metanorma-standoc 2.1.0 → 2.1.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1e7285f74dd3b2b9ab155b8bf7e334aeb45f453a20f1d4eaf94f39f42b9251d3
-  data.tar.gz: fdbad7c4cd96de56faacf360fdedaa0a0a5d24313c34b68ec2ab8e152db6a7c2
+  metadata.gz: 322efef58275c5330db01582e8ead5ea9c5c496e8a6c5e96aafd8c8413a4412f
+  data.tar.gz: f88ff574288ee6159fd0e4b05eb076767daede15afd3187d2de05dc1aae295bf
 SHA512:
-  metadata.gz: ce9033847b44804a9ba150db5e4840e1d3dafef077a9c5af52defd74321333c66a3053abc589ae02ee3c28d6fe42afbd568cdef8b1a529e5b2a97ea46ec95ae0
-  data.tar.gz: 6f498786e8222646f72fe88193797b0b0a23c30d1478cf78fad27862f19ece9c054ebe72c891ad671f5c03b4c8f06a659195cc77f359f517b1ffea02203ad22b
+  metadata.gz: f88c019751884a354f2217c7f67798365e4dfa4fdf15f573f8fa95d98b65abb483da40e6a5f8bd6c1fedb76a8d5d1fe72417cff24b94224bd5ecb625389dd926
+  data.tar.gz: 4eaa00800dc54f99af38e012923472d50215308e8ba96d2ebe8414ffb9e38c266a0de29e18fc80ba2536d82c7f47ea0df8bea8b892d7610c7d67b46d088bf430

data/lib/metanorma/standoc/base.rb

@@ -171,7 +171,9 @@ module Metanorma
       end
 
       def doctype(node)
-        node.attr("doctype")&.gsub(/\s+/, "-")&.downcase
+        ret = node.attr("doctype")&.gsub(/\s+/, "-")&.downcase || "standard"
+        ret = "standard" if ret == "article"
+        ret
       end
 
       def front(node, xml)

