metanorma-standoc 2.4.5 → 2.4.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 8989224ddc5bb169f13c102066f005b7c72e0d6ff0cd7a2c2122ef57ed7f6923
-  data.tar.gz: ff257fb02f7feb16d680f111ae39c8806abaff034e087659586bbd43bc34a6e5
+  metadata.gz: 66a2d55d6c29842acb1b636afe7cf434c1769692a4b81ddee42fe77abf706f2a
+  data.tar.gz: 77cc311b6679b34314274cb136807747a2e029eb7ced3bdd5830c16732a15d28
 SHA512:
-  metadata.gz: 837c2c91f59f93374b747c23b71f5740b1e64d0e29e06d8e7e4076bb7c0ce24700b29b17434ece743d1e1fbcd7b6fa5b245526207525f2b6cbfb6c5a2be6dd29
-  data.tar.gz: ab6f75ce5add0b40f99d47a47b123ccfc090291b5e6adb4f1c41687e541f53f111dff9fecb6c6ecb46ff5aeaee8ad0af97bd444da8f9b00525b210fa8c1d7430
+  metadata.gz: c65b3f99e33a4ad4452b05cc3c0d6c9e545179d50bc5e6bf8ff8a718287a3dfa2b344c233c7fb3adf1c95835aa0370910740b0262c8266f26dd9b8920ea074bc
+  data.tar.gz: 347dbf45508a7ab22323e24ba46abc08c1c4f1d091b17823fe941d844726774b85c0fa9c82a8399c08ac763fbb6d2cf5bb3e5e9f554e21d6e2a3e0abf39dc7e4

data/lib/metanorma/standoc/base.rb

@@ -73,6 +73,8 @@ module Metanorma
         @sourcecode_markup_start = node.attr("sourcecode-markup-start") || "{{{"
         @sourcecode_markup_end = node.attr("sourcecode-markup-end") || "}}}"
         @datauriimage = node.attr("data-uri-image") != "false"
+        @blockunnumbered = (node.attr("block-unnumbered") || "").split(",")
+          .map(&:strip)
       end
 
       def init_reqt(node)

data/lib/metanorma/standoc/blocks_notes.rb

@@ -38,7 +38,7 @@ module Metanorma
 
       def todo_attrs(node)
         date = node.attr("date") || Date.today.iso8601.gsub(/\+.*$/, "")
-        date += "T00:00:00Z" unless /T/.match? date
+        date += "T00:00:00Z" unless date.include?("T")
         attr_code(id_attr(node)
           .merge(reviewer: node.attr("reviewer") || node.attr("source") ||
                  "(Unknown)",

data/lib/metanorma/standoc/cleanup.rb

@@ -7,6 +7,7 @@ require_relative "./cleanup_footnotes"
 require_relative "./cleanup_ref"
 require_relative "./cleanup_asciibib"
 require_relative "./cleanup_boilerplate"
+require_relative "./cleanup_bibdata"
 require_relative "./cleanup_section"
 require_relative "./cleanup_terms"
 require_relative "./cleanup_symbols"
@@ -26,6 +27,8 @@ module Metanorma
       def cleanup(xmldoc)
         element_name_cleanup(xmldoc)
         passthrough_cleanup(xmldoc)
+        unnumbered_blocks_cleanup(xmldoc)
+        metadata_cleanup(xmldoc) # feeds: boilerplate_cleanup
         sections_cleanup(xmldoc) # feeds: obligations_cleanup, toc_cleanup,
         # floatingtitle_cleanup
         obligations_cleanup(xmldoc)
@@ -70,7 +73,6 @@ module Metanorma
         svgmap_cleanup(xmldoc) # feeds: img_cleanup
         boilerplate_cleanup(xmldoc)
         toc_cleanup(xmldoc)
-        metadata_cleanup(xmldoc)
         smartquotes_cleanup(xmldoc)
         variant_cleanup(xmldoc)
         para_cleanup(xmldoc)

