metanorma 1.6.1 → 1.6.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 284943566222cba73b11d131aaa2ce1da77851a8ed5acd60bd11675f78c99eb3
-  data.tar.gz: 6cd2ed331a36e1168169e495183cf92205076b7305fa8e1383424b844f5c114e
+  metadata.gz: ca433d59b111534c114a96b7ce386ccf0267a70cedcbfbe61ef11cf2f42737d3
+  data.tar.gz: c1a73eb056acb7910d255c8348273dd9abf79dd06ef4826ac3ab3d5474543242
 SHA512:
-  metadata.gz: 7a8d401ebdbe17ab843c6ed40113bc4a4b29f5fb78fc7ac0adfb5fb3f1e91056f892efcfb5fc5e26cf74676f0f93e23d3c5933ed88422a7b34f8bef80466ba16
-  data.tar.gz: 1a37a1b0598b2b20dfbbfabed2762d2e5671949f47eb7a830931ba8660edd7a4cf28b551ca42951762edaa21e1bdf5160aff3162163cdca02f3c221e1a3705a3
+  metadata.gz: 2e57292cdc5217cb985be7beb617035ee8edcfb0c9ee3c38cd2c6dc49e71c63468c05d3e6c3000985e93e94a4313f7cf77b24a24c4affe80778f3727c3c7a5b8
+  data.tar.gz: b1a33f6b99669f384af4426cf75e3052f480c0878ec9f1d5716661bc5340e228273f6993930aacba3b1bfe6da7f166791718b0c02c0c71440f83d8642e019355

data/lib/metanorma/collection.rb

@@ -9,16 +9,11 @@ require_relative "util"
 module Metanorma
   # Metanorma collection of documents
   class Collection
-    # @return [String]
     attr_reader :file
 
     # @return [Array<String>] documents-inline to inject the XML into
     #   the collection manifest; documents-external to keeps them outside
-    attr_accessor :directives
-
-    # @return [Hash<String, Metanorma::Document>]
-    attr_accessor :documents, :bibdatas, :coverpage
-
+    attr_accessor :directives, :documents, :bibdatas, :coverpage
     attr_accessor :disambig, :manifest
 
     # @param file [String] path to source file
@@ -92,8 +87,7 @@ module Metanorma
         elem.each_with_object([]) do |(k, v), m|
           m << "<#{k}>#{obj_to_xml(v)}</#{k}>"
         end.join
-      else elem
-      end
+      else elem end
     end
 
     def render(opts)
@@ -115,7 +109,7 @@ module Metanorma
       private
 
       def parse_xml(file)
-        xml = Nokogiri::XML File.read(file, encoding: "UTF-8") { |c| c.huge }
+        xml = Nokogiri::XML(File.read(file, encoding: "UTF-8"), &:huge)
         (b = xml.at("/xmlns:metanorma-collection/xmlns:bibdata")) and
           bd = Relaton::Cli.parse_xml(b)
         mnf_xml = xml.at("/xmlns:metanorma-collection/xmlns:manifest")
@@ -133,15 +127,14 @@ module Metanorma
       # TODO refine
       def directives_from_xml(dir)
         dir.each_with_object([]) do |d, m|
-          if d.at("./xmlns:value")
-            m << x.xpath("./xmlns:value").map(&:text)
-          elsif d.at("./*")
-            out = d.elements.each_with_object({}) do |e, ret|
-              ret[e.name] = e.children.to_xml
-            end
-            m << out
-          else m << d.children.to_xml
-          end
+          m << if d.at("./xmlns:value")
+                 x.xpath("./xmlns:value").map(&:text)
+               elsif d.at("./*")
+                 d.elements.each_with_object({}) do |e, ret|
+                   ret[e.name] = e.children.to_xml
+                 end
+               else d.children.to_xml
+               end
         end
       end
 
@@ -174,8 +167,7 @@ module Metanorma
       # @param xml [Nokogiri::XML::Element, nil]
       # @return [String, nil]
       def pref_final_content(xml)
-        return unless xml
-
+        xml or return
         <<~CONT
 
             == #{xml.at('title')&.text}
@@ -212,13 +204,12 @@ module Metanorma
     # @return [String] XML
     def sections(cnt)
       c = Asciidoctor.convert(cnt, backend: doctype.to_sym, header_footer: true)
-      Nokogiri::XML(c).at("//xmlns:sections").children.to_xml
+      Nokogiri::XML(c, &:huge).at("//xmlns:sections").children.to_xml
     end
 