data/lib/metanorma/standoc/biblio.rng

@@ -614,12 +614,103 @@
     <optional>
       <ref name="fetched"/>
     </optional>
-    <choice>
-      <oneOrMore>
-        <ref name="btitle"/>
-      </oneOrMore>
+    <optional>
       <ref name="formattedref"/>
-    </choice>
+    </optional>
+    <oneOrMore>
+      <ref name="btitle"/>
+    </oneOrMore>
+    <zeroOrMore>
+      <ref name="bsource"/>
+    </zeroOrMore>
+    <oneOrMore>
+      <ref name="docidentifier"/>
+    </oneOrMore>
+    <optional>
+      <ref name="docnumber"/>
+    </optional>
+    <zeroOrMore>
+      <ref name="bdate"/>
+    </zeroOrMore>
+    <zeroOrMore>
+      <ref name="contributor"/>
+    </zeroOrMore>
+    <optional>
+      <ref name="edition"/>
+    </optional>
+    <zeroOrMore>
+      <ref name="version"/>
+    </zeroOrMore>
+    <zeroOrMore>
+      <ref name="biblionote"/>
+    </zeroOrMore>
+    <zeroOrMore>
+      <ref name="language"/>
+    </zeroOrMore>
+    <zeroOrMore>
+      <ref name="script"/>
+    </zeroOrMore>
+    <zeroOrMore>
+      <ref name="bibabstract"/>
+    </zeroOrMore>
+    <optional>
+      <ref name="status"/>
+    </optional>
+    <zeroOrMore>
+      <ref name="copyright"/>
+    </zeroOrMore>
+    <zeroOrMore>
+      <ref name="docrelation"/>
+    </zeroOrMore>
+    <zeroOrMore>
+      <ref name="series"/>
+    </zeroOrMore>
+    <optional>
+      <ref name="medium"/>
+    </optional>
+    <zeroOrMore>
+      <ref name="bplace"/>
+    </zeroOrMore>
+    <zeroOrMore>
+      <ref name="bprice"/>
+    </zeroOrMore>
+    <zeroOrMore>
+      <ref name="extent"/>
+    </zeroOrMore>
+    <optional>
+      <ref name="bibliographic_size"/>
+    </optional>
+    <zeroOrMore>
+      <ref name="accesslocation"/>
+    </zeroOrMore>
+    <zeroOrMore>
+      <ref name="license"/>
+    </zeroOrMore>
+    <zeroOrMore>
+      <ref name="bclassification"/>
+    </zeroOrMore>
+    <zeroOrMore>
+      <ref name="bkeyword"/>
+    </zeroOrMore>
+    <optional>
+      <ref name="validity"/>
+    </optional>
+  </define>
+  <define name="ReducedBibliographicItem">
+    <optional>
+      <attribute name="type">
+        <ref name="BibItemType"/>
+      </attribute>
+    </optional>
+    <optional>
+      <ref name="fetched"/>
+    </optional>
+    <optional>
+      <ref name="formattedref"/>
+    </optional>
+    <zeroOrMore>
+      <ref name="btitle"/>
+    </zeroOrMore>
     <zeroOrMore>
       <ref name="bsource"/>
     </zeroOrMore>
@@ -638,9 +729,9 @@
     <optional>
       <ref name="edition"/>
     </optional>
-    <optional>
+    <zeroOrMore>
       <ref name="version"/>
-    </optional>
+    </zeroOrMore>
     <zeroOrMore>
       <ref name="biblionote"/>
     </zeroOrMore>
@@ -833,6 +924,12 @@
           <data type="boolean"/>
         </attribute>
       </optional>
+      <optional>
+        <attribute name="language"/>
+      </optional>
+      <optional>
+        <attribute name="script"/>
+      </optional>
       <text/>
     </element>
   </define>
@@ -986,36 +1083,34 @@
           <ref name="SeriesType"/>
         </attribute>
       </optional>
-      <choice>
+      <optional>
         <ref name="formattedref"/>
-        <group>
-          <ref name="btitle"/>
-          <optional>
-            <ref name="bplace"/>
-          </optional>
-          <optional>
-            <ref name="seriesorganization"/>
-          </optional>
-          <optional>
-            <ref name="abbreviation"/>
-          </optional>
-          <optional>
-            <ref name="seriesfrom"/>
-          </optional>
-          <optional>
-            <ref name="seriesto"/>
-          </optional>
-          <optional>
-            <ref name="seriesnumber"/>
-          </optional>
-          <optional>
-            <ref name="seriespartnumber"/>
-          </optional>
-          <optional>
-            <ref name="seriesrun"/>
-          </optional>
-        </group>
-      </choice>
+      </optional>
+      <ref name="btitle"/>
+      <optional>
+        <ref name="bplace"/>
+      </optional>
+      <optional>
+        <ref name="seriesorganization"/>
+      </optional>
+      <optional>
+        <ref name="abbreviation"/>
+      </optional>
+      <optional>
+        <ref name="seriesfrom"/>
+      </optional>
+      <optional>
+        <ref name="seriesto"/>
+      </optional>
+      <optional>
+        <ref name="seriesnumber"/>
+      </optional>
+      <optional>
+        <ref name="seriespartnumber"/>
+      </optional>
+      <optional>
+        <ref name="seriesrun"/>
+      </optional>
     </element>
   </define>
   <define name="SeriesType">
@@ -1174,7 +1269,7 @@
         </element>
       </optional>
       <element name="bibitem">
-        <ref name="BibliographicItem"/>
+        <ref name="ReducedBibliographicItem"/>
       </element>
       <choice>
         <zeroOrMore>
@@ -1199,9 +1294,9 @@
       <optional>
         <ref name="revision-date"/>
       </optional>
-      <zeroOrMore>
+      <optional>
         <ref name="draft"/>
-      </zeroOrMore>
+      </optional>
     </element>
   </define>
   <define name="vedition">

data/lib/metanorma/standoc/blocks.rb

@@ -38,9 +38,7 @@ module Metanorma
         role == "form" and return form(node)
         role == "definition" and return termdefinition(node)
         result = []
-        node.blocks.each do |b|
-          result << send(b.context, b)
-        end
+        node.blocks.each { |b| result << send(b.context, b) }
         result
       end
 