data/lib/metanorma/standoc/cleanup_bibdata.rb

@@ -0,0 +1,127 @@
+module Metanorma
+  module Standoc
+    module Cleanup
+      def bibdata_cleanup(xmldoc)
+        bibdata_anchor_cleanup(xmldoc)
+        bibdata_docidentifier_cleanup(xmldoc)
+        bibdata_embed_hdr_cleanup(xmldoc) # feeds bibdata_embed_id_cleanup
+        bibdata_embed_id_cleanup(xmldoc)
+        biblio_indirect_erefs(xmldoc, @internal_eref_namespaces&.uniq)
+      end
+
+      def bibdata_anchor_cleanup(xmldoc)
+        xmldoc.xpath("//bibdata//bibitem | //bibdata//note").each do |b|
+          b.delete("id")
+        end
+      end
+
+      def bibdata_docidentifier_cleanup(xmldoc)
+        ins = xmldoc.at("//bibdata/docidentifier")
+        xmldoc.xpath("//bibdata/docidentifier").each_with_index do |b, i|
+          i.zero? and next
+          ins.next = b.remove
+          ins = ins.next
+        end
+      end
+
+      def gather_indirect_erefs(xmldoc, prefix)
+        xmldoc.xpath("//eref[@type = '#{prefix}']")
+          .each_with_object({}) do |e, m|
+          e.delete("type")
+          m[e["bibitemid"]] = true
+        end.keys
+      end
+
+      def insert_indirect_biblio(xmldoc, refs, prefix)
+        i = xmldoc.at("bibliography") or
+          xmldoc.root << "<bibliography/>" and i = xmldoc.at("bibliography")
+        i = i.add_child("<references hidden='true' normative='false'/>").first
+        refs.each do |x|
+          i << <<~BIB
+            <bibitem id="#{x}" type="internal">
+            <docidentifier type="repository">#{x.sub(/^#{prefix}_/, "#{prefix}/")}</docidentifier>
+            </bibitem>
+          BIB
+        end
+      end
+
+      def indirect_eref_to_xref(eref, ident)
+        loc = eref.at("./localityStack[locality[@type = 'anchor']]")
+          &.remove&.text ||
+          eref.at("./locality[@type = 'anchor']")&.remove&.text || ident
+        eref.name = "xref"
+        eref.delete("bibitemid")
+        eref.delete("citeas")
+        eref["target"] = loc
+        unless eref.document.at("//*[@id = '#{loc}']")
+          eref.children = %(** Missing target #{loc})
+          eref["target"] = ident
+        end
+      end
+
+      def resolve_local_indirect_erefs(xmldoc, refs, prefix)
+        refs.each_with_object([]) do |r, m|
+          id = r.sub(/^#{prefix}_/, "")
+          n = xmldoc.at("//*[@id = '#{id}']")
+          if n&.at("./ancestor-or-self::*[@type = '#{prefix}']")
+            xmldoc.xpath("//eref[@bibitemid = '#{r}']").each do |e|
+              indirect_eref_to_xref(e, id)
+            end
+          else m << r
+          end
+        end
+      end
+
+      def biblio_indirect_erefs(xmldoc, prefixes)
+        prefixes&.each do |prefix|
+          refs = gather_indirect_erefs(xmldoc, prefix)
+          refs = resolve_local_indirect_erefs(xmldoc, refs, prefix)
+          refs.empty? and next
+          insert_indirect_biblio(xmldoc, refs, prefix)
+        end
+      end
+
+      def bibdata_embed_hdr_cleanup(xmldoc)
+        (@embed_hdr.nil? || @embed_hdr.empty?) and return
+        xmldoc.at("//bibdata") << "<relation type='derivedFrom'>" \
+                                  "#{hdr2bibitem(@embed_hdr.first)}</relation>"
+      end
+
+      def hdr2bibitem(hdr)
+        xml = Asciidoctor
+          .convert(hdr[:text], backend: hdr2bibitem_type(hdr),
+                               header_footer: true)
+        b = Nokogiri::XML(xml).at("//xmlns:bibdata")
+        b.name = "bibitem"
+        b.delete("type")
+        embed_recurse(b, hdr)
+        b.to_xml
+      end
+
+      def hdr2bibitem_type(hdr)
+        m = /:mn-document-class: (\S+)/.match(hdr[:text])
+        if m then m[1].to_sym
+        else Processor.new.asciidoctor_backend
+        end
+      end
+
+      def embed_recurse(bibitem, node)
+        node[:child].map { |x| hdr2bibitem(x) }.each do |x|
+          bibitem << "<relation type='derivedFrom'>#{x}</relation>"
+        end
+      end
+
+      def bibdata_embed_id_cleanup(xmldoc)
+        @embed_id.nil? and return
+        bibdata = xmldoc.at("//bibdata")
+        @embed_id.each do |d|
+          bibdata = bibdata.at("./relation[@type = 'derivedFrom']/bibitem")
+          ident = bibdata.at("./docidentifier[@primary = 'true']") ||
+            bibdata.at("./docidentifier")
+          xmldoc.xpath("//xref[@target = '#{d}'][normalize-space(text()) = '']")
+            .each { |x| x << ident.text }
+        end
+      end
+    end
+  end
+end

