metanorma 1.2.8 → 1.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 72a5ff12b5b2793f4d8b8ced7b5cdc747a7b59cb938300f1e288150787d82951
-  data.tar.gz: 2a8f1a0fda7ceed680324833c18ab595d9de381aab36f3a5848040743029dfc8
+  metadata.gz: c6503cb40980d520fe14a15845b75b748a6cdd40c9206e2218676ac9f2500ff5
+  data.tar.gz: cb169e73038bae1a2c8d70170af1eb4956342056bd4d2c88341f0f08d86462ee
 SHA512:
-  metadata.gz: d8ad8063e83a0d51494cb9d69f552676b27ac2022c9d477af6e12902bc2a9c7ceddc07d51965c594ab3ddf92839da6071d8a407836774e26c676b2fd3c9d584d
-  data.tar.gz: 18b394b6fd082a3072d543b120656ee181a68ce4c1a4cf3f0551f53719e07a96ec1b2140d61fac86e328370464e024df38872a0d85bc0b95aee588bb3e844982
+  metadata.gz: 66fb9df1d800763aafccdbb696d755627dfda9d29530200d048f91b59f46e081ff3ffe5f5504579d54ef2667ae33d821d1642329bf21eec6b04e1503e992454c
+  data.tar.gz: d46d5ba951e50d47e75cd86c627c65958a490a4cb6e74ba4f61523b301caa53ae6eddc57a72dd58653b0b5a2894110ff5e0d55d795ac764ef303d0f607424667

data/lib/metanorma/collection.rb

@@ -168,6 +168,7 @@ module Metanorma
       return unless Array(@directives).include? "documents-inline"
 
       documents.each_with_index do |(_, d), i|
+        next if d.attachment
         id = format("doc%<index>09d", index: i)
         builder.send("doc-container", id: id) { |b| d.to_xml b }
       end

