isodoc 1.7.6 → 1.8.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: bf6b7672b0923b4266fb04a8badd638b3838ede377e1b712f8d30e4c1aa6061f
-  data.tar.gz: 90c92ce493296887b9215384762a5a19bee7d5c2d275bab6f696b06593b60fa3
+  metadata.gz: ec504b3b905ec960bef74398b3789ab5b7eb248dc05a7c3f64590db77d31cac6
+  data.tar.gz: c2de4c1589a7ff282715f4930a1851393358a64fd407c5a1e3bda02beddc0782
 SHA512:
-  metadata.gz: 1f102fb8d85bce4966bd70cac156c3f27c9308e00619993f79b0d9a32f168c03a25cc2c9860782e370654c70a9288760ddaaff60e75dad280ad992bf0bfbea24
-  data.tar.gz: 4d8f3b39550d2ff82a63a58ff9a549a7dc349fe2c294c2736ae5c88251db187ce8a8223ea47a939f1c4aa9e39a9109bc6aacca0b66629a69d3eeabf0613e13c9
+  metadata.gz: 34a1c43f6f34f90aa326ed7ff50ede791eadaf22c05ebf0f4183fb2cb3c708bd4f0512572b638a94e236b807f5a873986c24579fe126eb88ad15ef1e178c64d2
+  data.tar.gz: 23fbe4f295729ad565ab787903e9c4b49388671766b91477dc3d26302b8499da892f77d6e034ce4f6cc4ba1b0956c8dbb051e94a3a336acdb661d1e6d30e584b

data/isodoc.gemspec

@@ -29,7 +29,7 @@ Gem::Specification.new do |spec|
   spec.required_ruby_version = Gem::Requirement.new(">= 2.5.0")
 
   spec.add_dependency "asciimath"
-  spec.add_dependency "html2doc", "~> 1.1.1"
+  spec.add_dependency "html2doc", "~> 1.2.0"
   spec.add_dependency "htmlentities", "~> 4.3.4"
   spec.add_dependency "liquid", "~> 4"
   # spec.add_dependency "metanorma", ">= 1.2.0"

data/lib/isodoc/class_utils.rb

@@ -27,5 +27,28 @@ module IsoDoc
       end.join
       Liquid::Template.parse(doc)
     end
+
+    def case_strict(text, casing, script)
+      return text unless %w(Latn Cyrl Grek Armn).include?(script)
+
+      letters = text.chars
+      case casing
+      when "capital" then letters.first.upcase!
+      when "lowercase" then letters.first.downcase!
+      end
+      letters.join
+    end
+
+    def case_with_markup(linkend, casing, script)
+      seen = false
+      xml = Nokogiri::XML("<root>#{linkend}</root>")
+      xml.traverse do |b|
+        next unless b.text? && !seen
+
+        b.replace(Common::case_strict(b.text, casing, script))
+        seen = true
+      end
+      xml.root.children.to_xml
+    end
   end
 end

data/lib/isodoc/convert.rb

@@ -102,6 +102,7 @@ module IsoDoc
       @bookmarks_allocated = { "X" => true }
       @fn_bookmarks = {}
       @baseassetpath = options[:baseassetpath]
+      @aligncrosselements = options[:aligncrosselements]
     end
 
     def tmpimagedir_suffix
@@ -118,6 +119,7 @@ module IsoDoc
 
     def convert1(docxml, filename, dir)
       @xrefs.parse docxml
+      bibitem_lookup(docxml)
       noko do |xml|
         xml.html **{ lang: @lang.to_s } do |html|
           html.parent.add_namespace("epub", "http://www.idpf.org/2007/ops")
@@ -129,6 +131,13 @@ module IsoDoc
       end.join("\n")
     end
 
