metanorma-standoc 2.2.5 → 2.2.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 12bb8cacaa61b510652a691e358e62e0e2e4f4ea1ac37dcff09f394b097051a9
-  data.tar.gz: 714fa0c071097b0817e3b95c5066fbd181634f1b7f09cff04e2c1e5eb45b385a
+  metadata.gz: c59c6e517b305ec3559d4948767575ba032d877714800712e883bc2a0a561e6b
+  data.tar.gz: eeb0aaf07ac2ea303921b2ed855981b834f9fb8de4e06b708611852fb7465123
 SHA512:
-  metadata.gz: 4a45f4013300b1c263e7f99ff50605e3576157afaacb4710c7a1f9b14c2e1ac6a624737be9a87e6a4ca3205756451b9a7cacde8df4bc8491d508121d5e821772
-  data.tar.gz: 0c5bdb3f7a121998082b101488352bba31cad34abedf1f7fb1845cf25438c671545e3c65d09ec6a5bf06a075c8f311edbe4f61d210f470b1da009c71c8ea7627
+  metadata.gz: 7a91ae88a07c545c1c03f0845d94f38db34cc1128d7c7274eb49cbd2dd509ea8ca02da007e1de78b0d91d773c7eec6798206613eacda9e6322ebc3b0ab67f37b
+  data.tar.gz: 58af55addb78a590e2389dbce06cc29977bda866924c80e042047b85877f9374407dfb5f2b1679b8766bfd12ed0ce9e4fa7a339b88d333099666e5b169afcf09

data/lib/isodoc/html/htmlstyle.css

@@ -368,7 +368,7 @@ div.figure {
     font-weight: 700;
     font-size: 1em;
     text-align: center; }
-  div.figure > img {
+  div.figure > img, div.figure > svg {
     margin-left: auto;
     margin-right: auto;
     display: block;

data/lib/metanorma/standoc/base.rb

@@ -41,6 +41,7 @@ module Metanorma
       def init_vars
         @fn_number ||= 0
         @refids = Set.new
+        @anchor_alias = {}
         @anchors = {}
         @internal_eref_namespaces = []
         @seen_headers = []
@@ -56,6 +57,7 @@ module Metanorma
         @boilerplateauthority = node.attr("boilerplate-authority")
         @embed_hdr = node.attr("embed_hdr")
         @document_scheme = node.attr("document-scheme")
+        @xrefstyle = node.attr("xrefstyle")
       end
 
       def init_processing(node)
@@ -129,9 +131,7 @@ module Metanorma
 
       def document1(node)
         init(node)
-        ret = makexml(node)
-          .to_xml(encoding: "UTF-8", indent: 2,
-                  save_with: Nokogiri::XML::Node::SaveOptions::AS_XML)
+        ret = to_xml(makexml(node))
         outputs(node, ret) unless node.attr("nodoc") || !node.attr("docfile")
         ret
       end
@@ -162,8 +162,7 @@ module Metanorma
 
       def clean_abort(msg, file = nil)
         if file
-          doc = file.to_xml(encoding: "UTF-8", indent: 2,
-                            save_with: Nokogiri::XML::Node::SaveOptions::AS_XML)
+          doc = to_xml(file)
           File.open("#{@filename}.xml.abort", "w:UTF-8") { |f| f.write(doc) }
         end
         clean_exit

data/lib/metanorma/standoc/basicdoc.rng

@@ -1,10 +1,5 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <grammar xmlns="http://relaxng.org/ns/structure/1.0" datatypeLibrary="http://www.w3.org/2001/XMLSchema-datatypes">
-  <include href="biblio.rng">
-    <start>
-      <ref name="document"/>
-    </start>
-  </include>
   <define name="document">
     <element name="document">
       <optional>
@@ -1125,26 +1120,4 @@
       </zeroOrMore>
     </element>
   </define>
-  <define name="ext">
-    <element name="ext">
-      <ref name="BibDataExtensionType"/>
-    </element>
-  </define>
-  <define name="BibDataExtensionType">
-    <ref name="doctype"/>
-  </define>
-  <define name="doctype">
-    <element name="doctype">
-      <ref name="DocumentType"/>
-    </element>
-  </define>
-  <define name="DocumentType">
-    <value>document</value>
-  </define>
-  <define name="BibData">
-    <ref name="BibliographicItem"/>
-    <optional>
-      <ref name="ext"/>
-    </optional>
-  </define>
 </grammar>

