Merge pull request #825 from metanorma/features/private-charset

Features/private charset

lib/metanorma/standoc/anchor.rb

@@ -0,0 +1,112 @@
+require "asciidoctor/extensions"
+
+module Metanorma
+  module Standoc
+    module Inline
+      def inline_anchor(node)
+        case node.type
+        when :ref then inline_anchor_ref node
+        when :xref then inline_anchor_xref node
+        when :link then inline_anchor_link node
+        when :bibref then inline_anchor_bibref node
+        end
+      end
+
+      def inline_anchor_ref(node)
+        noko do |xml|
+          xml.bookmark nil, **attr_code(id: node.id)
+        end.join
+      end
+
+      def inline_anchor_xref(node)
+        noko do |xml|
+          attrs = inline_anchor_xref_attrs(node)
+          c = attrs[:text]
+          attrs.delete(:text) unless c.nil?
+          xml.xref **attr_code(attrs) do |x|
+            x << c
+          end
+        end.join
+      end
+
+      def inline_anchor_xref_attrs(node)
+        text = concatenate_attributes_to_xref_text(node)
+        m = inline_anchor_xref_match(text)
+        t = node.target.gsub(/^#/, "").gsub(%r{(\.xml|\.adoc)(#.*$)}, "\\2")
+        m.nil? and return { target: t, type: "inline", text: text }
+        inline_anchor_xref_attrs1(m, t, text)
+      end
+
+      def concatenate_attributes_to_xref_text(node)
+        node.attributes.each_with_object([]) do |(k, v), m|
+          %w(path fragment refid).include?(k) and next
+          m << "#{k}=#{v}%"
+        end.map { |x| x.sub(/%+/, "%") }.join + (node.text || "")
+      end
+
+      def inline_anchor_xref_attrs1(match, target, text)
+        { target: target, hidden: match[:hidden],
+          type: match[:fn].nil? ? "inline" : "footnote",
+          case: match[:case]&.sub(/%$/, ""),
+          style: match[:style]&.sub(/^style=/, "")&.sub(/%$/, "") || @xrefstyle,
+          droploc: match[:drop].nil? && match[:drop2].nil? ? nil : true,
+          text: inline_anchor_xref_text(match, text) }
+      end
+
+      def inline_anchor_xref_match(text)
+        /^(?:hidden%(?<hidden>[^,]+),?)?
+          (?<style>style=[^%]+%)?
+          (?<drop>droploc%)?(?<case>capital%|lowercase%)?(?<drop2>droploc%)?
+          (?<fn>fn:?\s*)?(?<text>.*)$/x.match text
+      end
+
+      def inline_anchor_xref_text(match, text)
+        if %i[case fn drop drop2 hidden style].any? { |x| !match[x].nil? }
+          match[:text]
+        else text
+        end
+      end
+
+      def inline_anchor_link(node)
+        contents = node.text
+        contents = "" if node.target.gsub(%r{^mailto:}, "") == node.text
+        attributes = { target: node.target, alt: node.attr("title"),
+                       "update-type": node.attr("updatetype") ||
+                         node.attr("update-type") }
+        noko do |xml|
+          xml.link **attr_code(attributes) do |l|
+            l << contents
+          end
+        end.join
+      end
+
+      def inline_anchor_bibref(node)
+        eref_contents =
+          @c.decode(node.text || node.target || node.id)
+          &.sub(/^\[?([^\[\]]+?)\]?$/, "[\\1]")
+        @refids << (node.target || node.id)
+        noko do |xml|
+          xml.ref **attr_code(id: node.target || node.id) do |r|
+            r << eref_contents
+          end
+        end.join
+      end
+
+      def inline_callout(node)
+        noko do |xml|
+          xml.callout node.text
+        end.join
+      end
+
+      def inline_footnote(node)
+        @fn_number ||= 0
+        noko do |xml|
+          @fn_number += 1
+          xml.fn reference: @fn_number do |fn|
+            fn.p { |p| p << node.text }
+          end
+        end.join
+      end
+    end
+  end
+end