+    def bibitem_lookup(docxml)
+      @bibitems = docxml.xpath(ns("//references/bibitem"))
+        .each_with_object({}) do |b, m|
+        m[b["id"]] = b
+      end
+    end
+
     def metadata_init(lang, script, i18n)
       @meta = Metadata.new(lang, script, i18n)
     end

data/lib/isodoc/function/cleanup.rb

@@ -165,6 +165,10 @@ module IsoDoc
             t << a.remove
           end
         end
+        table_footnote_cleanup_propagate(docxml)
+      end
+
+      def table_footnote_cleanup_propagate(docxml)
         docxml.xpath("//p[not(self::*[@class])]"\
                      "[ancestor::*[@class = 'TableFootnote']]").each do |p|
           p["class"] = "TableFootnote"

data/lib/isodoc/function/inline.rb

@@ -41,10 +41,8 @@ module IsoDoc
       end
 
       def eref_target(node)
-        href = "##{node['bibitemid']}"
-        url = node.at(ns("//bibitem[@id = '#{node['bibitemid']}']/"\
-                         "uri[@type = 'citation']"))
-        return href unless url
+        return "##{node['bibitemid']}" unless url = @bibitems[node["bibitemid"]]
+          &.at(ns("./uri[@type = 'citation']"))
 
         href = suffix_url(url.text)
         anchor = node&.at(ns(".//locality[@type = 'anchor']"))&.text&.strip

data/lib/isodoc/function/utils.rb

@@ -211,7 +211,7 @@ module IsoDoc
         c = HTMLEntities.new
         text.split(/([<>])/).each_slice(4).map do |a|
           a[0] = c.encode(c.decode(a[0]), :hexadecimal)
-          a[1] = c.encode(c.decode(a[1]), :hexadecimal)
+          a[2] = c.encode(c.decode(a[2]), :hexadecimal) if a.size >= 3
           a
         end.join
       end

data/lib/isodoc/html_function/html.rb

@@ -5,6 +5,7 @@ module IsoDoc
   module HtmlFunction
     module Html
       def convert1(docxml, filename, dir)
+        bibitem_lookup(docxml)
         noko do |xml|
           xml.html **{ lang: @lang.to_s } do |html|
             info docxml, nil

data/lib/isodoc/html_function/postprocess.rb

@@ -1,5 +1,6 @@
 require "isodoc/html_function/mathvariant_to_plain"
 require_relative "postprocess_footnotes"
+require "metanorma-utils"
 
 module IsoDoc
   module HtmlFunction
@@ -74,7 +75,8 @@ module IsoDoc
         head = docxml.at("//*[local-name() = 'head']")
         head << htmlstylesheet(@htmlstylesheet)
         s = htmlstylesheet(@htmlstylesheet_override) and head << s
-        @bare and head << "<style>body {margin-left: 2em; margin-right: 2em;}</style>"
+        @bare and
+          head << "<style>body {margin-left: 2em; margin-right: 2em;}</style>"
         docxml
       end
 
@@ -167,11 +169,7 @@ module IsoDoc
       end
 
       def datauri(img)
-        type = img["src"].split(".")[-1]
-        supertype = type == "xml" ? "application" : "image"
-        bin = IO.binread(image_localfile(img))
-        data = Base64.strict_encode64(bin)
-        img["src"] = "data:#{supertype}/#{type};base64,#{data}"
+        img["src"] = Metanorma::Utils::datauri(img["src"], @localdir)
       end
 
       def image_suffix(img)

data/lib/isodoc/metadata_date.rb

@@ -6,7 +6,7 @@ module IsoDoc
                    vote-ended}.freeze
 
     def months
-    {
+      {
         "01": @labels["month_january"],
         "02": @labels["month_february"],
         "03": @labels["month_march"],
@@ -19,30 +19,32 @@ module IsoDoc
         "10": @labels["month_october"],
         "11": @labels["month_november"],
         "12": @labels["month_december"],
-    }
+      }
     end
 
     def monthyr(isodate)
       m = /(?<yr>\d\d\d\d)-(?<mo>\d\d)/.match isodate
       return isodate unless m && m[:yr] && m[:mo]
+
       l10n("#{months[m[:mo].to_sym]} #{m[:yr]}",
-                                   @lang, @script)
+           @lang, @script)
     end
 
