isodoc 2.10.0 → 3.0.0

data/lib/isodoc/presentation_function/terms.rb

@@ -1,103 +1,25 @@
 module IsoDoc
   class PresentationXMLConvert < ::IsoDoc::Convert
-    def designation(docxml)
-      docxml.xpath(ns("//term")).each { |t| merge_second_preferred(t) }
-      docxml.xpath(ns("//preferred | //admitted | //deprecates"))
-        .each { |p| designation1(p) }
-      docxml.xpath(ns("//deprecates")).each { |d| deprecates(d) }
-      docxml.xpath(ns("//admitted")).each { |d| admits(d) }
-    end
-
-    def deprecates(elem)
-      elem.children.first.previous = @i18n.l10n("#{@i18n.deprecated}: ")
-    end
-
-    def admits(elem); end
-
-    def merge_second_preferred(term)
-      pref = nil
-      term.xpath(ns("./preferred[expression/name]")).each_with_index do |p, i|
-        (i.zero? and pref = p) or merge_second_preferred1(pref, p)
-      end
-    end
-
-    def merge_second_preferred1(pref, second)
-      merge_preferred_eligible?(pref, second) or return
-      n1 = pref.at(ns("./expression/name"))
-      n2 = second.remove.at(ns("./expression/name"))
-      n1.children = l10n("#{to_xml(n1.children)}; #{Common::to_xml(n2.children)}")
-    end
-
-    def merge_preferred_eligible?(first, second)
-      firstex = first.at(ns("./expression")) || {}
-      secondex = second.at(ns("./expression")) || {}
-      first["geographic-area"] == second["geographic-area"] &&
-        firstex["language"] == secondex["language"] &&
-        !first.at(ns("./pronunciation | ./grammar")) &&
-        !second.at(ns("./pronunciation | ./grammar"))
-    end
-
-    def designation1(desgn)
-      s = desgn.at(ns("./termsource"))
-      name = desgn.at(ns("./expression/name | ./letter-symbol/name | " \
-                         "./graphical-symbol")) or return
-      designation_annotate(desgn, name)
-      s and desgn.next = s
-    end
-
-    def designation_annotate(desgn, name)
-      designation_boldface(desgn)
-      designation_field(desgn, name)
-      g = desgn.at(ns("./expression/grammar")) and
-        name << ", #{designation_grammar(g).join(', ')}"
-      designation_localization(desgn, name)
-      designation_pronunciation(desgn, name)
-      designation_bookmarks(desgn, name)
-      desgn.children = name.children
-    end
-
-    def designation_boldface(desgn)
-      desgn.name == "preferred" or return
-      name = desgn.at(ns("./expression/name | ./letter-symbol/name")) or return
-      name.children = "<strong>#{name.children}</strong>"
-    end
-
-    def designation_field(desgn, name)
-      f = desgn.xpath(ns("./field-of-application | ./usage-info"))
-        &.map { |u| to_xml(u.children) }&.join(", ")
-      f&.empty? and return nil
-      name << ", &#x3c;#{f}&#x3e;"
-    end
-
-    def designation_grammar(grammar)
-      ret = []
-      grammar.xpath(ns("./gender | ./number")).each do |x|
-        ret << @i18n.grammar_abbrevs[x.text]
+    def termcontainers(docxml)
+      docxml.xpath(ns("//term")).each do |t|
+        %w(preferred admitted deprecates related definition termsource)
+          .each do |w|
+          d = t.at(ns("./#{w}[last()]")) and d.after("<fmt-#{w}/>")
+        end
       end
-      %w(isPreposition isParticiple isAdjective isVerb isAdverb isNoun)
-        .each do |x|
-        grammar.at(ns("./#{x}[text() = 'true']")) and
-          ret << @i18n.grammar_abbrevs[x]
+      docxml.xpath(ns("//termsource")).each do |s|
+        s["id"] ||= "_#{UUIDTools::UUID.random_create}"
       end
-      ret
-    end
-
-    def designation_localization(desgn, name)
-      loc = [desgn&.at(ns("./expression/@language"))&.text,
-             desgn&.at(ns("./expression/@script"))&.text,
-             desgn&.at(ns("./@geographic-area"))&.text].compact
-      loc.empty? and return
-      name << ", #{loc.join(' ')}"
     end
 