@@ -80,14 +78,17 @@ module Metanorma
 
       def term_example(node)
         noko do |xml|
-          xml.termexample **attr_code(id_attr(node)) do |ex|
+          xml.termexample **attr_code(id_attr(node)
+            .merge(
+              keepasterm: node.option?("termexample") || nil,
+            )) do |ex|
             wrap_in_para(node, ex)
           end
         end.join("\n")
       end
 
       def example(node)
-        return term_example(node) if in_terms?
+        return term_example(node) if in_terms? || node.option?("termexample")
 
         role = node.role || node.attr("style")
         %w(recommendation requirement permission).include?(role) and
@@ -237,10 +238,11 @@ module Metanorma
       end
 
       def pass(node)
+        c = HTMLEntities.new
         noko do |xml|
           xml.passthrough **attr_code(formats:
                                       node.attr("format") || "metanorma") do |p|
-            p << HTMLEntities.new.encode(node.content, :basic, :hexadecimal)
+            p << c.encode(c.decode(node.content), :basic, :hexadecimal)
           end
         end
       end

data/lib/metanorma/standoc/blocks_notes.rb

@@ -8,6 +8,7 @@ module Metanorma
             number: node.attr("number"),
             subsequence: node.attr("subsequence"),
             "keep-separate": node.attr("keep-separate"),
+            keepasterm: node.option?("termnote") ? "true" : nil,
           )))
       end
 
@@ -61,6 +62,8 @@ module Metanorma
       end
 
       def note(node)
+        return termnote(node) if node.option?("termnote")
+
         noko do |xml|
           xml.note **note_attrs(node) do |c|
             wrap_in_para(node, c)

data/lib/metanorma/standoc/cleanup.rb

@@ -25,6 +25,7 @@ module Metanorma
     module Cleanup
       def cleanup(xmldoc)
         element_name_cleanup(xmldoc)
+        passthrough_cleanup(xmldoc)
         sections_cleanup(xmldoc)
         obligations_cleanup(xmldoc)
         table_cleanup(xmldoc)

data/lib/metanorma/standoc/cleanup_footnotes.rb

@@ -5,12 +5,15 @@ require "json"
 module Metanorma
   module Standoc
     module Cleanup
-      def footnote_content(fn)
-        c = fn.children.respond_to?(:to_xml) ? fn.children.to_xml : fn.children
+      def footnote_content(fnote)
+        c = if fnote.children.respond_to?(:to_xml)
+              fnote.children.to_xml
+            else fn.children
+            end
         c.gsub(/ id="[^"]+"/, "")
       end
 
-      # include footnotes inside figure if they are the only content 
+      # include footnotes inside figure if they are the only content
       # of the paras following
       def figure_footnote_cleanup(xmldoc)
         nomatches = false
@@ -18,7 +21,9 @@ module Metanorma
           q = "//figure/following-sibling::*[1][self::p and *[1][self::fn]]"
           nomatches = true
           xmldoc.xpath(q).each do |s|
-            next if s.children.map { |c| c.text? && /[[:alpha:]]/.match(c.text) }.any?
+            next if s.children.map do |c|
+                      c.text? && /[[:alpha:]]/.match(c.text)
+                    end.any?
 
             s.previous_element << s.first_element_child.remove
             s.remove
@@ -27,16 +32,16 @@ module Metanorma
         end
       end
 
-      def table_footnote_renumber1(fn, i, seen)
-        content = footnote_content(fn)
+      def table_footnote_renumber1(fnote, i, seen)
+        content = footnote_content(fnote)
         if seen[content] then outnum = seen[content]
         else
           i += 1
           outnum = i
           seen[content] = outnum
         end
-        fn["reference"] = (outnum - 1 + "a".ord).chr
-        fn["table"] = true
+        fnote["reference"] = (outnum - 1 + "a".ord).chr
+        fnote["table"] = true
         [i, seen]
       end
 