     def MMMddyyyy(isodate)
       return nil if isodate.nil?
+
       arr = isodate.split("-")
-      date = if arr.size == 1 and (/^\d+$/.match isodate)
-               Date.new(*arr.map(&:to_i)).strftime("%Y")
-             elsif arr.size == 2
-               Date.new(*arr.map(&:to_i)).strftime("%B %Y")
-             else
-               Date.parse(isodate).strftime("%B %d, %Y")
-             end
+      if arr.size == 1 && (/^\d+$/.match isodate)
+        Date.new(*arr.map(&:to_i)).strftime("%Y")
+      elsif arr.size == 2
+        Date.new(*arr.map(&:to_i)).strftime("%B %Y")
+      else
+        Date.parse(isodate).strftime("%B %d, %Y")
+      end
     end
 
     def bibdate(isoxml, _out)
-      isoxml.xpath(ns('//bibdata/date')).each do |d|
+      isoxml.xpath(ns("//bibdata/date")).each do |d|
         set("#{d['type'].gsub(/-/, '_')}date".to_sym, Common::date_range(d))
       end
     end

data/lib/isodoc/presentation_function/bibdata.rb

@@ -10,7 +10,7 @@ module IsoDoc
         "</localized-strings>"
     end
 
-     def docid_prefixes(docxml)
+    def docid_prefixes(docxml)
       docxml.xpath(ns("//references/bibitem/docidentifier")).each do |i|
         i.children = @xrefs.klass.docid_prefix(i["type"], i.text)
       end
@@ -97,7 +97,7 @@ module IsoDoc
 
     # https://stackoverflow.com/a/31822406
     def blank?(elem)
-      elem.nil? || elem.respond_to?(:empty?) && elem.empty?
+      elem.nil? || (elem.respond_to?(:empty?) && elem.empty?)
     end
 
     def trim_hash(hash)

data/lib/isodoc/presentation_function/block.rb

@@ -54,12 +54,6 @@ module IsoDoc
       end
     end
 
-    def termexample(docxml)
-      docxml.xpath(ns("//termexample")).each do |f|
-        example1(f)
-      end
-    end
-
     def example1(elem)
       n = @xrefs.get[elem["id"]]
       lbl = if n.nil? || n[:label].nil? || n[:label].empty?
@@ -89,36 +83,6 @@ module IsoDoc
       prefix_name(elem, "", lbl, "name")
     end
 
-    def termnote(docxml)
-      docxml.xpath(ns("//termnote")).each do |f|
-        termnote1(f)
-      end
-    end
-
-    # introduce name element
-    def termnote1(elem)
-      lbl = l10n(@xrefs.anchor(elem["id"], :label) || "???")
-      prefix_name(elem, "", lower2cap(lbl), "name")
-    end
-
-    def termdefinition(docxml)
-      docxml.xpath(ns("//term[definition]")).each do |f|
-        termdefinition1(f)
-      end
-    end
-
-    def termdefinition1(elem)
-      return unless elem.xpath(ns("./definition")).size > 1
-
-      d = elem.at(ns("./definition"))
-      d = d.replace("<ol><li>#{d.children.to_xml}</li></ol>").first
-      elem.xpath(ns("./definition")).each do |f|
-        f = f.replace("<li>#{f.children.to_xml}</li>").first
-        d << f
-      end
-      d.wrap("<definition></definition>")
-    end
-
     def recommendation(docxml)
       docxml.xpath(ns("//recommendation")).each do |f|
         recommendation1(f, lower2cap(@i18n.recommendation))

data/lib/isodoc/presentation_function/inline.rb

@@ -1,5 +1,4 @@
 require "metanorma-utils"
-require_relative "./concept"
 
 module IsoDoc
   class PresentationXMLConvert < ::IsoDoc::Convert
@@ -7,13 +6,16 @@ module IsoDoc
       l10n("#{@xrefs.anchor(container, :xref)}, #{linkend}")
     end
 