     # @param builder [Nokogiri::XML::Builder]
     def doccontainer(builder)
-      return unless Array(@directives).include? "documents-inline"
-
+      Array(@directives).include? "documents-inline" or return
       documents.each_with_index do |(_, d), i|
         doccontainer1(builder, d, i)
       end
@@ -235,16 +226,13 @@ module Metanorma
       end
     end
 
-    # @return [String]
     def doctype
       @doctype ||= fetch_doctype || "standoc"
     end
 
-    # @return [String]
     def fetch_doctype
       docid = @bibdata.docidentifier.first
-      return unless docid
-
+      docid or return
       docid.type&.downcase || docid.id&.sub(/\s.*$/, "")&.downcase
     end
   end

data/lib/metanorma/collection_fileparse.rb

@@ -1,82 +1,40 @@
 module Metanorma
   # XML collection renderer
   class CollectionRenderer
-=begin
-    # map locality type and label (e.g. "clause" "1") to id = anchor for
-    # a document
-    # Note: will only key clauses, which have unambiguous reference label in
-    # locality. Notes, examples etc with containers are just plunked against
-    # UUIDs, so that their IDs can at least be registered to be tracked
-    # as existing.
-    def read_anchors(xml)
-      xrefs = @isodoc.xref_init(@lang, @script, @isodoc, @isodoc.i18n,
-                                { locale: @locale })
-      xrefs.parse xml
-      xrefs.get.each_with_object({}) do |(k, v), ret|
-        read_anchors1(k, v, ret)
-      end
-    end
-
-    def read_anchors1(key, val, ret)
-      val[:type] ||= "clause"
-      ret[val[:type]] ||= {}
-      index = if val[:container] || val[:label].nil? || val[:label].empty?
-                UUIDTools::UUID.random_create.to_s
-              else val[:label]
-              end
-      ret[val[:type]][index] = key
-      ret[val[:type]][val[:value]] = key if val[:value]
-    end
-
-    # @param id [String]
-    # @param read [Boolean]
-    # @return [Array<String, nil>]
-    def xml_file(id, read)
-      file = @xml.at(ns("//doc-container[@id = '#{id}']")).to_xml if read
-      filename = "#{id}.html"
-      [file, filename]
-    end
-=end
-
     # @param bib [Nokogiri::XML::Element]
     # @param identifier [String]
     def update_bibitem(bib, identifier)
       docid = get_bibitem_docid(bib, identifier) or return
       newbib = dup_bibitem(docid, bib)
+      _file, url = @files
+        .targetfile_id(docid, relative: true, read: false,
+                              doc: !@files.get(docid, :attachment))
+      dest = newbib.at("./docidentifier") || newbib.at(ns("./docidentifier"))
+      dest.previous = "<uri type='citation'>#{url}</uri>"
       bib.replace(newbib)
-      #_file, url = targetfile(@files[docid], relative: true, read: false, doc: !@files[docid][:attachment])
-      _file, url = @files.targetfile_id(docid, relative: true, read: false, doc: !@files.get(docid,:attachment))
-      uri_node = Nokogiri::XML::Node.new "uri", newbib.document
-      uri_node[:type] = "citation"
-      uri_node.content = url
-      newbib.at(ns("./docidentifier")).previous = uri_node
     end
 
     def get_bibitem_docid(bib, identifier)
       # IDs for repo references are untyped by default
       docid = bib.at(ns("./docidentifier[not(@type)]")) ||
         bib.at(ns("./docidentifier"))
-      docid &&= @c.decode(@isodoc
-          .docid_prefix(docid["type"], docid.children.to_xml)).gsub(/\s/, " ")
-      #if @files[docid] then docid
-      if @files.get(docid) || @files.get("X #{docid}") then docid
+      docid &&= docid_prefix(docid)
+      if @files.get(docid) then docid
       else
-        #require "debug"; binding.b
         fail_update_bibitem(docid, identifier)
         nil
       end
     end
 