@@ -84,7 +89,7 @@ module Metanorma
 
       def footnote_block_cleanup(xmldoc)
         xmldoc.xpath("//footnoteblock").each do |f|
-          f.name = 'fn'
+          f.name = "fn"
           if id = xmldoc.at("//*[@id = '#{f.text}']")
             f.children = id.remove.children
           else

data/lib/metanorma/standoc/cleanup_inline.rb

@@ -185,6 +185,23 @@ module Metanorma
         Digest::MD5.hexdigest("#{elem.path}////#{elem.text}")
           .sub(/^(.{8})(.{4})(.{4})(.{4})(.{12})$/, "_\\1-\\2-\\3-\\4-\\5")
       end
+
+      def passthrough_cleanup(doc)
+        doc.xpath("//passthrough-inline").each do |p|
+          p.name = "passthrough"
+          p.children = select_odd_chars(p.children.to_xml)
+        end
+        doc.xpath("//identifier").each do |p|
+          p.children = select_odd_chars(p.children.to_xml)
+        end
+      end
+
+      private
+
+      # skip ZWNJ inserted to prevent regexes operating in asciidoctor
+      def select_odd_chars(text)
+        text.gsub(/(?!&)([[:punct:]])\u200c/, "\\1")
+      end
     end
   end
 end

data/lib/metanorma/standoc/cleanup_reqt.rb

@@ -53,7 +53,7 @@ module Metanorma
                    permission).include?(text.name))) ||
           (text.text.strip.empty? && !text.at(".//xref | .//eref | .//link"))
 
-        t = Nokogiri::XML::Element.new("description", reqt)
+        t = Nokogiri::XML::Element.new("description", reqt.document)
         text.before(t)
         t.children = text.remove
       end

data/lib/metanorma/standoc/cleanup_terms.rb

@@ -93,7 +93,9 @@ module Metanorma
       def termnote_example_cleanup(xmldoc)
         %w(note example).each do |w|
           xmldoc.xpath("//term#{w}[not(ancestor::term)]").each do |x|
-            x.name = w
+            if x["keepasterm"] then x.delete("keepasterm")
+            else x.name = w
+            end
           end
         end
       end

data/lib/metanorma/standoc/cleanup_text.rb

@@ -5,8 +5,11 @@ module Metanorma
         text = result.flatten.map { |l| l.sub(/\s*\Z/, "") } * "\n"
         !@keepasciimath and text = asciimath2mathml(text)
         text = text.gsub(/\s+<fn /, "<fn ")
-        text.gsub(%r{<passthrough\s+formats="metanorma">([^<]*)
-                  </passthrough>}mx) { HTMLEntities.new.decode($1) }
+        %w(passthrough passthrough-inline).each do |v|
+          text.gsub!(%r{<#{v}\s+formats="metanorma">([^<]*)
+                    </#{v}>}mx) { HTMLEntities.new.decode($1) }
+        end
+        text
       end
 
       def smartquotes_cleanup(xmldoc)
@@ -33,7 +36,8 @@ module Metanorma
 
       def uninterrupt_quotes_around_xml_skip(elem)
         !(/\A['"]/.match?(elem.text) &&
-          elem.previous.ancestors("pre, tt, sourcecode, stem, figure, bibdata")
+          elem.previous.ancestors("pre, tt, sourcecode, stem, figure, bibdata,
+                                  passthrough, identifer")
           .empty? &&
           ((elem.previous.text.strip.empty? &&
             !empty_tag_with_text_content?(elem.previous)) ||
@@ -70,7 +74,8 @@ module Metanorma
           empty_tag_with_text_content?(x) and prev = "dummy"
           next unless x.text?
 