data/lib/metanorma/standoc/cleanup_block.rb

@@ -12,6 +12,7 @@ module Metanorma
          "ancestor::table or ancestor::bibdata)]"].each do |w|
           inject_id(xmldoc, w)
         end
+        xmldoc.xpath("//p[not(text()) and not(node())]").each(&:remove)
       end
 
       def inject_id(xmldoc, path)
@@ -67,11 +68,10 @@ module Metanorma
       # examples containing only figures become subfigures of figures
       def subfigure_cleanup(xmldoc)
         xmldoc.xpath("//example[figure]").each do |e|
-          next unless e.elements.reject do |m|
+          e.elements.reject do |m|
             %w(name figure index note).include?(m.name) ||
               (m.name == "dl" && m["key"] == "true")
-          end.empty?
-
+          end.empty? or next
           e.name = "figure"
         end
       end
@@ -100,8 +100,7 @@ module Metanorma
       # (so there was no way of making that block include the note)
       def note_cleanup(xmldoc)
         xmldoc.xpath("//note").each do |n|
-          next if n["keep-separate"] == "true" || !n.ancestors("table").empty?
-
+          n["keep-separate"] == "true" || !n.ancestors("table").empty? and next
           prev = n.previous_element || next
           n.parent = prev if ELEMS_ALLOW_NOTES.include? prev.name
         end
@@ -129,7 +128,7 @@ module Metanorma
 
       def merge_annotations_into_sourcecode(xmldoc)
         xmldoc.xpath("//sourcecode").each do |x|
-          while x&.next_element&.name == "annotation"
+          while x.next_element&.name == "annotation"
             x.next_element.parent = x
           end
         end
@@ -143,10 +142,8 @@ module Metanorma
       def sourcecode_cleanup(xmldoc)
         xmldoc.xpath("//sourcecode").each do |x|
           x.traverse do |n|
-            next unless n.text?
-            next unless /#{Regexp.escape(@sourcecode_markup_start)}/
-              .match?(n.text)
-
+            n.text? or next
+            /#{Regexp.escape(@sourcecode_markup_start)}/.match?(n.text) or next
             n.replace(sourcecode_markup(n))
           end
         end
@@ -164,8 +161,7 @@ module Metanorma
                           #{Regexp.escape(@sourcecode_markup_end)})/x)
           .each_slice(4).map.with_object([]) do |a, acc|
           acc << safe_noko(a[0], node.document)
-          next unless a.size == 4
-
+          a.size == 4 or next
           acc << Asciidoctor.convert(
             a[2], doctype: :inline, backend: (self&.backend&.to_sym || :standoc)
           )
@@ -174,7 +170,7 @@ module Metanorma
 
       def form_cleanup(xmldoc)
         xmldoc.xpath("//select").each do |s|
-          while s&.next_element&.name == "option"
+          while s.next_element&.name == "option"
             s << s.next_element
           end
         end
@@ -200,8 +196,7 @@ module Metanorma
       end
 
       def include_indexterm?(elem)
-        return false if elem.nil?
-
+        elem.nil? and return false
         !%w(image literal sourcecode).include?(elem.name)
       end
 
@@ -234,6 +229,14 @@ module Metanorma
           s.delete("type")
         end
       end
+
+      def unnumbered_blocks_cleanup(xmldoc)
+        @blockunnumbered&.each do |b|
+          xmldoc.xpath("//#{b}").each do |e|
+            e["unnumbered"] ||= "true"
+          end
+        end
+      end
     end
   end
 end