data/lib/metanorma/collection_fileparse.rb

@@ -0,0 +1,213 @@
+module Metanorma
+  # XML collection renderer
+  class CollectionRenderer
+    # 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)
+      ret = {}
+      xrefs = @isodoc.xref_init(@lang, @script, @isodoc, @isodoc.i18n, {})
+      xrefs.parse xml
+      xrefs.get.each do |k, v|
+        ret[v[:type]] ||= {}
+        index = v[:container] || v[:label].nil? || v[:label].empty? ?
+          UUIDTools::UUID.random_create.to_s : v[:label]
+        ret[v[:type]][index] = k
+      end
+      ret
+    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
+
+    # @param bib [Nokogiri::XML::Element]
+    # @param identifier [String]
+    def update_bibitem(bib, identifier) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
+      docid = bib&.at(ns("./docidentifier"))&.text
+      unless @files[docid]
+        error = "[metanorma] Cannot find crossreference to document #{docid} "\
+          "in document #{identifier}."
+        @log.add("Cross-References", nil, error)
+        Util.log(error, :warning)
+        return
+      end
+      id = bib["id"]
+      newbib = bib.replace(@files[docid][:bibdata])
+      newbib.name = "bibitem"
+      newbib["id"] = id
+      newbib["hidden"] = "true"
+      newbib&.at(ns("./ext"))&.remove
+      _file, url = targetfile(@files[docid], false, !@files[docid][:attachment])
+      uri_node = Nokogiri::XML::Node.new "uri", newbib
+      uri_node[:type] = "citation"
+      uri_node.content = url
+      newbib.at(ns("./docidentifier")).previous = uri_node
+    end
+
+    # Resolves direct links to other files in collection
+    # (repo(current-metanorma-collection/x),
+    # and indirect links to other files in collection
+    # (bibitem[@type = 'internal'] pointing to a file anchor
+    # in another file in the collection)
+    # @param file [String] XML content
+    # @param identifier [String] docid
+    # @param internal_refs [Hash{String=>Hash{String=>String}] schema name to anchor to filename
+    # @return [String] XML content
+    def update_xrefs(file, identifier, internal_refs)
+      docxml = Nokogiri::XML(file)
+      update_indirect_refs_to_docs(docxml, internal_refs)
+      add_document_suffix(identifier, docxml)
+      update_direct_refs_to_docs(docxml, identifier)
+      svgmap_resolve(datauri_encode(docxml))
+      docxml.xpath(ns("//references[not(./bibitem[not(@hidden) or @hidden = 'false'])]")).each do |f|
+        f["hidden"] = "true"
+      end
+      docxml.to_xml
+    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)
+      isodoc = IsoDoc::Convert.new({})
+      docxml.xpath(ns("//svgmap//eref")).each do |e|
+        href = isodoc.eref_target(e)
+        next if href == "##{e['bibitemid']}"
+
+        if href.match(/^#/)
+          next unless docxml.at("//*[@id = '#{href.sub(/^#/, '')}']")
+        end
+        e["target"] = href.strip
+        e.name = "link"
+        e&.elements&.remove
+      end
+      Metanorma::Utils::svgmap_rewrite(docxml, "")
+    end
+
+    # repo(current-metanorma-collection/ISO 17301-1:2016)
+    # replaced by bibdata of "ISO 17301-1:2016" in situ as bibitem.
+    # Any erefs to that bibitem id are replaced with relative URL
+    # Preferably with anchor, and is a job to realise dynamic lookup
+    # of localities.
+    def update_direct_refs_to_docs(docxml, identifier)
+      docxml.xpath(ns("//bibitem[not(ancestor::bibitem)]")).each do |b|
+        docid = b&.at(ns("./docidentifier[@type = 'repository']"))&.text
+        next unless docid && %r{^current-metanorma-collection/}.match(docid)
+        update_bibitem(b, identifier)
+        update_anchors(b, docxml, docid)
+      end
+    end
+
+    # 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)
+      internal_refs.each do |schema, ids|
+        ids.each do |id, file|
+          update_indirect_refs_to_docs1(docxml, schema, id, file)
+        end
+      end
+    end
+
+    def update_indirect_refs_to_docs1(docxml, schema, id, file)
+      docxml.xpath(ns("//eref[@bibitemid = '#{schema}_#{id}']")).each do |e|
+        e["citeas"] = file
+      end
+      docid = docxml.at(ns("//bibitem[@id = '#{schema}_#{id}']/"\
+                           "docidentifier[@type = 'repository']")) or return
+      docid.children = "current-metanorma-collection/#{file}"
+      docid.previous = "<docidentifier type='X'>#{file}</docidentifier>"
+    end
+
+    # update crossrefences to other documents, to include disambiguating document suffix on id
+    def update_anchors(bib, docxml, _id) # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
+      docid = bib&.at(ns("./docidentifier"))&.text
+      docxml.xpath("//xmlns:eref[@citeas = '#{docid}']").each do |e|
+        if @files[docid]
+          update_anchor_loc(bib, e, docid)
+        else
+          e << "<strong>** Unresolved reference to document #{docid}, id #{e['bibitemid']}</strong>"
+        end
+      end
+    end
+
+    def update_anchor_loc(bib, e, docid)
+      loc = e.at(ns(".//locality[@type = 'anchor']")) or
+        return update_anchor_create_loc(bib, e, docid)
+      document_suffix = Metanorma::Utils::to_ncname(docid)
+      ref = loc.at(ns("./referenceFrom")) || return
+      anchor = "#{ref.text}_#{document_suffix}"
+      return unless @files[docid][:anchors].inject([]) do |m, (_, x)|
+        m+= x.values
+      end.include?(anchor)
+      ref.content = anchor
+    end
+
+    # if there is a crossref to another document, with no anchor, retrieve the
+    # anchor given the locality, and insert it into the crossref
+    def update_anchor_create_loc(bib, e, docid)
+      ins = e.at(ns("./localityStack")) || return
+      type = ins&.at(ns("./locality/@type"))&.text
+      ref = ins&.at(ns("./locality/referenceFrom"))&.text
+      (anchor = @files[docid][:anchors][type][ref]) || return
+      ref_from = Nokogiri::XML::Node.new "referenceFrom", bib
+      ref_from.content = anchor.sub(/^_/, "")
+      locality = Nokogiri::XML::Node.new "locality", bib
+      locality[:type] = "anchor"
+      locality.add_child ref_from
+      ins << locality
+    end
+
+    # gather internal bibitem references
+    def gather_internal_refs
+      @files.each_with_object({}) do |(_, x), refs|
+        next if x[:attachment]
+        file, _ = targetfile(x, 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
+      end
+    end
+
+    # resolve file location for the target of each internal reference
+    def locate_internal_refs
+      refs = gather_internal_refs
+      @files.each do |identifier, x|
+        next if x[:attachment]
+
+        file, _filename = targetfile(x, true)
+        docxml = Nokogiri::XML(file)
+        refs.each do |schema, ids|
+          ids.each_key do |id|
+            n = docxml.at("//*[@id = '#{id}']") and
+              n.at("./ancestor-or-self::*[@type = '#{schema}']") and
+              refs[schema][id] = identifier
+          end
+        end
+      end
+      refs.each do |schema, ids|
+        ids.each do |id, key|
+          key == true and refs[schema][id] = "Missing:#{schema}:#{id}"
+        end
+      end
+      refs
+    end
+  end
+end