lib/metanorma/standoc/cleanup.rb

@@ -158,8 +158,7 @@ def variant_space_cleanup(xmldoc)
       end
 
       def metadata_cleanup(xmldoc)
-        return if @metadata_attrs.nil? || @metadata_attrs.empty?
-
+        (@metadata_attrs.nil? || @metadata_attrs.empty?) and return
         ins = add_misc_container(xmldoc)
         ins << @metadata_attrs
       end

lib/metanorma/standoc/cleanup_toc.rb

@@ -78,8 +78,7 @@ def toc_cleanup_clause_entry(xmldoc, list)
       end
 
       def toc_metadata(xmldoc)
-        return unless @htmltoclevels || @doctoclevels || @toclevels
-
+        @htmltoclevels || @doctoclevels || @toclevels or return
         ins = add_misc_container(xmldoc)
         toc_metadata1(ins)
       end

lib/metanorma/standoc/converter.rb

@@ -5,6 +5,7 @@
 require_relative "lists"
 require_relative "ref"
 require_relative "inline"
+require_relative "anchor"
 require_relative "blocks"
 require_relative "section"
 require_relative "table"

lib/metanorma/standoc/inline.rb

@@ -1,4 +1,3 @@
-require "asciidoctor/extensions"
 require "unicode2latex"
 require "mime/types"
 require "base64"
@@ -8,115 +7,6 @@
 module Metanorma
   module Standoc
     module Inline
-      def refid?(ref)
-        @refids.include? ref
-      end
-
-      def inline_anchor(node)
-        case node.type
-        when :ref then inline_anchor_ref node
-        when :xref then inline_anchor_xref node
-        when :link then inline_anchor_link node
-        when :bibref then inline_anchor_bibref node
-        end
-      end
-
-      def inline_anchor_ref(node)
-        noko do |xml|
-          xml.bookmark nil, **attr_code(id: node.id)
-        end.join
-      end
-
-      def inline_anchor_xref(node)
-        noko do |xml|
-          attrs = inline_anchor_xref_attrs(node)
-          c = attrs[:text]
-          attrs.delete(:text) unless c.nil?
-          xml.xref **attr_code(attrs) do |x|
-            x << c
-          end
-        end.join
-      end
-
-      def inline_anchor_xref_attrs(node)
-        text = concatenate_attributes_to_xref_text(node)
-        m = inline_anchor_xref_match(text)
-        t = node.target.gsub(/^#/, "").gsub(%r{(\.xml|\.adoc)(#.*$)}, "\\2")
-        m.nil? and return { target: t, type: "inline", text: text }
-        inline_anchor_xref_attrs1(m, t, text)
-      end
-
-      def concatenate_attributes_to_xref_text(node)
-        node.attributes.each_with_object([]) do |(k, v), m|
-          %w(path fragment refid).include?(k) and next
-          m << "#{k}=#{v}%"
-        end.map { |x| x.sub(/%+/, "%") }.join + (node.text || "")
-      end
-
-      def inline_anchor_xref_attrs1(match, target, text)
-        { target: target, hidden: match[:hidden],
-          type: match[:fn].nil? ? "inline" : "footnote",
-          case: match[:case]&.sub(/%$/, ""),
-          style: match[:style]&.sub(/^style=/, "")&.sub(/%$/, "") || @xrefstyle,
-          droploc: match[:drop].nil? && match[:drop2].nil? ? nil : true,
-          text: inline_anchor_xref_text(match, text) }
-      end
-
-      def inline_anchor_xref_match(text)
-        /^(?:hidden%(?<hidden>[^,]+),?)?
-          (?<style>style=[^%]+%)?
-          (?<drop>droploc%)?(?<case>capital%|lowercase%)?(?<drop2>droploc%)?
-          (?<fn>fn:?\s*)?(?<text>.*)$/x.match text
-      end
-
-      def inline_anchor_xref_text(match, text)
-        if %i[case fn drop drop2 hidden style].any? { |x| !match[x].nil? }
-          match[:text]
-        else text
-        end
-      end
-
-      def inline_anchor_link(node)
-        contents = node.text
-        contents = "" if node.target.gsub(%r{^mailto:}, "") == node.text
-        attributes = { target: node.target, alt: node.attr("title"),
-                       "update-type": node.attr("updatetype") ||
-                         node.attr("update-type") }
-        noko do |xml|
-          xml.link **attr_code(attributes) do |l|
-            l << contents
-          end
-        end.join
-      end
-
-      def inline_anchor_bibref(node)
-        eref_contents =
-          @c.decode(node.text || node.target || node.id)
-          &.sub(/^\[?([^\[\]]+?)\]?$/, "[\\1]")
-        @refids << (node.target || node.id)
-        noko do |xml|
-          xml.ref **attr_code(id: node.target || node.id) do |r|
-            r << eref_contents
-          end
-        end.join
-      end
-
-      def inline_callout(node)
-        noko do |xml|
-          xml.callout node.text
-        end.join
-      end
-
-      def inline_footnote(node)
-        @fn_number ||= 0
-        noko do |xml|
-          @fn_number += 1
-          xml.fn reference: @fn_number do |fn|
-            fn.p { |p| p << node.text }
-          end
-        end.join
-      end
-
       def inline_break(node)
         noko do |xml|
           xml << node.text
@@ -137,8 +27,8 @@ def thematic_break(_node)
 