-          x.ancestors("pre, tt, sourcecode, stem, figure, bibdata").empty? and
+          x.ancestors("pre, tt, sourcecode, stem, figure, bibdata, passthrough,
+                      identifier").empty? and
             dumb2smart_quotes1(x, prev)
           prev = x.text if x.ancestors("index").empty?
         end

data/lib/metanorma/standoc/cleanup_xref.rb

@@ -68,9 +68,10 @@ module Metanorma
       end
 
       def xref_to_eref(elem)
+        c = HTMLEntities.new
         elem["bibitemid"] = elem["target"]
         if ref = @anchors&.dig(elem["target"], :xref)
-          elem["citeas"] = HTMLEntities.new.encode(ref, :hexadecimal)
+          elem["citeas"] = c.decode(ref)
         else
           elem["citeas"] = ""
           xref_to_eref1(elem)

data/lib/metanorma/standoc/converter.rb

@@ -36,6 +36,7 @@ module Metanorma
         inline_macro Metanorma::Standoc::DomainTermInlineMacro
         inline_macro Metanorma::Standoc::InheritInlineMacro
         inline_macro Metanorma::Standoc::HTML5RubyMacro
+        inline_macro Metanorma::Standoc::IdentifierInlineMacro
         inline_macro Metanorma::Standoc::ConceptInlineMacro
         inline_macro Metanorma::Standoc::AutonumberInlineMacro
         inline_macro Metanorma::Standoc::VariantInlineMacro
@@ -63,6 +64,7 @@ module Metanorma
         block Metanorma::Plugin::Lutaml::LutamlDiagramBlock
         block Metanorma::Standoc::PseudocodeBlockMacro
         preprocessor Metanorma::Standoc::EmbedIncludeProcessor
+        preprocessor Metanorma::Standoc::NamedEscapePreprocessor
       end
 
       include ::Asciidoctor::Converter

data/lib/metanorma/standoc/inline.rb

@@ -79,7 +79,8 @@ module Metanorma
       end
 
       def inline_anchor_bibref(node)
-        eref_contents = (node.text || node.target || node.id)
+        eref_contents = HTMLEntities.new
+          .decode(node.text || node.target || node.id)
           &.sub(/^\[?([^\[\]]+?)\]?$/, "[\\1]")
         @refids << (node.target || node.id)
         noko do |xml|

data/lib/metanorma/standoc/isodoc.rng

@@ -282,6 +282,9 @@
             <ref name="MultilingualRenderingType"/>
           </attribute>
         </optional>
+        <optional>
+          <ref name="tname"/>
+        </optional>
         <oneOrMore>
           <ref name="ul_li"/>
         </oneOrMore>
@@ -324,6 +327,9 @@
             </choice>
           </attribute>
         </optional>
+        <optional>
+          <ref name="tname"/>
+        </optional>
         <oneOrMore>
           <ref name="li"/>
         </oneOrMore>
@@ -360,6 +366,9 @@
             <ref name="MultilingualRenderingType"/>
           </attribute>
         </optional>
+        <optional>
+          <ref name="tname"/>
+        </optional>
         <oneOrMore>
           <ref name="dt"/>
           <ref name="dd"/>
@@ -694,6 +703,9 @@
       <optional>
         <attribute name="tag"/>
       </optional>
+      <optional>
+        <attribute name="type"/>
+      </optional>
       <optional>
         <attribute name="multilingual-rendering">
           <ref name="MultilingualRenderingType"/>
@@ -729,6 +741,9 @@
         <optional>
           <attribute name="tag"/>
         </optional>
+        <optional>
+          <attribute name="type"/>
+        </optional>
         <optional>
           <attribute name="multilingual-rendering">
             <ref name="MultilingualRenderingType"/>
@@ -2050,6 +2065,7 @@
       <value>compare</value>
       <value>contrast</value>
       <value>see</value>
+      <value>seealso</value>
     </choice>
   </define>
   <define name="deprecates">