-    def designation_pronunciation(desgn, name)
-      f = desgn.at(ns("./expression/pronunciation")) or return
-      name << ", /#{to_xml(f.children)}/"
-    end
-
-    def designation_bookmarks(desgn, name)
-      desgn.xpath(ns(".//bookmark")).each do |b|
-        name << b.remove
+    def termcleanup(docxml)
+      docxml.xpath(ns("//term")).each do |t|
+        %w(preferred admitted deprecates related definition termsource)
+          .each do |w|
+          t.xpath(ns("./#{w}//fmt-name | ./#{w}//fmt-xref-label")).each(&:remove)
+          f = t.at(ns(".//fmt-#{w}"))
+          f&.children&.empty? and f.remove
+        end
       end
     end
 
@@ -109,34 +31,46 @@ module IsoDoc
       docxml.xpath(ns("//termnote")).each { |f| termnote1(f) }
     end
 
+    def termnote_delim(_elem)
+      l10n(": ")
+    end
+
     def termnote1(elem)
-      lbl = l10n(@xrefs.anchor(elem["id"], :label) || "???")
-      prefix_name(elem, "", lower2cap(lbl), "name")
+      lbl = termnote_label(elem)
+      prefix_name(elem, { label: termnote_delim(elem) }, lower2cap(lbl), "name")
+    end
+
+    def termnote_label(elem)
+      @xrefs.anchor(elem["id"], :label) || "???"
     end
 
     def termdefinition(docxml)
-      docxml.xpath(ns("//term[definition]")).each do |f|
-        termdefinition1(f)
-      end
+      docxml.xpath(ns("//term[definition]")).each { |f| termdefinition1(f) }
     end
 
     def termdefinition1(elem)
-      unwrap_definition(elem)
-      multidef(elem) if elem.xpath(ns("./definition")).size > 1
+      d = elem.xpath(ns("./definition"))
+      d1 = elem.at(ns("./fmt-definition"))
+      if d.size > 1 then multidef(elem, d, d1)
+      else singledef(elem, d, d1)
+      end
+      unwrap_definition(elem, d1)
+      termdomain(elem, d1)
     end
 
-    def multidef(elem)
-      d = elem.at(ns("./definition"))
-      d = d.replace("<ol><li>#{to_xml(d.children)}</li></ol>").first
-      elem.xpath(ns("./definition")).each do |f|
-        f = f.replace("<li>#{to_xml(f.children)}</li>").first
-        d << f
+    def multidef(_elem, defn, fmt_defn)
+      ret = defn.each_with_object([]) do |f, m|
+        m << "<li>#{to_xml(semx_fmt_dup(f))}</li>"
       end
-      d.wrap("<definition></definition>")
+      fmt_defn << "<ol>#{ret.join("\n")}</ol>"
     end
 
-    def unwrap_definition(elem)
-      elem.xpath(ns("./definition")).each do |d|
+    def singledef(_elem, defn, fmt_defn)
+      fmt_defn << semx_fmt_dup(defn.first)
+    end
+
+    def unwrap_definition(_elem, fmt_defn)
+      fmt_defn.xpath(ns(".//semx[@element = 'definition']")).each do |d|
         %w(verbal-definition non-verbal-representation).each do |e|
           v = d&.at(ns("./#{e}"))
           v&.replace(v.children)
@@ -144,19 +78,62 @@ module IsoDoc
       end
     end
 