data/lib/metanorma/standoc/biblio-standoc.rng

@@ -0,0 +1,164 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<grammar xmlns="http://relaxng.org/ns/structure/1.0">
+  <!--
+    Add-ons to biblio.rnc for standoc model: defines the extension point BibDataExtensionType
+    of relaton
+    
+    Specialisations as for biblio.rnc. Extension point can be redefined completely for a flavour of standoc
+    (SDO); but other elements in Bibdata can only be extended (more specialised vocabularies for Bibdata)
+  -->
+  <include href="biblio.rng">
+    <define name="BibData">
+      <ref name="BibliographicItem"/>
+      <optional>
+        <ref name="ext"/>
+      </optional>
+    </define>
+  </include>
+  <define name="ext">
+    <element name="ext">
+      <ref name="BibDataExtensionType"/>
+    </element>
+  </define>
+  <define name="BibDataExtensionType">
+    <optional>
+      <attribute name="schema-version"/>
+    </optional>
+    <ref name="doctype"/>
+    <optional>
+      <ref name="docsubtype"/>
+    </optional>
+    <optional>
+      <ref name="editorialgroup"/>
+    </optional>
+    <zeroOrMore>
+      <ref name="ics"/>
+    </zeroOrMore>
+    <zeroOrMore>
+      <ref name="structuredidentifier"/>
+    </zeroOrMore>
+  </define>
+  <define name="doctype">
+    <element name="doctype">
+      <optional>
+        <attribute name="abbreviation"/>
+      </optional>
+      <ref name="DocumentType"/>
+    </element>
+  </define>
+  <define name="DocumentType">
+    <value>document</value>
+  </define>
+  <define name="docsubtype">
+    <element name="subdoctype">
+      <ref name="DocumentSubtype"/>
+    </element>
+  </define>
+  <define name="DocumentSubtype">
+    <text/>
+  </define>
+  <define name="editorialgroup">
+    <element name="editorialgroup">
+      <oneOrMore>
+        <ref name="technical-committee"/>
+      </oneOrMore>
+    </element>
+  </define>
+  <define name="technical-committee">
+    <element name="technical-committee">
+      <ref name="IsoWorkgroup"/>
+    </element>
+  </define>
+  <define name="IsoWorkgroup">
+    <optional>
+      <attribute name="number"/>
+    </optional>
+    <optional>
+      <attribute name="type"/>
+    </optional>
+    <optional>
+      <attribute name="identifier"/>
+    </optional>
+    <optional>
+      <attribute name="prefix"/>
+    </optional>
+    <text/>
+  </define>
+  <define name="ics">
+    <element name="ics">
+      <element name="code">
+        <text/>
+      </element>
+      <optional>
+        <element name="text">
+          <text/>
+        </element>
+      </optional>
+    </element>
+  </define>
+  <define name="structuredidentifier">
+    <element name="structuredidentifier">
+      <optional>
+        <attribute name="type"/>
+      </optional>
+      <oneOrMore>
+        <element name="agency">
+          <text/>
+        </element>
+      </oneOrMore>
+      <optional>
+        <element name="class">
+          <text/>
+        </element>
+      </optional>
+      <element name="docnumber">
+        <text/>
+      </element>
+      <optional>
+        <element name="partnumber">
+          <text/>
+        </element>
+      </optional>
+      <optional>
+        <element name="edition">
+          <text/>
+        </element>
+      </optional>
+      <optional>
+        <element name="version">
+          <text/>
+        </element>
+      </optional>
+      <optional>
+        <element name="supplementtype">
+          <text/>
+        </element>
+      </optional>
+      <optional>
+        <element name="supplementnumber">
+          <text/>
+        </element>
+      </optional>
+      <optional>
+        <element name="amendment">
+          <text/>
+        </element>
+      </optional>
+      <optional>
+        <element name="corrigendum">
+          <text/>
+        </element>
+      </optional>
+      <optional>
+        <element name="language">
+          <text/>
+        </element>
+      </optional>
+      <optional>
+        <element name="year">
+          <text/>
+        </element>
+      </optional>
+    </element>
+  </define>
+</grammar>