+    def anchor_value(id)
+      @xrefs.anchor(id, :value) || @xrefs.anchor(id, :label) ||
+        @xrefs.anchor(id, :xref)
+    end
+
     def anchor_linkend(node, linkend)
       if node["citeas"].nil? && node["bibitemid"]
         return @xrefs.anchor(node["bibitemid"], :xref) || "???"
       elsif node["target"] && node["droploc"]
-        return @xrefs.anchor(node["target"], :value) ||
-            @xrefs.anchor(node["target"], :label) ||
-            @xrefs.anchor(node["target"], :xref) || "???"
+        return anchor_value(node["target"]) || "???"
       elsif node["target"] && !/.#./.match(node["target"])
         linkend = anchor_linkend1(node)
       end
@@ -27,14 +29,15 @@ module IsoDoc
       (container && get_note_container_id(node) != container &&
        @xrefs.get[node["target"]]) and
         linkend = prefix_container(container, linkend, node["target"])
-      capitalise_xref(node, linkend)
+      capitalise_xref(node, linkend, anchor_value(node["target"]))
     end
 
-    def capitalise_xref(node, linkend)
-      return linkend unless %w(Latn Cyrl Grek).include? @script
-      return linkend&.capitalize if node["case"] == "capital"
-      return linkend&.downcase if node["case"] == "lowercase"
-      return linkend if linkend[0, 1].match?(/\p{Upper}/)
+    def capitalise_xref(node, linkend, label)
+      linktext = linkend.gsub(/<[^>]+>/, "")
+      (label && !label.empty? && /^#{Regexp.escape(label)}/.match?(linktext)) ||
+        linktext[0, 1].match?(/\p{Upper}/) and return linkend
+      node["case"] and
+        return Common::case_with_markup(linkend, node["case"], @script)
 
       capitalise_xref1(node, linkend)
     end
@@ -43,7 +46,7 @@ module IsoDoc
       prec = nearest_block_parent(node).xpath("./descendant-or-self::text()") &
         node.xpath("./preceding::text()")
       if prec.empty? || /(?!<[^.].)\.\s+$/.match(prec.map(&:text).join)
-        linkend&.capitalize
+        Common::case_with_markup(linkend, "capital", @script)
       else linkend
       end
     end