+    def termdomain(elem, fmt_defn)
+      d = elem.at(ns(".//domain")) or return
+      p = fmt_defn.at(ns(".//p")) or return
+      d1 = semx_fmt_dup(d)
+      p.add_first_child "&lt;#{to_xml(d1)}&gt;  "
+    end
+
     def termsource(docxml)
-      docxml.xpath(ns("//termsource")).each { |f| termsource_modification(f) }
-      docxml.xpath(ns("//termsource")).each { |f| termsource1(f) }
+      copy_baselevel_termsource(docxml)
+      # TODO should I wrap fmt-definition//termsource in fmt-termsource, in order to preserve termsource attributes?
+      docxml.xpath(ns("//fmt-termsource/termsource | //fmt-definition//termsource | //fmt-preferred//termsource | //fmt-admitted//termsource | //fmt-deprecates//termsource"))
+        .each do |f|
+        termsource_modification(f)
+      end
+      docxml.xpath(ns("//fmt-preferred//fmt-termsource | //fmt-admitted//fmt-termsource | //fmt-deprecates//fmt-termsource"))
+        .each do |f|
+          termsource_designation(f)
+        end
+      docxml.xpath(ns("//fmt-termsource/termsource | //fmt-definition//termsource | //fmt-preferred//termsource | //fmt-admitted//termsource | //fmt-deprecates//termsource"))
+        .each do |f|
+        f.parent and termsource1(f)
+      end
+    end
+
+    def termsource_designation(fmtsource)
+      p = fmtsource.previous_element
+          p&.name == "p" or return
+          p << " "
+          p << fmtsource.children
+    end
+
+    def copy_baselevel_termsource(docxml)
+      docxml.xpath(ns("//term[termsource]")).each do |x|
+        s = x.xpath(ns("./termsource"))
+        s1 = x.at(ns("./fmt-termsource"))
+        s.each { |ss| s1 << ss.clone }
+        strip_duplicate_ids(nil, s, s1)
+        %w(status type).each { |a| s[0][a] and s1[a] = s[0][a] }
+      end
     end
 
     def termsource1(elem)
+      ret = [semx_fmt_dup(elem)]
       while elem&.next_element&.name == "termsource"
-        elem << "; #{to_xml(elem.next_element.remove.children)}"
+        ret << semx_fmt_dup(elem.next_element.remove)
       end
-      elem.children = l10n("[#{@i18n.source}: #{to_xml(elem.children).strip}]")
+      s = ret.map { |x| to_xml(x) }.map(&:strip).join("; ")
+      termsource_label(elem, s)
+    end
+
+    def termsource_label(elem, sources)
+      elem.replace(l10n("[#{@i18n.source}: #{sources}]"))
     end
 
     def termsource_modification(elem)
+      elem.xpath(".//text()[normalize-space() = '']").each(&:remove)
       origin = elem.at(ns("./origin"))
       s = termsource_status(elem["status"]) and origin.next = l10n(", #{s}")
       termsource_add_modification_text(elem.at(ns("./modification")))
@@ -164,10 +141,13 @@ module IsoDoc
 
     def termsource_add_modification_text(mod)
       mod or return
-      mod.text.strip.empty? or mod.previous = " &#x2014; "
-      mod.elements.size == 1 and
-        mod.elements[0].replace(mod.elements[0].children)
-      mod.replace(mod.children)
+      if mod.text.strip.empty?
+        mod.remove
+        return
+      end
+      mod.previous = " &#x2014; "
+      mod.elements.size == 1 and mod.children = to_xml(mod.elements[0].children)
+      mod.replace(semx_fmt_dup(mod))
     end
 
     def termsource_status(status)
@@ -176,5 +156,14 @@ module IsoDoc
       when "adapted" then @i18n.adapted
       end
     end
+
+    def term(docxml)
+      docxml.xpath(ns("//term")).each { |f| term1(f) }
+    end
+
+    def term1(elem)
+      lbl = @xrefs.anchor(elem["id"], :label) or return
+      prefix_name(elem, {}, "#{lbl}#{clausedelim}", "name")
+    end
   end
 end

