">')
+                .sub(%r{</main>.*$}m, "</main>")
+        expect(html).to(be_equivalent_to(output))
+      end
+    end
+
+    context "when complex `mathvariant` combinations" do
+      it "converts mathvariant text chars into associated plain chars" do
+        FileUtils.rm_f "test.html"
+        FileUtils.rm_rf "test_htmlimages"
+        input = <<~INPUT
+          <?xml version="1.0" encoding="UTF-8"?>
+          <iso-standard xmlns="https://www.metanorma.org/ns/iso" type="semantic" version="1.5.14">
+            <sections>
+              <clause id="_clause" inline-header="false" obligation="normative">
+                <title>Clause</title>
+                <p id="_20514f5a-9f86-454e-b6ce-927f65ba6441">
+                  <stem type="MathML">
+                    <math xmlns="http://www.w3.org/1998/Math/MathML">
+                      <mstyle mathvariant="sans-serif">
+                        <mfrac>
+                          <mrow>
+                            <mrow>
+                              <mi>n</mi>
+                              <mfenced open="(" close=")">
+                                <mrow>
+                                  <mstyle mathvariant="bold">
+                                    <mrow>
+                                      <mi>n</mi>
+                                      <mo>+</mo>
+                                      <mn>1</mn>
+                                      <mo>+</mo>
+                                      <mstyle mathvariant="italic">
+                                        <mi>x</mi>
+                                      </mstyle>
+                                    </mrow>
+                                  </mstyle>
+                                </mrow>
+                              </mfenced>
+                            </mrow>
+                          </mrow>
+                          <mrow>
+                            <mstyle mathvariant="bold">
+                              <mrow>
+                                <mi>y</mi>
+                                <mo>+</mo>
+                                <mstyle mathvariant="fraktur">
+                                  <mi>z</mi>
+                                </mstyle>
+                              </mrow>
+                            </mstyle>
+                          </mrow>
+                        </mfrac>
+                      </mstyle>
+                    </math>
+                  </stem>
+                </p>
+              </clause>
+            </sections>
+          </iso-standard>
+        INPUT
+        output = <<~OUTPUT
+          <main class="main-section"><button onclick="topFunction()" id="myBtn" title="Go to top">Top</button>
+            <p class="zzSTDTitle1"></p>
+            <div id="_clause">
+              <h1>Clause</h1>
+              <p id="_20514f5a-9f86-454e-b6ce-927f65ba6441">
+              <span class="stem"><math xmlns="http://www.w3.org/1998/Math/MathML">
+                  <mstyle mathvariant="sans-serif">
+                    <mfrac>
+                      <mrow>
+                        <mrow>
+                          <mi>&#x1D5C7;</mi>
+                          <mfenced open="(" close=")">
+                            <mrow>
+                              <mstyle mathvariant="bold">
+                                <mrow>
+                                  <mi>&#x1D5FB;</mi>
+                                  <mo>+</mo>
+                                  <mn>&#x1D7ED;</mn>
+                                  <mo>+</mo>
+                                  <mstyle mathvariant="italic">
+                                    <mi>&#x1D66D;</mi>
+                                  </mstyle>
+                                </mrow>
+                              </mstyle>
+                            </mrow>
+                          </mfenced>
+                        </mrow>
+                      </mrow>
+                      <mrow>
+                        <mstyle mathvariant="bold">
+                          <mrow>
+                            <mi>&#x1D606;</mi>
+                            <mo>+</mo>
+                            <mstyle mathvariant="fraktur">
+                              <mi>&#x1D59F;</mi>
+                            </mstyle>
+                          </mrow>
+                        </mstyle>
+                      </mrow>
+                    </mfrac>
+                  </mstyle>
+                </math></span>
+            </p>
+            </div>
+          </main>
+        OUTPUT
+        IsoDoc::HtmlConvert.new({}).convert("test", input, false)
+        html = File.read("test.html")
+                .sub(/^.*<main class="main-section">/m, '<main class="main-section">')
+                .sub(%r{</main>.*$}m, "</main>")
+        expect(html).to(be_equivalent_to(output))
+      end
+    end
+  end
 
   it "moves images in HTML with no file suffix" do
     FileUtils.rm_f "test.html"
@@ -1047,7 +1210,7 @@ TOCLEVEL
     OUTPUT
   end
 
-  it "deals with image captions (Word)" do 
+  it "deals with image captions (Word)" do
 FileUtils.rm_f "test.doc"
     FileUtils.rm_f "test.html"
     IsoDoc::WordConvert.new({wordstylesheet: "spec/assets/word.css", htmlstylesheet: "spec/assets/html.scss"}).convert("test", <<~"INPUT", false)
@@ -1366,7 +1529,7 @@ expect(xmlpp(html.sub(/^.*<body /m, "<body ").sub(%r{</body>.*$}m, "</body>"))).
            <div class='Section3' id=''>
              <h1 class='IntroTitle'>Preface 1</h1>
              <p align='center' style='text-align:center;' class='MsoNormal'>
-               This is a 
+               This is a
                <p class='MsoNormal'>
                  <br clear='all' class='section'/>
                </p>
@@ -1412,7 +1575,7 @@ expect(xmlpp(html.sub(/^.*<body /m, "<body ").sub(%r{</body>.*$}m, "</body>"))).
                <p class='Note'>
                  <span class='note_label'/>
                  <span style='mso-tab-count:1'>&#xA0; </span>
-                 For further information on the Foreword, see 
+                 For further information on the Foreword, see
                  <b>ISO/IEC Directives, Part 2, 2016, Clause 12.</b>
                </p>
                <p class='Note'>

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: isodoc
 version: !ruby/object:Gem::Version
-  version: 1.3.1
+  version: 1.4.0
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2020-12-14 00:00:00.000000000 Z
+date: 2020-12-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: asciimath
@@ -374,6 +374,7 @@ files:
 - lib/isodoc/html_function/comments.rb
 - lib/isodoc/html_function/footnotes.rb
 - lib/isodoc/html_function/html.rb
+- lib/isodoc/html_function/mathvariant_to_plain.rb
 - lib/isodoc/html_function/postprocess.rb
 - lib/isodoc/i18n.rb
 - lib/isodoc/metadata.rb