-    def fail_update_bibitem(docid, identifier)
-      error = "[metanorma] Cannot find crossreference to document #{docid} " \
-              "in document #{identifier}."
-      @log&.add("Cross-References", nil, error)
-      Util.log(error, :warning)
+    def docid_prefix(docid)
+      type = docid["type"]
+      type == "metanorma-collection" and type = nil
+      @c.decode(@isodoc
+          .docid_prefix(type, docid.children.to_xml)).gsub(/\s/, " ")
     end
 
     def dup_bibitem(docid, bib)
-      #newbib = @files[docid][:bibdata].dup
-      newbib = @files.get(docid,:bibdata).dup
+      newbib = @files.get(docid, :bibdata).dup
       newbib.name = "bibitem"
       newbib["hidden"] = "true"
       newbib&.at("./*[local-name() = 'ext']")&.remove
@@ -95,16 +53,24 @@ module Metanorma
     #   anchor to filename
     # @return [String] XML content
     def update_xrefs(file, identifier, internal_refs)
-      docxml = Nokogiri::XML(file) { |config| config.huge }
+      docxml = file.is_a?(String) ? Nokogiri::XML(file, &:huge) : file
       supply_repo_ids(docxml)
       update_indirect_refs_to_docs(docxml, internal_refs)
+      ids = @files.get(identifier, :ids)
       @files.add_document_suffix(identifier, docxml)
+      svgmap_resolve(datauri_encode(docxml), ids)
       update_direct_refs_to_docs(docxml, identifier)
-      svgmap_resolve(datauri_encode(docxml))
       hide_refs(docxml)
+      @files.get(identifier, :sectionsplit_output) and eref2link(docxml)
       docxml.to_xml
     end
 
+    def eref2link(docxml)
+      isodoc = IsoDoc::PresentationXMLConvert.new({})
+      isodoc.bibitem_lookup(docxml)
+      isodoc.eref2link(docxml)
+    end
+
     def hide_refs(docxml)
       docxml.xpath(ns("//references[bibitem][not(./bibitem[not(@hidden) or " \
                       "@hidden = 'false'])]")).each do |f|
@@ -114,41 +80,31 @@ module Metanorma
 
     def supply_repo_ids(docxml)
       docxml.xpath(ns("//bibitem[not(ancestor::bibitem)]")).each do |b|
-        next if b&.at(ns("./docidentifier[@type = 'repository']"))
-
+        b.at(ns("./docidentifier[@type = 'repository']")) and next
         b.xpath(ns("./docidentifier")).each do |docid|
-          #id = @c.decode(@isodoc
-          #.docid_prefix(docid["type"], docid.children.to_xml))
           id = @isodoc
-          .docid_prefix(docid["type"], docid.children.to_xml)
-          #@files[id] or next
+            .docid_prefix(docid["type"], docid.children.to_xml)
           @files.get(id) or next
           docid.next = "<docidentifier type='repository'>" \
-                       "current-metanorma-collection/#{id}"
+                       "current-metanorma-collection/#{id}</docidentifier>"
         end
       end
     end
 
-    def datauri_encode(docxml)
-      docxml.xpath(ns("//image")).each do |i|
-        i["src"] = Metanorma::Utils::datauri(i["src"])
-      end
-      docxml
-    end
-
-    def svgmap_resolve(docxml)
+    def svgmap_resolve(docxml, ids)
       isodoc = IsoDoc::PresentationXMLConvert.new({})
       isodoc.bibitem_lookup(docxml)
       docxml.xpath(ns("//svgmap//eref")).each do |e|
-        svgmap_resolve1(e, isodoc)
+        svgmap_resolve1(e, isodoc, docxml, ids)
       end
       Metanorma::Utils::svgmap_rewrite(docxml, "")
     end
 
-    def svgmap_resolve1(eref, isodoc)
+    def svgmap_resolve1(eref, isodoc, _docxml, ids)
       href = isodoc.eref_target(eref)
       return if href == "##{eref['bibitemid']}" ||
-        (href =~ /^#/ && !docxml.at("//*[@id = '#{href.sub(/^#/, '')}']"))
+        # (href =~ /^#/ && !docxml.at("//*[@id = '#{href.sub(/^#/, '')}']"))
+        (href =~ /^#/ && !ids[href.sub(/^#/, "")])
 