data/lib/isodoc/presentation_function/title.rb

@@ -0,0 +1,107 @@
+require_relative "refs"
+
+module IsoDoc
+  class PresentationXMLConvert < ::IsoDoc::Convert
+    def middle_title(docxml)
+      s = docxml.at(ns("//sections")) or return
+      t = @meta.get[:doctitle]
+      t.nil? || t.empty? and return
+      s.add_first_child "<p class='zzSTDTitle1'>#{t}</p>"
+    end
+
+    def missing_title(docxml)
+      docxml.xpath(ns("//definitions[not(./title)]")).each do |d|
+        # should only be happening for subclauses
+        d.add_first_child "<title>#{@i18n.symbols}</title>"
+      end
+      docxml.xpath(ns("//foreword[not(./title)]")).each do |d|
+        d.add_first_child "<title>#{@i18n.foreword}</title>"
+      end
+    end
+
+    def floattitle(docxml)
+      docxml.xpath(ns(".//floating-title")).each { |f| floattitle1(f) }
+    end
+
+    # TODO not currently doing anything with the @depth attribute of floating-title
+    def floattitle1(elem)
+      elem["id"] ||= "_#{UUIDTools::UUID.random_create}"
+      p = elem.dup
+      p.children = "<semx element='floating-title' source='#{elem['id']}'>" \
+        "#{to_xml(p.children)}</semx>"
+      elem.next = p
+      p.name = "p"
+      p["type"] = "floating-title"
+      transfer_id(elem, p)
+    end
+
+    def preceding_floating_titles(node, idx)
+      out = node.xpath("./preceding-sibling::*")
+        .reverse.each_with_object([]) do |p, m|
+        %w(note admonition p floating-title).include?(p.name) or break m
+        m << p
+      end
+      #require 'debug'; out.empty? or binding.b
+      out.reject { |c| c["displayorder"] }.reverse_each do |c|
+          skip_display_order?(c) and next
+        c["displayorder"] = idx
+        idx += 1
+      end
+      idx
+    end
+
+    def clausetitle(docxml)
+      cjk_extended_title(docxml)
+    end
+
+    def cjk_search
+      lang = %w(zh ja ko).map { |x| "@language = '#{x}'" }.join(" or ")
+      %(Hans Hant Jpan Hang Kore).include?(@script) and
+        lang += " or not(@language)"
+      lang
+    end
+
+    def cjk_extended_title(doc)
+      l = cjk_search
+      doc.xpath(ns("//bibdata/title[#{l}] | //floating-title[#{l}] | " \
+                   "//fmt-title[@depth = '1' or not(@depth)][#{l}]"))
+        .each do |t|
+        t.text.size < 4 or next
+        t.traverse do |n|
+          n.text? or next
+          n.replace(@i18n.cjk_extend(n.text))
+        end
+      end
+    end
+
+    def preceding_floats(clause)
+      ret = []
+      p = clause
+      while prev = p.previous_element
+        if prev.name == "floating-title"
+          ret << prev
+          p = prev
+        else break end
+      end
+      ret
+    end
+
+    def toc_title(docxml)
+      docxml.at(ns("//preface/clause[@type = 'toc']")) and return
+      ins = toc_title_insert_pt(docxml) or return
+      id = UUIDTools::UUID.random_create.to_s
+      ins.previous = <<~CLAUSE
+        <clause type = 'toc' id='_#{id}'><fmt-title depth='1'>#{@i18n.table_of_contents}</fmt-title></clause>
+      CLAUSE
+    end
+
+    def toc_title_insert_pt(docxml)
+      ins = docxml.at(ns("//preface")) ||
+        docxml.at(ns("//sections | //annex | //bibliography"))
+          &.before("<preface> </preface>")
+          &.previous_element or return nil
+      ins.children.empty? and ins << " "
+      ins.children.first
+    end
+  end
+end

data/lib/isodoc/presentation_function/xrefs.rb

@@ -1,5 +1,18 @@
 module IsoDoc
   class PresentationXMLConvert < ::IsoDoc::Convert
+    def fmt_ref(docxml)
+      docxml.xpath(ns("//xref | //eref | //origin | //link")).each do |x|
+        sem_xml_descendant?(x) and next
+        tag = x.name
+        y = Nokogiri::XML::Node.new("fmt-#{tag}", x.document)
+        x.attributes.each_key { |a| y[a] = x[a] }
+        n = semx_fmt_dup(x) # semx/fmt-xref for ease of processing
+        n.children.each { |c| y << c }
+        n << y
+        x.next = n
+      end
+    end
+
     def prefix_container(container, linkend, node, target)
       prefix_container?(container, node) or return linkend
       container_container = @xrefs.anchor(container, :container, false)
@@ -7,20 +20,21 @@ module IsoDoc
         prefix_container(container_container,
                          anchor_xref(node, container, container: true),
                          node, target)
-      l10n(@i18n.nested_xref.sub("%1", container_label)
-        .sub("%2", linkend))
+      l10n(connectives_spans(@i18n.nested_xref
+        .sub("%1", "<span class='fmt-xref-container'>#{container_label}</span>")
+        .sub("%2", linkend)))
     end
 