       def latex_parse1(text, block)
         lxm_input = Unicode2LaTeX.unicode2latex(@c.decode(text))
-        results = Plurimath::Math.parse(lxm_input, "latex")
-          .to_mathml(display_style: block)
+        results = Plurimath::Math.parse(lxm_input, "latex").
+          to_mathml(display_style: block)
         if results.nil?
           @log.add("Math", nil,
                    "latexmlmath failed to process equation:\n#{lxm_input}")
@@ -164,8 +54,8 @@ def latex_parse(text, xml, block)
         latex = latex_parse1(text, block) or
           return xml.stem type: "MathML", block: block
         xml.stem type: "MathML", block: block do |s|
-          math = Nokogiri::XML.fragment(latex.sub(/<\?[^>]+>/, ""))
-            .elements[0]
+          math = Nokogiri::XML.fragment(latex.sub(/<\?[^>]+>/, "")).
+            elements[0]
           math.delete("alttext")
           s.parent.children = math
           s << "<latexmath>#{text}</latexmath>"
@@ -202,13 +92,28 @@ def inline_quoted(node)
               xml.span style: node.role.sub(/^css /, "") do |s|
                 s << node.text
               end
+            when /:/
+              xml.span **attr_code(hash2styles(node.role)) do |s|
+                s << node.text
+              end
             else
               xml << node.text
             end
           end
         end.join
       end
 
+      def hash2styles(role)
+        CSV.parse_line(role, liberal_parsing: true).
+          each_with_object({}) do |r, m|
+          kv = r.split(":", 2).map(&:strip)
+          case kv[0]
+          when "custom-charset"
+            m[kv[0]] = kv[1]
+          end
+        end
+      end
+
       def image_attributes(node)
         nodetarget = node.attr("target") || node.target
         if Gem.win_platform? && /^[a-zA-Z]:/.match?(nodetarget)

lib/metanorma/standoc/utils.rb

@@ -69,11 +69,11 @@ def link_unwrap(para)
       end
 
       def xml_encode(text)
-        @c.encode(text, :basic, :hexadecimal)
-          .gsub("&amp;gt;", ">").gsub("&amp;lt;", "<").gsub("&amp;amp;", "&")
-          .gsub("&gt;", ">").gsub("&lt;", "<").gsub("&amp;", "&")
-          .gsub("&quot;", '"').gsub("&#xa;", "\n").gsub("&amp;#", "&#")
-          .gsub("&apos;", "'")
+        @c.encode(text, :basic, :hexadecimal).
+          gsub("&amp;gt;", ">").gsub("&amp;lt;", "<").gsub("&amp;amp;", "&").
+          gsub("&gt;", ">").gsub("&lt;", "<").gsub("&amp;", "&").
+          gsub("&quot;", '"').gsub("&#xa;", "\n").gsub("&amp;#", "&#").
+          gsub("&apos;", "'")
       end
 
       # wrapped in <sections>
@@ -104,6 +104,10 @@ def grkletters(text)
                       psi|omega)\b/xi, "&\\1;")
       end
 
+      def refid?(ref)
+        @refids.include? ref
+      end
+
       module_function :adoc2xml
 
       class EmptyAttr