       eref["target"] = href.strip
       eref.name = "link"
@@ -161,21 +117,21 @@ module Metanorma
     # Preferably with anchor, and is a job to realise dynamic lookup
     # of localities.
     def update_direct_refs_to_docs(docxml, identifier)
-      erefs = collect_erefs(docxml)
-      docxml.xpath(ns("//bibitem[not(ancestor::bibitem)]")).each do |b|
+      @ncnames = {}
+      erefs = Util::gather_citeases(docxml)
+      docxml.xpath(ns("//bibitem")).each do |b|
         docid = b.at(ns("./docidentifier[@type = 'repository']"))
         (docid && %r{^current-metanorma-collection/}.match(docid.text)) or next
         update_bibitem(b, identifier)
         docid = docid_to_citeas(b) or next
-        erefs[:citeas][docid] and update_anchors(b, docxml, docid)
+        erefs[docid] and update_anchors(b, docid, erefs[docid])
       end
     end
 
     def docid_to_citeas(bib)
       docid = bib.at(ns("./docidentifier[@primary = 'true']")) ||
         bib.at(ns("./docidentifier")) or return
-      @c.decode(@isodoc
-          .docid_prefix(docid["type"], docid.children.to_xml))
+      docid_prefix(docid)
     end
 
     def collect_erefs(docxml)
@@ -189,31 +145,35 @@ module Metanorma
     # Resolve erefs to a container of ids in another doc,
     # to an anchor eref (direct link)
     def update_indirect_refs_to_docs(docxml, internal_refs)
+      bibitems = Util::gather_bibitems(docxml)
+      erefs = Util::gather_bibitemids(docxml)
       internal_refs.each do |schema, ids|
         ids.each do |id, file|
-          update_indirect_refs_to_docs1(docxml, schema, id, file)
+          update_indirect_refs_to_docs1(docxml, "#{schema}_#{id}",
+                                        file, bibitems, erefs)
         end
       end
     end
 
-    def update_indirect_refs_to_docs1(docxml, schema, id, file)
-      docxml.xpath(ns("//eref[@bibitemid = '#{schema}_#{id}']")).each do |e|
+    def update_indirect_refs_to_docs1(_docxml, key, file, bibitems, erefs)
+      erefs[key]&.each do |e|
+        # docxml.xpath(ns("//eref[@bibitemid = '#{key}']")).each do |e|
         e["citeas"] = file
-        if a = e.at(ns(".//locality[@type = 'anchor']/referenceFrom"))
+        a = e.at(ns(".//locality[@type = 'anchor']/referenceFrom")) and
           a.children = "#{a.text}_#{Metanorma::Utils::to_ncname(file)}"
-        end
       end
-      docid = docxml.at(ns("//bibitem[@id = '#{schema}_#{id}']/" \
-                           "docidentifier[@type = 'repository']")) or return
+      docid = bibitems[key]&.at(ns("./docidentifier[@type = 'repository']")) or
+        return
       docid.children = "current-metanorma-collection/#{file}"
-      docid.previous = "<docidentifier type='X'>#{file}</docidentifier>"
+      docid.previous =
+        "<docidentifier type='metanorma-collection'>#{file}</docidentifier>"
     end
 
     # update crossrefences to other documents, to include
     # disambiguating document suffix on id
-    def update_anchors(bib, docxml, docid) # rubocop:disable Metrics/AbcSize
-      docxml.xpath("//xmlns:eref[@citeas = '#{docid}']").each do |e|
-        #if @files[docid] then update_anchor_loc(bib, e, docid)
+    def update_anchors(bib, docid, erefs)
+      #docxml.xpath("//xmlns:eref[@citeas = '#{docid}']").each do |e|
+      erefs.each do |e|
         if @files.get(docid) then update_anchor_loc(bib, e, docid)
         else
           e << "<strong>** Unresolved reference to document #{docid} " \
@@ -223,13 +183,12 @@ module Metanorma
     end
 
     def update_anchor_loc(bib, eref, docid)
-      loc = eref.at(ns(".//locality[@type = 'anchor']")) or
+      loc = eref.at(".//xmlns:locality[@type = 'anchor']") or
         return update_anchor_create_loc(bib, eref, docid)
-      document_suffix = Metanorma::Utils::to_ncname(docid)
-      ref = loc.at(ns("./referenceFrom")) or return
-      anchor = "#{ref.text}_#{document_suffix}"
-      #return unless @files[docid][:anchors].inject([]) do |m, (_, x)|
-      return unless @files.get(docid,:anchors).inject([]) do |m, (_, x)|
+      @ncnames[docid] ||= Metanorma::Utils::to_ncname(docid)
+      ref = loc.at("./xmlns:referenceFrom") or return
+      anchor = "#{ref.text}_#{@ncnames[docid]}"
+      return unless @files.get(docid, :anchors).inject([]) do |m, (_, x)|
         m += x.values
       end.include?(anchor)
 