     def anchor_value(id)
-      @xrefs.anchor(id, :bare_xref) || @xrefs.anchor(id, :value) ||
-        @xrefs.anchor(id, :label) || @xrefs.anchor(id, :xref)
+      @xrefs.anchor(id, :bare_xref) || @xrefs.anchor(id, :label) ||
+        @xrefs.anchor(id, :value) || @xrefs.anchor(id, :xref)
     end
 
     def anchor_linkend(node, linkend)
       node["style"] == "id" and
         return anchor_id_postproc(node)
       node["citeas"].nil? && node["bibitemid"] and
-        return @xrefs.anchor(node["bibitemid"], :xref) || "???"
+        return citeas_cleanup(@xrefs.anchor(node["bibitemid"], :xref)) || "???"
       node.at(ns("./location")) and
         return combine_xref_locations(node) || "???"
       node["target"] && node["droploc"] and
@@ -55,14 +69,15 @@ module IsoDoc
 
     def anchor_xref_short(node, target, container)
       if (l = node["label"]) && !container
-        @i18n.l10n("#{l} #{anchor_value(target)}")
+        v = anchor_value(target)
+        @i18n.l10n(%[<span class="fmt-element-name">#{l}</span> #{v}])
       else @xrefs.anchor(target, :xref)
       end
     end
 
     def anchor_xref_full(num, title)
       (!title.nil? && !title.empty?) or return nil
-      l10n("#{num}, #{title}")
+      l10n("#{num}<span class='fmt-comma'>,</span> #{title}")
     end
 
     def prefix_container?(container, node)
@@ -116,7 +131,7 @@ module IsoDoc
       node.xpath(ns("./location")).each_with_object([]) do |l, m|
         type = @xrefs.anchor(l["target"], :type)
         m << { conn: l["connective"], target: l["target"],
-               type: type, node: l, elem: @xrefs.anchor(l["target"], :elem),
+               type:, node: l, elem: @xrefs.anchor(l["target"], :elem),
                container: @xrefs.anchor(l["target"], :container, false) ||
                  %w(termnote).include?(type) }
       end
@@ -124,7 +139,7 @@ module IsoDoc
 
     def loc2xref(entry)
       if entry[:target]
-        "<xref nested='true' target='#{entry[:target]}'>#{entry[:label]}</xref>"
+        "<fmt-xref nested='true' target='#{entry[:target]}'>#{entry[:label]}</fmt-xref>"
       else
         entry[:label]
       end
@@ -132,20 +147,25 @@ module IsoDoc
 
     def combine_conn(list)
       list.size == 1 and list.first[:label]
-      if list[1..-1].all? { |l| l[:conn] == "and" }
-        @i18n.boolean_conj(list.map { |l| loc2xref(l) }, "and")
-      elsif list[1..-1].all? { |l| l[:conn] == "or" }
-        @i18n.boolean_conj(list.map { |l| loc2xref(l) }, "or")
+      if list[1..].all? { |l| l[:conn] == "and" }
+        connectives_spans(@i18n.boolean_conj(list.map do |l|
+          loc2xref(l)
+        end, "and"))
+      elsif list[1..].all? { |l| l[:conn] == "or" }
+        connectives_spans(@i18n.boolean_conj(list.map do |l|
+          loc2xref(l)
+        end, "or"))
       else
         ret = loc2xref(list[0])
-        list[1..-1].each { |l| ret = i18n_chain_boolean(ret, l) }
+        list[1..].each { |l| ret = i18n_chain_boolean(ret, l) }
         ret
       end
     end
 
     def i18n_chain_boolean(value, entry)
-      @i18n.send("chain_#{entry[:conn]}").sub("%1", value)
-        .sub("%2", loc2xref(entry))
+      connectives_spans(@i18n.send("chain_#{entry[:conn]}")
+        .sub("%1", value)
+        .sub("%2", loc2xref(entry)))
     end
 
     def can_conflate_xref_rendering?(locs)
@@ -156,7 +176,7 @@ module IsoDoc
     end
 
     def capitalise_xref(node, linkend, label)
-      linktext = linkend.gsub(/<[^>]+>/, "")
+      linktext = linkend.gsub(/<[^<>]+>/, "")
       (label && !label.empty? && /^#{Regexp.escape(label)}/.match?(linktext)) ||
         linktext[0, 1].match?(/\p{Upper}/) and return linkend
       node["case"] and

data/lib/isodoc/presentation_xml_convert.rb

@@ -7,6 +7,7 @@ require_relative "presentation_function/erefs"
 require_relative "presentation_function/inline"
 require_relative "presentation_function/math"
 require_relative "presentation_function/section"
+require_relative "presentation_function/index"
 require_relative "presentation_function/bibdata"
 require_relative "presentation_function/metadata"
 