data/lib/metanorma/standoc/biblio.rng

@@ -33,9 +33,10 @@
       <param name="pattern">([\+\-]?\d{4})((-?)((0[1-9]|1[0-2])((-?)([12]\d|0[1-9]|3[01]))?|W([0-4]\d|5[0-2])(-?[1-7])?|(00[1-9]|0[1-9]\d|[12]\d{2}|3([0-5]\d|6[1-6]))))?</param>
     </data>
   </define>
-  <start>
-    <ref name="bibitem"/>
-  </start>
+  <!-- start = bibitem -->
+  <define name="BibData">
+    <ref name="BibliographicItem"/>
+  </define>
   <define name="status">
     <element name="status">
       <ref name="stage"/>
@@ -170,27 +171,30 @@
   </define>
   <define name="contributor">
     <element name="contributor">
-      <zeroOrMore>
+      <oneOrMore>
         <ref name="role"/>
-      </zeroOrMore>
+      </oneOrMore>
       <ref name="ContributorInfo"/>
     </element>
   </define>
   <define name="role">
     <element name="role">
-      <optional>
-        <attribute name="type">
-          <choice>
-            <value>author</value>
-            <value>performer</value>
-            <value>publisher</value>
-            <value>editor</value>
-            <value>adapter</value>
-            <value>translator</value>
-            <value>distributor</value>
-          </choice>
-        </attribute>
-      </optional>
+      <attribute name="type">
+        <choice>
+          <value>author</value>
+          <value>performer</value>
+          <value>publisher</value>
+          <value>editor</value>
+          <value>adapter</value>
+          <value>translator</value>
+          <value>distributor</value>
+          <value>realizer</value>
+          <value>owner</value>
+          <value>authorizer</value>
+          <value>enabler</value>
+          <value>subject</value>
+        </choice>
+      </attribute>
       <zeroOrMore>
         <ref name="roledescription"/>
       </zeroOrMore>
@@ -512,6 +516,17 @@
   </define>
   <define name="localityStack">
     <element name="localityStack">
+      <optional>
+        <attribute name="connective">
+          <choice>
+            <value>and</value>
+            <value>or</value>
+            <value>from</value>
+            <value>to</value>
+            <value/>
+          </choice>
+        </attribute>
+      </optional>
       <zeroOrMore>
         <ref name="locality"/>
       </zeroOrMore>
@@ -524,6 +539,17 @@
   </define>
   <define name="sourceLocalityStack">
     <element name="sourceLocalityStack">
+      <optional>
+        <attribute name="connective">
+          <choice>
+            <value>and</value>
+            <value>or</value>
+            <value>from</value>
+            <value>to</value>
+            <value/>
+          </choice>
+        </attribute>
+      </optional>
       <zeroOrMore>
         <ref name="sourceLocality"/>
       </zeroOrMore>
@@ -626,6 +652,9 @@
         <ref name="BibItemType"/>
       </attribute>
     </optional>
+    <optional>
+      <attribute name="schema-version"/>
+    </optional>
     <optional>
       <ref name="fetched"/>
     </optional>
@@ -720,6 +749,9 @@
         <ref name="BibItemType"/>
       </attribute>
     </optional>
+    <optional>
+      <attribute name="schema-version"/>
+    </optional>
     <optional>
       <ref name="fetched"/>
     </optional>