data/lib/isodoc/presentation_function/terms.rb

@@ -0,0 +1,179 @@
+module IsoDoc
+  class PresentationXMLConvert < ::IsoDoc::Convert
+    def concept(docxml)
+      docxml.xpath(ns("//concept")).each { |f| concept1(f) }
+    end
+
+    def concept1(node)
+      xref = node&.at(ns("./xref/@target"))&.text or
+        return concept_render(node, ital: node["ital"] || "true",
+                                    ref: node["ref"] || "true",
+                                    linkref: node["linkref"] || "true",
+                                    linkmention: node["linkmention"] || "false")
+      if node.at(ns("//definitions//dt[@id = '#{xref}']"))
+        concept_render(node, ital: node["ital"] || "false",
+                             ref: node["ref"] || "false",
+                             linkref: node["linkref"] || "true",
+                             linkmention: node["linkmention"] || "false")
+      else concept_render(node, ital: node["ital"] || "true",
+                                ref: node["ref"] || "true",
+                                linkref: node["linkref"] || "true",
+                                linkmention: node["linkmention"] || "false")
+      end
+    end
+
+    def concept_render(node, opts)
+      node&.at(ns("./refterm"))&.remove
+      r = node.at(ns("./renderterm"))
+      ref = node.at(ns("./xref | ./eref | ./termref"))
+      ref && opts[:ref] != "false" and r&.next = " "
+      opts[:ital] == "true" and r&.name = "em"
+      if opts[:linkmention] == "true" && !r.nil? && !ref.nil?
+        ref2 = ref.clone
+        r2 = r.clone
+        r.replace(ref2).children = r2
+      end
+      concept1_ref(node, ref, opts)
+      if opts[:ital] == "false"
+        r = node.at(ns(".//renderterm"))
+        r&.replace(r&.children)
+      end
+      node.replace(node.children)
+    end
+
+    def concept1_ref(_node, ref, opts)
+      ref.nil? and return
+      return ref.remove if opts[:ref] == "false"
+
+      r = concept1_ref_content(ref)
+      ref = r.at("./descendant-or-self::xmlns:xref | "\
+                 "./descendant-or-self::xmlns:eref | "\
+                 "./descendant-or-self::xmlns:termref")
+      %w(xref eref).include? ref&.name and get_linkend(ref)
+      if opts[:linkref] == "false" && %w(xref eref).include?(ref&.name)
+        ref.replace(ref.children)
+      end
+    end
+
+    def concept1_ref_content(ref)
+      if non_locality_elems(ref).select do |c|
+           !c.text? || /\S/.match(c)
+         end.empty?
+        ref.replace(@i18n.term_defined_in.sub(/%/,
+                                              ref.to_xml))
+      else ref.replace("[#{ref.to_xml}]")
+      end
+    end
+
+    def related(docxml)
+      docxml.xpath(ns("//related")).each { |f| related1(f) }
+    end
+
+    def related1(node)
+      p = node.at(ns("./preferred"))
+      ref = node.at(ns("./xref | ./eref | ./termref"))
+      label = @i18n.relatedterms[node["type"]].upcase
+      node.replace(l10n("<p><strong>#{label}:</strong> "\
+                        "<em>#{p.to_xml}</em> (#{ref.to_xml})</p>"))
+    end
+
+    def designation(docxml)
+      docxml.xpath(ns("//preferred | //admitted | //deprecates")).each do |p|
+        designation1(p)
+      end
+      docxml.xpath(ns("//term")).each do |t|
+        merge_second_preferred(t)
+      end
+    end
+
+    def merge_second_preferred(term)
+      pref = nil
+      term.xpath(ns("./preferred")).each_with_index do |p, i|
+        if i.zero? then pref = p
+        else
+          pref << l10n("; #{p.children.to_xml}")
+          p.remove
+        end
+      end
+    end
+
+    def designation1(desgn)
+      s = desgn.at(ns("./termsource"))
+      name = desgn.at(ns("./expression/name | ./letter-symbol/name | "\
+                         "./graphical-symbol")) or return
+
+      g = desgn.at(ns("./expression/grammar")) and
+        name << " #{designation_grammar(g).join(', ')}"
+      desgn.children = name.children
+      s and desgn.next = s
+    end
+
+    def designation_grammar(grammar)
+      ret = []
+      grammar.xpath(ns("./gender")).each do |x|
+        ret << @i18n.grammar_abbrevs[x.text]
+      end
+      %w(isPreposition isParticiple isAdjective isVerb isAdverb isNoun)
+        .each do |x|
+        grammar.at(ns("./#{x}[text() = 'true']")) and
+          ret << @i18n.grammar_abbrevs[x]
+      end
+      ret
+    end
+
+        def definition1(elem)
+          nodes = Nokogiri::XML::NodeSet.new(elem.document)
+          v = elem&.at(ns("./verbaldefinition"))&.children and nodes += v
+          n = elem&.at(ns("./nonverbalrepresentation"))&.children and nodes += n
+          elem.children = nodes
+        end
+
+            def termexample(docxml)
+      docxml.xpath(ns("//termexample")).each do |f|
+        example1(f)
+      end
+    end
+
+                def termnote(docxml)
+      docxml.xpath(ns("//termnote")).each do |f|
+        termnote1(f)
+      end
+    end
+
+    # introduce name element
+    def termnote1(elem)
+      lbl = l10n(@xrefs.anchor(elem["id"], :label) || "???")
+      prefix_name(elem, "", lower2cap(lbl), "name")
+    end
+
+    def termdefinition(docxml)
+      docxml.xpath(ns("//term[definition]")).each do |f|
+        termdefinition1(f)
+      end
+    end
+
+    def termdefinition1(elem)
+      unwrap_definition(elem)
+      multidef(elem) if elem.xpath(ns("./definition")).size > 1
+      end
+
+    def multidef(elem)
+      d = elem.at(ns("./definition"))
+      d = d.replace("<ol><li>#{d.children.to_xml}</li></ol>").first
+      elem.xpath(ns("./definition")).each do |f|
+        f = f.replace("<li>#{f.children.to_xml}</li>").first
+        d << f
+      end
+      d.wrap("<definition></definition>")
+    end
+
+           def unwrap_definition(elem)
+             elem.xpath(ns("./definition")).each do |d|
+          nodes = Nokogiri::XML::NodeSet.new(elem.document)
+          v = d&.at(ns("./verbaldefinition"))&.children and nodes += v
+          n = d&.at(ns("./nonverbalrepresentation"))&.children and nodes += n
+          d.children = nodes
+             end
+        end
+  end
+end