data/lib/metanorma/standoc/lists.rb

@@ -38,9 +38,8 @@ module Metanorma
 
         noko do |xml|
           xml.ul **ul_attrs(node) do |xml_ul|
-            node.items.each do |item|
-              ul_li(xml_ul, item)
-            end
+            list_caption(node, xml_ul)
+            node.items.each { |item| ul_li(xml_ul, item) }
           end
         end.join("\n")
       end
@@ -62,6 +61,7 @@ module Metanorma
       def olist(node)
         noko do |xml|
           xml.ol **ol_attrs(node) do |xml_ol|
+            list_caption(node, xml_ol)
             node.items.each { |item| li(xml_ol, item) }
           end
         end.join("\n")
@@ -98,6 +98,7 @@ module Metanorma
       def dlist(node)
         noko do |xml|
           xml.dl **dl_attrs(node) do |xml_dl|
+            list_caption(node, xml_dl)
             node.items.each do |terms, dd|
               dt(terms, xml_dl)
               dd(dd, xml_dl)
@@ -115,6 +116,10 @@ module Metanorma
           end
         end.join("\n")
       end
+
+      def list_caption(node, out)
+        a = node.title and out.name { |n| n << a }
+      end
     end
   end
 end

data/lib/metanorma/standoc/macros.rb

@@ -130,7 +130,7 @@ module Metanorma
 
       def process(parent, target, attrs)
         out = Asciidoctor::Inline.new(parent, :quoted, attrs["text"]).convert
-        %{<autonumber type=#{target}>#{out}</autonumber>}
+        %{<autonumber type='#{target}'>#{out}</autonumber>}
       end
     end
 
@@ -143,9 +143,9 @@ module Metanorma
         /^(?<lang>[^-]*)(?:-(?<script>.*))?$/ =~ target
         out = Asciidoctor::Inline.new(parent, :quoted, attrs["text"]).convert
         if script
-          %{<variant lang=#{lang} script=#{script}>#{out}</variant>}
+          %{<variant lang='#{lang}' script='#{script}'>#{out}</variant>}
         else
-          %{<variant lang=#{lang}>#{out}</variant>}
+          %{<variant lang='#{lang}'>#{out}</variant>}
         end
       end
     end
@@ -191,6 +191,7 @@ module Metanorma
       end
     end
 
+    # inject ZWNJ to prevent Asciidoctor from attempting regex substitutions
     class PassInlineMacro < Asciidoctor::Extensions::InlineMacroProcessor
       use_dsl
       named :"pass-format"
@@ -198,7 +199,21 @@ module Metanorma
       def process(parent, target, attrs)
         format = target || "metanorma"
         out = Asciidoctor::Inline.new(parent, :quoted, attrs[1]).convert
-        %{<passthrough formats="#{format}">#{out}</passthrough>}
+          .gsub(/((?![<>&])[[:punct:]])/, "\\1&#x200c;")
+        %{<passthrough-inline formats="#{format}">#{out}</passthrough-inline>}
+      end
+    end
+
+    class IdentifierInlineMacro < Asciidoctor::Extensions::InlineMacroProcessor
+      use_dsl
+      named :identifier
+      parse_content_as :raw
+      using_format :short
+
+      def process(parent, _target, attrs)
+        out = Asciidoctor::Inline.new(parent, :quoted, attrs["text"]).convert
+          .gsub(/((?![<>&])[[:punct:]])/, "\\1&#x200c;")
+        %{<identifier>#{out}</identifier>}
       end
     end
 
@@ -213,5 +228,17 @@ module Metanorma
                       type: :xref, target: "_#{UUIDTools::UUID.random_create}")
       end
     end
+
+    class NamedEscapePreprocessor < Asciidoctor::Extensions::Preprocessor
+      def process(_document, reader)
+        c = HTMLEntities.new
+        lines = reader.readlines.map do |l|
+          l.split(/(&[A-Za-z][^;]*;)/).map do |s|
+            /^&[A-Za-z]/.match?(s) ? c.encode(c.decode(s), :hexadecimal) : s
+          end.join
+        end
+        ::Asciidoctor::Reader.new lines
+      end
+    end
   end
 end

data/lib/metanorma/standoc/macros_form.rb

@@ -56,7 +56,7 @@ module Metanorma
         m = %w(disabled value)
           .map { |a| attr[a] ? " #{a}='#{attr[a]}'" : nil }.compact
         out = Asciidoctor::Inline.new(parent, :quoted, attr["text"]).convert
-        %{<option #{m.join}">#{out}</option>}
+        %{<option #{m.join}>#{out}</option>}
       end
     end
   end