@@ -18,10 +19,13 @@ module IsoDoc
       super
     end
 
-    def convert1(docxml, _filename, _dir)
+    def convert1(docxml, filename, dir)
+      @outputdir = dir
+      @outputfile = Pathname.new(filename).basename.to_s
       docid_prefixes(docxml) # feeds @xrefs.parse citation processing
       @xrefs.parse docxml
-      info docxml, nil
+      @xrefs.klass.meta = @meta
+      @xrefs.klass.info docxml, nil
       conversions(docxml)
       docxml.root["type"] = "presentation"
       docxml.to_xml.gsub("<", "<").gsub(">", ">")
@@ -35,7 +39,7 @@ module IsoDoc
     end
 
     def conversions(docxml)
-      semantic_xml_insert(docxml)
+      #semantic_xml_insert(docxml)
       metadata docxml
       bibdata docxml
       @xrefs.parse docxml
@@ -52,20 +56,21 @@ module IsoDoc
       # feeds middle_title
       # triggers xrefs reparse, so put references before all other sections,
       # which alter titles and thus can alter xrefs
-      rearrange_clauses docxml # feeds toc, display_order, clausetitle, clause, middle_title
+      rearrange_clauses docxml # feeds toc, display_order, clausetitle,
+      # clause, middle_title
       middle_title docxml
+      missing_title docxml # feeds clause
       annex docxml
       clause docxml # feeds clausetitle
       term docxml
-      index docxml
       clausetitle docxml # feeds floattitle
       floattitle docxml # feeds rearrange_clauses
+      index docxml # fed by strip_duplicate_ids
       toc docxml
       display_order docxml
     end
 
     def block(docxml)
-      amend docxml
       table docxml
       figure docxml
       sourcecode docxml
@@ -74,20 +79,26 @@ module IsoDoc
       note docxml
       admonition docxml
       source docxml
+      ul docxml
       ol docxml
+      dl docxml
+      quote docxml
       permission docxml
       requirement docxml
       recommendation docxml
       requirement_render docxml
+      amend docxml
     end
 
     def inline(docxml)
       bibitem_lookup(docxml) # feeds citeas
-      citeas docxml # feeds xref, eref, origin, quotesource
+      fmt_ref docxml # feeds citeas, xref, eref, origin, concept
+      citeas docxml # feeds xref, eref, origin, concept
       xref docxml
       eref docxml # feeds eref2link
       origin docxml # feeds eref2link
-      quotesource docxml # feeds eref2link
+      #quotesource docxml # feeds eref2link
+      concept docxml
       eref2link docxml
       mathml docxml
       ruby docxml
@@ -99,15 +110,18 @@ module IsoDoc
     end
 
     def terms(docxml)
+      termcontainers docxml
       termexample docxml
       termnote docxml
       termdefinition docxml
       designation docxml
       termsource docxml
-      concept docxml
+      #concept docxml
       related docxml
+      termcleanup docxml
     end
 
+    # KILL
     def semantic_xml_insert(xml)
       @semantic_xml_insert or return
       embed = embedable_semantic_xml(xml)
@@ -120,7 +134,7 @@ module IsoDoc
     def metanorma_extension_insert_pt(xml)
       xml.at(ns("//metanorma-extension")) ||
         xml.at(ns("//bibdata"))&.after("<metanorma-extension/>")
-        &.next_element ||
+          &.next_element ||
         xml.root.elements.first.before("<metanorma-extension/>")
           .previous_element
     end

data/lib/isodoc/version.rb

@@ -1,3 +1,3 @@
 module IsoDoc
-  VERSION = "2.10.0".freeze
+  VERSION = "3.0.0".freeze
 end