data/lib/metanorma/collection_fileprocess.rb

@@ -2,6 +2,7 @@
 
 require "isodoc"
 require "metanorma-utils"
+require_relative "collection_fileparse"
 
 module Metanorma
   # XML collection renderer
@@ -15,11 +16,9 @@ module Metanorma
       files = {}
       @xml.xpath(ns("//docref")).each do |d|
         identifier = d.at(ns("./identifier")).text
-        files[identifier] = if d["fileref"]
-                              { type: "fileref",
-                                ref: File.join(path, d["fileref"]) }
-                            else { type: "id", ref: d["id"] }
-                            end
+        files[identifier] = file_entry(d, path)
+        next if files[identifier][:attachment]
+
         file, _filename = targetfile(files[identifier], true)
         xml = Nokogiri::XML(file)
         add_document_suffix(identifier, xml)
@@ -29,6 +28,17 @@ module Metanorma
       files
     end
 
+    def file_entry(docref, path)
+      ret = if docref["fileref"]
+              { type: "fileref", ref: File.join(path, docref["fileref"]),
+                rel_path: docref["fileref"] }
+            else
+              { type: "id", ref: docref["id"] }
+            end
+      ret[:attachment] = docref["attachment"] if docref["attachment"]
+      ret
+    end
+
     def add_suffix_to_attributes(doc, suffix, tag_name, attribute_name)
       doc.xpath(ns("//#{tag_name}[@#{attribute_name}]")).each do |elem|
         elem.attributes[attribute_name].value =
@@ -38,199 +48,42 @@ module Metanorma
 
     def add_document_suffix(identifier, doc)
       document_suffix = Metanorma::Utils::to_ncname(identifier)
-      [%w[* id],
-      %w[* bibitemid],
-      %w[review from],
-      %w[review to],
-      %w[index to],
-      %w[xref target],
-      %w[callout target]]
-      .each do |(tag_name, attribute_name)|
+      [%w[* id], %w[* bibitemid], %w[review from],
+       %w[review to], %w[index to], %w[xref target],
+       %w[callout target]]
+        .each do |(tag_name, attribute_name)|
         add_suffix_to_attributes(doc, document_suffix, tag_name, attribute_name)
       end
     end
 
-    # map locality type and label (e.g. "clause" "1") to id = anchor for
-    # a document
-    def read_anchors(xml)
-      ret = {}
-      xrefs = @isodoc.xref_init(@lang, @script, @isodoc, @isodoc.i18n, {})
-      xrefs.parse xml
-      xrefs.get.each do |k, v|
-        ret[v[:type]] ||= {}
-        index = v[:container] || v[:label].nil? || v[:label].empty? ? 
-          UUIDTools::UUID.random_create.to_s : v[:label]
-        # Note: will only key clauses, which have unambiguous reference label in locality.
-        # Notes, examples etc with containers are just plunked agaisnt UUIDs, so that their
-        # IDs can at least be registered to be tracked as existing.
-        ret[v[:type]][index] = k
-      end
-      ret
-    end
-
     # return file contents + output filename for each file in the collection,
     # given a docref entry
     # @param data [Hash]
     # @param read [Boolean]
     # @return [Array<String, nil>]
-    def targetfile(data, read = false)
-      if data[:type] == "fileref" then ref_file data[:ref], read
+    def targetfile(data, read = false, doc = true)
+      if data[:type] == "fileref" then ref_file data[:ref], read, doc
       else xml_file data[:id], read
       end
     end
 