data/lib/metanorma/standoc/cleanup.rb

@@ -5,6 +5,7 @@ require_relative "./cleanup_block"
 require_relative "./cleanup_table"
 require_relative "./cleanup_footnotes"
 require_relative "./cleanup_ref"
+require_relative "./cleanup_asciibib"
 require_relative "./cleanup_biblio"
 require_relative "./cleanup_boilerplate"
 require_relative "./cleanup_section"
@@ -26,31 +27,33 @@ module Metanorma
       def cleanup(xmldoc)
         element_name_cleanup(xmldoc)
         passthrough_cleanup(xmldoc)
-        sections_cleanup(xmldoc)
+        sections_cleanup(xmldoc) # feeds: obligations_cleanup, toc_cleanup
         obligations_cleanup(xmldoc)
         para_index_cleanup(xmldoc)
         block_index_cleanup(xmldoc)
         table_cleanup(xmldoc)
         formula_cleanup(xmldoc)
         form_cleanup(xmldoc)
-        sourcecode_cleanup(xmldoc)
+        sourcecode_cleanup(xmldoc) # feeds: callout_cleanup
         figure_cleanup(xmldoc)
+        requirement_cleanup(xmldoc) # feeds: xref_cleanup
         element_name_cleanup(xmldoc)
-        ref_cleanup(xmldoc)
+        ref_cleanup(xmldoc) # feeds: bibitem_cleanup
         note_cleanup(xmldoc)
         clausebefore_cleanup(xmldoc)
         floatingtitle_cleanup(xmldoc)
-        bibitem_cleanup(xmldoc)
+        bibitem_cleanup(xmldoc) # feeds: normref_cleanup, biblio_cleanup,
+        # reference_names, bpart_cleanup
         normref_cleanup(xmldoc)
         biblio_cleanup(xmldoc)
         reference_names(xmldoc)
-        symbols_cleanup(xmldoc)
-        xref_cleanup(xmldoc)
-        concept_cleanup(xmldoc)
-        related_cleanup(xmldoc)
-        origin_cleanup(xmldoc)
+        symbols_cleanup(xmldoc) # feeds: termdef_cleanup
+        xref_cleanup(xmldoc) # feeds: concept_cleanup, origin_cleanup
+        concept_cleanup(xmldoc) # feeds: related_cleanup, termdef_cleanup
+        related_cleanup(xmldoc) # feeds: termdef_cleanup
+        origin_cleanup(xmldoc) # feeds: termdef_cleanup
         bookmark_cleanup(xmldoc)
-        termdef_cleanup(xmldoc)
+        termdef_cleanup(xmldoc) # feeds: iev_cleanup, term_index_cleanup
         RelatonIev::iev_cleanup(xmldoc, @bibdb)
         element_name_cleanup(xmldoc)
         term_index_cleanup(xmldoc)
@@ -60,10 +63,9 @@ module Metanorma
         footnote_cleanup(xmldoc)
         mathml_cleanup(xmldoc)
         script_cleanup(xmldoc)
-        docidentifier_cleanup(xmldoc)
-        requirement_cleanup(xmldoc)
+        docidentifier_cleanup(xmldoc) # feeds: bibdata_cleanup
         bibdata_cleanup(xmldoc)
-        svgmap_cleanup(xmldoc)
+        svgmap_cleanup(xmldoc) # feeds: img_cleanup
         boilerplate_cleanup(xmldoc)
         toc_cleanup(xmldoc)
         metadata_cleanup(xmldoc)
@@ -127,8 +129,8 @@ module Metanorma
 
       def variant_cleanup1(elem)
         elem.xpath("./variant").each do |n|
-          if n.at_xpath("preceding-sibling::node()"\
-                        "[not(self::text()[not(normalize-space())])][1]"\
+          if n.at_xpath("preceding-sibling::node()" \
+                        "[not(self::text()[not(normalize-space())])][1]" \
                         "[self::variantwrap]")
             n.previous_element << n
           else
@@ -151,8 +153,7 @@ module Metanorma
       def metadata_cleanup(xmldoc)
         return if @metadata_attrs.nil? || @metadata_attrs.empty?
 
-        ins = xmldoc.at("//misc-container") ||
-          xmldoc.at("//bibdata").after("<misc-container/>").next_element
+        ins = add_misc_container(xmldoc)
         ins << @metadata_attrs
       end
     end

data/lib/metanorma/standoc/cleanup_amend.rb

@@ -31,8 +31,8 @@ module Metanorma
           amend.children = "<description>#{amend.children.to_xml}</description>"
           return
         end
-        pre = q&.xpath("./preceding-sibling::*")&.remove
-        post = q&.xpath("./following-sibling::*")&.remove
+        pre = q.xpath("./preceding-sibling::*")&.remove
+        post = q.xpath("./following-sibling::*")&.remove
         pre.empty? or amend << "<description>#{pre.to_xml}</description>"
         amend << q.remove
         post.empty? or amend << "<description>#{post.to_xml}</description>"