@@ -243,41 +202,36 @@ module Metanorma
       type = ins.at(ns("./locality/@type"))&.text
       type = "clause" if type == "annex"
       ref = ins.at(ns("./locality/referenceFrom"))&.text
-      #anchor = @files[docid][:anchors].dig(type, ref) or return
-      anchor = @files.get(docid,:anchors).dig(type, ref) or return
+      anchor = @files.get(docid, :anchors).dig(type, ref) or return
       ins << "<locality type='anchor'><referenceFrom>#{anchor.sub(/^_/, '')}" \
              "</referenceFrom></locality>"
     end
 
     # gather internal bibitem references
     def gather_internal_refs
-      #@files.each.with_object({}) do |(_, x), refs|
       @files.keys.each_with_object({}) do |i, refs|
-        #x[:attachment] and next
-        @files.get(i,:attachment) and next
-        #file, = targetfile(x, read: true)
+        @files.get(i, :attachment) and next
         file, = @files.targetfile_id(i, read: true)
-        Nokogiri::XML(file)
-          .xpath(ns("//bibitem[@type = 'internal']/" \
-                    "docidentifier[@type = 'repository']")).each do |d|
-          a = d.text.split(%r{/}, 2)
-          a.size > 1 or next
-          refs[a[0]] ||= {}
-          refs[a[0]][a[1]] = true
-        end
+        gather_internal_refs1(file, refs)
+      end
+    end
+
+    def gather_internal_refs1(file, refs)
+      Nokogiri::XML(file, &:huge)
+        .xpath(ns("//bibitem[@type = 'internal']/" \
+                  "docidentifier[@type = 'repository']")).each do |d|
+        a = d.text.split(%r{/}, 2)
+        a.size > 1 or next
+        refs[a[0]] ||= {}
+        refs[a[0]][a[1]] = true
       end
     end
 
     # resolve file location for the target of each internal reference
     def locate_internal_refs
-      #require 'debug'; binding.b
       refs = gather_internal_refs
-      #@files.keys.reject { |k| @files[k][:attachment] }.each do |identifier|
-      @files.keys.reject { |k| @files.get(k,:attachment) }.each do |identifier|
-        #id = @c.decode(@isodoc.docid_prefix("", identifier.dup))
-        #locate_internal_refs1(refs, identifier, @files[id])
-        id = @isodoc.docid_prefix("", identifier.dup)
-        locate_internal_refs1(refs, identifier, id)
+      @files.keys.reject { |k| @files.get(k, :attachment) }.each do |ident|
+        locate_internal_refs1(refs, ident, @isodoc.docid_prefix("", ident.dup))
       end
       refs.each do |schema, ids|
         ids.each do |id, key|
@@ -287,19 +241,20 @@ module Metanorma
       refs
     end
 
-    #def locate_internal_refs1(refs, identifier, filedesc)
     def locate_internal_refs1(refs, identifier, ident)
-      #file, _filename = targetfile(filedesc, read: true)
-      file, _filename = @files.targetfile_id(ident, read: true)
-      xml = Nokogiri::XML(file) { |config| config.huge }
-      t = xml.xpath("//*/@id").each_with_object({}) { |i, x| x[i.text] = true }
+      t = locate_internal_refs1_prep(ident)
       refs.each do |schema, ids|
         ids.keys.select { |id| t[id] }.each do |id|
-          n = xml.at("//*[@id = '#{id}']") and
-            n.at("./ancestor-or-self::*[@type = '#{schema}']") and
+          t[id].at("./ancestor-or-self::*[@type = '#{schema}']") and
             refs[schema][id] = identifier
         end
       end
     end
+
+    def locate_internal_refs1_prep(ident)
+      file, _filename = @files.targetfile_id(ident, read: true)
+      xml = Nokogiri::XML(file, &:huge)
+      xml.xpath("//*[@id]").each_with_object({}) { |i, x| x[i["id"]] = i }
+    end
   end
 end