data/lib/metanorma/standoc/cleanup_boilerplate.rb

@@ -118,132 +118,58 @@ module Metanorma
       end
 
       def boilerplate(xml, conv)
+        # prevent infinite recursion of asciidoc boilerplate processing
+        xml.at("//metanorma-extension/semantic-metadata/" \
+               "headless[text() = 'true']") and return nil
         file = boilerplate_file(xml)
-        @boilerplateauthority and
-          file = File.join(@localdir, @boilerplateauthority)
+        @boilerplateauthority and file = File.join(@localdir,
+                                                   @boilerplateauthority)
         (!file.nil? and File.exist?(file)) or return
-        conv.populate_template(File.read(file, encoding: "UTF-8"), nil)
-      end
-
-      def bibdata_cleanup(xmldoc)
-        bibdata_anchor_cleanup(xmldoc)
-        bibdata_docidentifier_cleanup(xmldoc)
-        bibdata_embed_hdr_cleanup(xmldoc) # feeds bibdata_embed_id_cleanup
-        bibdata_embed_id_cleanup(xmldoc)
-        biblio_indirect_erefs(xmldoc, @internal_eref_namespaces&.uniq)
-      end
-
-      def bibdata_anchor_cleanup(xmldoc)
-        xmldoc.xpath("//bibdata//bibitem | //bibdata//note").each do |b|
-          b.delete("id")
-        end
-      end
-
-      def bibdata_docidentifier_cleanup(xmldoc)
-        ins = xmldoc.at("//bibdata/docidentifier")
-        xmldoc.xpath("//bibdata/docidentifier").each_with_index do |b, i|
-          i.zero? and next
-          ins.next = b.remove
-          ins = ins.next
-        end
-      end
-
-      def gather_indirect_erefs(xmldoc, prefix)
-        xmldoc.xpath("//eref[@type = '#{prefix}']")
-          .each_with_object({}) do |e, m|
-          e.delete("type")
-          m[e["bibitemid"]] = true
-        end.keys
-      end
-
-      def insert_indirect_biblio(xmldoc, refs, prefix)
-        i = xmldoc.at("bibliography") or
-          xmldoc.root << "<bibliography/>" and i = xmldoc.at("bibliography")
-        i = i.add_child("<references hidden='true' normative='false'/>").first
-        refs.each do |x|
-          i << <<~BIB
-            <bibitem id="#{x}" type="internal">
-            <docidentifier type="repository">#{x.sub(/^#{prefix}_/, "#{prefix}/")}</docidentifier>
-            </bibitem>
-          BIB
-        end
-      end
-
-      def indirect_eref_to_xref(eref, ident)
-        loc = eref.at("./localityStack[locality[@type = 'anchor']]")
-          &.remove&.text ||
-          eref.at("./locality[@type = 'anchor']")&.remove&.text || ident
-        eref.name = "xref"
-        eref.delete("bibitemid")
-        eref.delete("citeas")
-        eref["target"] = loc
-        unless eref.document.at("//*[@id = '#{loc}']")
-          eref.children = %(** Missing target #{loc})
-          eref["target"] = ident
-        end
-      end
-
-      def resolve_local_indirect_erefs(xmldoc, refs, prefix)
-        refs.each_with_object([]) do |r, m|
-          id = r.sub(/^#{prefix}_/, "")
-          n = xmldoc.at("//*[@id = '#{id}']")
-          if n&.at("./ancestor-or-self::*[@type = '#{prefix}']")
-            xmldoc.xpath("//eref[@bibitemid = '#{r}']").each do |e|
-              indirect_eref_to_xref(e, id)
-            end
-          else m << r
-          end
-        end
-      end
-
-      def biblio_indirect_erefs(xmldoc, prefixes)
-        prefixes&.each do |prefix|
-          refs = gather_indirect_erefs(xmldoc, prefix)
-          refs = resolve_local_indirect_erefs(xmldoc, refs, prefix)
-          refs.empty? and next
-          insert_indirect_biblio(xmldoc, refs, prefix)
-        end
-      end
-
-      def bibdata_embed_hdr_cleanup(xmldoc)
-        (@embed_hdr.nil? || @embed_hdr.empty?) and return
-        xmldoc.at("//bibdata") << "<relation type='derivedFrom'>" \
-                                  "#{hdr2bibitem(@embed_hdr.first)}</relation>"
-      end
-
-      def hdr2bibitem(hdr)
-        xml = Asciidoctor
-          .convert(hdr[:text], backend: hdr2bibitem_type(hdr),
-                               header_footer: true)
-        b = Nokogiri::XML(xml).at("//xmlns:bibdata")
-        b.name = "bibitem"
-        b.delete("type")
-        embed_recurse(b, hdr)
-        b.to_xml
-      end
-
-      def hdr2bibitem_type(hdr)
-        m = /:mn-document-class: (\S+)/.match(hdr[:text])
-        if m then m[1].to_sym
-        else Processor.new.asciidoctor_backend
-        end
-      end
-
-      def embed_recurse(bibitem, node)
-        node[:child].map { |x| hdr2bibitem(x) }.each do |x|
-          bibitem << "<relation type='derivedFrom'>#{x}</relation>"
+        b = conv.populate_template(boilerplate_read(file), nil)
+        boilerplate_file_convert(b)
+      end
+
+      def boilerplate_read(file)
+        ret = File.read(file, encoding: "UTF-8")
+        /\.adoc$/.match?(file) and
+          ret.gsub!(/(?<!\{)(\{\{[^{}]+\}\})(?!\})/, "pass:[\\1]")
+        ret
+      end
+
+      # If Asciidoctor, convert top clauses to tags and wrap in <boilerplate>
+      def boilerplate_file_convert(file)
+        Nokogiri::XML(file).root and return file
+        to_xml(boilerplate_file_restructure(file))
+      end
+
+      # If Asciidoctor, convert top clauses to tags and wrap in <boilerplate>
+      def boilerplate_file_restructure(file)
+        ret = adoc2xml(file, backend.to_sym)
+        boilerplate_xml_cleanup(ret)
+        ret.name = "boilerplate"
+        boilerplate_top_elements(ret)
+        ret
+      end
+
+      # remove Metanorma namespace, so generated doc containing boilerplate 
+      # can be queried consistently
+      # _\d+ anchor is assigned to titleless clauses, will clash with main doc
+      # instances of same
+      def boilerplate_xml_cleanup(xml)
+        ns = xml.namespace.href
+        xml.traverse do |n|
+          n.element? or next
+          n.namespace.href == ns and n.namespace = nil
+          /^_\d+$/.match?(n["id"]) and
+            n["id"] = "_#{UUIDTools::UUID.random_create}"
         end
       end
 