data/lib/metanorma/standoc/cleanup_asciibib.rb

@@ -0,0 +1,110 @@
+require "set"
+require "relaton_bib"
+
+module Metanorma
+  module Standoc
+    module Cleanup
+      def ref_dl_cleanup(xmldoc)
+        xmldoc.xpath("//clause[@bibitem = 'true']").each do |c|
+          bib = dl_bib_extract(c) or next
+          validate_ref_dl(bib, c)
+          bibitemxml = RelatonBib::BibliographicItem.from_hash(bib).to_xml or next
+          bibitem = Nokogiri::XML(bibitemxml)
+          bibitem.root["id"] = c["id"] if c["id"] && !/^_/.match(c["id"])
+          c.replace(bibitem.root)
+        end
+      end
+
+      # do not accept implicit id
+      def validate_ref_dl(bib, clause)
+        id = bib["id"]
+        id ||= clause["id"] unless /^_/.match?(clause["id"])
+        unless id
+          @log.add("Anchors", clause,
+                   "The following reference is missing an anchor:\n" \
+                   "#{clause.to_xml}")
+          return
+        end
+        @refids << id
+        validate_ref_dl1(bib, id, clause)
+      end
+
+      def validate_ref_dl1(bib, id, clause)
+        bib["title"] or
+          @log.add("Bibliography", clause, "Reference #{id} is missing a title")
+        bib["docid"] or
+          @log.add("Bibliography", clause,
+                   "Reference #{id} is missing a document identifier (docid)")
+      end
+
+      def extract_from_p(tag, bib, key)
+        return unless bib[tag]
+
+        "<#{key}>#{bib[tag].at('p').children}</#{key}>"
+      end
+
+      # if the content is a single paragraph, replace it with its children
+      # single links replaced with uri
+      def p_unwrap(para)
+        elems = para.elements
+        if elems.size == 1 && elems[0].name == "p"
+          link_unwrap(elems[0]).children.to_xml.strip
+        else
+          para.to_xml.strip
+        end
+      end
+
+      def link_unwrap(para)
+        elems = para.elements
+        if elems.size == 1 && elems[0].name == "link"
+          para.at("./link").replace(elems[0]["target"].strip)
+        end
+        para
+      end
+
+      def dd_bib_extract(dtd)
+        return nil if dtd.children.empty?
+
+        dtd.at("./dl") and return dl_bib_extract(dtd)
+        elems = dtd.remove.elements
+        return p_unwrap(dtd) unless elems.size == 1 &&
+          %w(ol ul).include?(elems[0].name)
+
+        elems[0].xpath("./li").each_with_object([]) do |li, ret|
+          ret << p_unwrap(li)
+        end
+      end
+
+      def add_to_hash(bib, key, val)
+        Metanorma::Utils::set_nested_value(bib, key.split("."), val)
+      end
+
+      # definition list, with at most one level of unordered lists
+      def dl_bib_extract(clause, nested = false)
+        dl = clause.at("./dl") or return
+        key = ""
+        bib = dl.xpath("./dt | ./dd").each_with_object({}) do |dtd, m|
+          (dtd.name == "dt" and key = dtd.text.sub(/:+$/, "")) and next
+          add_to_hash(m, key, dd_bib_extract(dtd))
+        end
+        clause.xpath("./clause").each do |c1|
+          key = c1&.at("./title")&.text&.downcase&.strip
+          next unless %w(contributor relation series).include? key
+
+          add_to_hash(bib, key, dl_bib_extract(c1, true))
+        end
+        dl_bib_extract_title(bib, clause, nested)
+      end
+
+      def dl_bib_extract_title(bib, clause, nested)
+        (!nested && clause.at("./title")) or return bib
+        title = clause.at("./title").remove.children.to_xml
+        bib["title"] = [bib["title"]] if bib["title"].is_a?(Hash) ||
+          bib["title"].is_a?(String)
+        bib["title"] ||= []
+        bib["title"] << title if !title.empty?
+        bib
+      end
+    end
+  end
+end