data/lib/isodoc/presentation_xml_convert.rb

@@ -1,4 +1,5 @@
 require_relative "presentation_function/block"
+require_relative "presentation_function/terms"
 require_relative "presentation_function/inline"
 require_relative "presentation_function/math"
 require_relative "presentation_function/section"
@@ -25,6 +26,7 @@ module IsoDoc
       @xrefs.parse docxml
       section docxml
       block docxml
+      terms docxml
       inline docxml
     end
 
@@ -46,17 +48,13 @@ module IsoDoc
       sourcecode docxml
       formula docxml
       example docxml
-      termexample docxml
       note docxml
-      termnote docxml
-      termdefinition docxml
       permission docxml
       requirement docxml
       recommendation docxml
     end
 
     def inline(docxml)
-      concept docxml
       xref docxml
       eref docxml
       origin docxml
@@ -65,6 +63,15 @@ module IsoDoc
       variant docxml
     end
 
+    def terms(docxml)
+      termexample docxml
+      termnote docxml
+      termdefinition docxml
+      designation docxml
+      concept docxml
+      related docxml
+    end
+
     def postprocess(result, filename, _dir)
       toXML(result, filename)
       @files_to_delete.each { |f| FileUtils.rm_rf f }

data/lib/isodoc/version.rb

@@ -1,3 +1,3 @@
 module IsoDoc
-  VERSION = "1.7.6".freeze
+  VERSION = "1.8.1".freeze
 end

data/lib/isodoc/word_function/body.rb

@@ -54,11 +54,11 @@ module IsoDoc
       end
 
       def para_class(_node)
-        classtype = nil
-        classtype = "Note" if @note
-        classtype = "MsoCommentText" if in_comment
-        classtype = "Sourcecode" if @annotation
-        classtype
+        return "Sourcecode" if @annotation
+        return "MsoCommentText" if @in_comment
+        return "Note" if @note
+
+        nil
       end
 
       def para_parse(node, out)
@@ -86,21 +86,31 @@ module IsoDoc
       end
 
       def dl_parse(node, out)
+        return super unless node.ancestors("table, dl").empty?
+
+        dl_parse_table(node, out)
+      end
+
+      def dl_parse_table(node, out)
         out.table **{ class: "dl" } do |v|
           node.elements.select { |n| dt_dd? n }.each_slice(2) do |dt, dd|
-            v.tr do |tr|
-              tr.td **{ valign: "top", align: "left" } do |term|
-                dt_parse(dt, term)
-              end
-              tr.td **{ valign: "top" } do |listitem|
-                dd.children.each { |n| parse(n, listitem) }
-              end
-            end
+            dl_parse_table1(v, dt, dd)
           end
           dl_parse_notes(node, v)
         end
       end
 
+      def dl_parse_table1(table, dterm, ddefn)
+        table.tr do |tr|
+          tr.td **{ valign: "top", align: "left" } do |term|
+            dt_parse(dterm, term)
+          end
+          tr.td **{ valign: "top" } do |listitem|
+            ddefn.children.each { |n| parse(n, listitem) }
+          end
+        end
+      end
+
       def dl_parse_notes(node, out)
         return if node.elements.reject { |n| dt_dd? n }.empty?
 
@@ -120,8 +130,8 @@ module IsoDoc
         dl
       end
 
+      # get rid of footnote link, it is in diagram
       def figure_aside_process(fig, aside, key)
-        # get rid of footnote link, it is in diagram
         fig&.at("./a[@class='TableFootnoteRef']")&.remove
         fnref = fig.at(".//span[@class='TableFootnoteRef']/..")
         tr = key.add_child("<tr></tr>").first

data/lib/isodoc/word_function/comments.rb

@@ -1,10 +1,6 @@
 module IsoDoc
   module WordFunction
     module Comments
-      def in_comment
-        @in_comment
-      end
-
       def comments(div)
         return if @comments.empty?