-      def bibdata_embed_id_cleanup(xmldoc)
-        @embed_id.nil? and return
-        bibdata = xmldoc.at("//bibdata")
-        @embed_id.each do |d|
-          bibdata = bibdata.at("./relation[@type = 'derivedFrom']/bibitem")
-          ident = bibdata.at("./docidentifier[@primary = 'true']") ||
-            bibdata.at("./docidentifier")
-          xmldoc.xpath("//xref[@target = '#{d}'][normalize-space(text()) = '']")
-            .each { |x| x << ident.text }
+      def boilerplate_top_elements(xml)
+        xml.elements.each do |e|
+          (t = e.at("./title") and /-statement$/.match?(t.text)) or next
+          e.name = t.remove.text
+          e.keys.each { |a| e.delete(a) } # rubocop:disable Style/HashEachMethods
         end
       end
     end

data/lib/metanorma/standoc/processor.rb

@@ -5,7 +5,7 @@ module Metanorma
     class Processor < Metanorma::Processor
       class << self
         attr_reader :asciidoctor_backend
-            end
+      end
 
       def initialize # rubocop:disable Lint/MissingSuper
         @short = :standoc
@@ -22,7 +22,7 @@ module Metanorma
       end
 
       def version
-        "Metanorma::Standoc #{Metanorma::Standoc::VERSION}/"\
+        "Metanorma::Standoc #{Metanorma::Standoc::VERSION}/" \
           "IsoDoc #{IsoDoc::VERSION}"
       end
 

data/lib/metanorma/standoc/table.rb

@@ -35,7 +35,7 @@ module Metanorma
 
         xml_table.colgroup do |cg|
           node.columns.each do |col|
-            cg.col **{ width: "#{col.attr 'colpcwidth'}%" }
+            cg.col width: "#{col.attr 'colpcwidth'}%"
           end
         end
       end

data/lib/metanorma/standoc/utils.rb

@@ -76,6 +76,17 @@ module Metanorma
           .gsub(/'/, "'")
       end
 
+      # wrapped in <sections>
+      def adoc2xml(text, flavour)
+        Nokogiri::XML(text).root and return text
+        c = Asciidoctor.convert("= X\nA\n:semantic-metadata-headless: true\n" \
+                                ":novalid:\n\n#{text}\n",
+                                backend: flavour, header_footer: true)
+        Nokogiri::XML(c).at("//xmlns:sections")
+      end
+
+      module_function :adoc2xml
+
       class EmptyAttr
         def attr(_any_attribute)
           nil

data/lib/metanorma/standoc/version.rb

@@ -19,6 +19,6 @@ module Metanorma
   end
 
   module Standoc
-    VERSION = "2.4.5".freeze
+    VERSION = "2.4.7".freeze
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: metanorma-standoc
 version: !ruby/object:Gem::Version
-  version: 2.4.5
+  version: 2.4.7
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2023-06-05 00:00:00.000000000 Z
+date: 2023-06-20 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: asciidoctor
@@ -475,6 +475,7 @@ files:
 - lib/metanorma/standoc/cleanup.rb
 - lib/metanorma/standoc/cleanup_amend.rb
 - lib/metanorma/standoc/cleanup_asciibib.rb
+- lib/metanorma/standoc/cleanup_bibdata.rb
 - lib/metanorma/standoc/cleanup_block.rb
 - lib/metanorma/standoc/cleanup_boilerplate.rb
 - lib/metanorma/standoc/cleanup_footnotes.rb