     # @param ref [String]
     # @param read [Boolean]
+    # @param doc [Boolean]
     # @return [Array<String, nil>]
-    def ref_file(ref, read)
+    def ref_file(ref, read, doc)
       file = File.read(ref, encoding: "utf-8") if read
-      filename = ref.sub(/\.xml$/, ".html")
+      filename = ref.dup
+      filename.sub!(/\.xml$/, ".html") if doc
       [file, filename]
     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
-
-    # @param bib [Nokogiri::XML::Element]
-    # @param identifier [String]
-    def update_bibitem(bib, identifier) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
-      docid = bib&.at(ns("./docidentifier"))&.text
-      unless @files[docid]
-        error = "[metanorma] Cannot find crossreference to document #{docid} in document #{identifier}."
-        @log.add("Cross-References", nil, error)
-        Util.log(error, :warning)
-        return
-      end
-      id = bib["id"]
-      newbib = bib.replace(@files[docid][:bibdata])
-      newbib.name = "bibitem"
-      newbib["id"] = id
-      newbib["hidden"] = "true"
-      newbib&.at(ns("./ext"))&.remove
-      _file, url = targetfile(@files[docid], false)
-      uri_node = Nokogiri::XML::Node.new "uri", newbib
-      uri_node[:type] = "citation"
-      uri_node.content = url
-      newbib.at(ns("./docidentifier")).previous = uri_node
-    end
-
-    # Resolves direct links to other files in collection (repo(current-metanorma-collection/x),
-    # and indirect links to other files in collection (bibitem[@type = 'internal'] pointing to a file anchor
-    # in another file in the collection)
-    # @param file [String] XML content
-    # @param identifier [String] docid
-    # @param internal_refs [Hash{String=>Hash{String=>String}] schema name to anchor to filename
-    # @return [String] XML content
-    def update_xrefs(file, identifier, internal_refs)
-      docxml = Nokogiri::XML(file)
-      update_indirect_refs_to_docs(docxml, internal_refs)
-      add_document_suffix(identifier, docxml)
-      update_direct_refs_to_docs(docxml, identifier)
-      svgmap_resolve(datauri_encode(docxml))
-      docxml.xpath(ns("//references[not(./bibitem[not(@hidden) or @hidden = 'false'])]")).each do |f|
-        f["hidden"] = "true"
-      end
-      docxml.to_xml
-    end
-
-    def datauri_encode(docxml)
-      docxml.xpath(ns("//image")).each { |i| i["src"] = Metanorma::Utils::datauri(i["src"]) }
-      docxml
-    end
-
-    def svgmap_resolve(docxml)
-      isodoc = IsoDoc::Convert.new({})
-      docxml.xpath(ns("//svgmap//eref")).each do |e|
-        href = isodoc.eref_target(e)
-        next if href == "#" + e["bibitemid"]
-        if href.match(/^#/)
-          next unless docxml.at("//*[@id = '#{href.sub(/^#/, '')}']")
-        end
-        e["target"] = href.strip
-        e.name = "link"
-        e&.elements&.remove
-      end
-      Metanorma::Utils::svgmap_rewrite(docxml, "")
-    end
-
-    # repo(current-metanorma-collection/ISO 17301-1:2016)
-    # replaced by bibdata of "ISO 17301-1:2016" in situ as bibitem.
-    # Any erefs to that bibitem id are replaced with relative URL
-    # Preferably with anchor, and is a job to realise dynamic lookup of localities.
-    def update_direct_refs_to_docs(docxml, identifier)
-      docxml.xpath(ns("//bibitem[not(ancestor::bibitem)]")).each do |b|
-        docid = b&.at(ns("./docidentifier[@type = 'repository']"))&.text
-        next unless docid && %r{^current-metanorma-collection/}.match(docid)
-        update_bibitem(b, identifier)
-        update_anchors(b, docxml, docid)
-      end
-    end
-
-    # 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)
-      internal_refs.each do |schema, ids|
-        ids.each do |id, file|
-          update_indirect_refs_to_docs1(docxml, schema, id, file)
-        end
-      end
-    end
-
-    def update_indirect_refs_to_docs1(docxml, schema, id, file)
-      docxml.xpath(ns("//eref[@bibitemid = '#{schema}_#{id}']")).each do |e|
-        e["citeas"] = file
-      end
-      docid = docxml.at(ns("//bibitem[@id = '#{schema}_#{id}']/docidentifier[@type = 'repository']")) or return
-      docid.children = "current-metanorma-collection/#{file}"
-      docid.previous = "<docidentifier type='X'>#{file}</docidentifier>"
-    end
-
-    # update crossrefences to other documents, to include disambiguating document suffix on id
-    def update_anchors(bib, docxml, _id) # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
-      docid = bib&.at(ns("./docidentifier"))&.text
-      docxml.xpath("//xmlns:eref[@citeas = '#{docid}']").each do |e|
-        if @files[docid]
-          update_anchor_loc(bib, e, docid)
-        else
-          e << "<strong>** Unresolved reference to document #{docid}, id #{e['bibitemid']}</strong>"
-        end
-      end
-    end
-
-    def update_anchor_loc(bib, e, docid)
-      loc = e.at(ns(".//locality[@type = 'anchor']")) or return update_anchor_create_loc(bib, e, docid)
-      document_suffix = Metanorma::Utils::to_ncname(docid)
-      ref = loc.at(ns("./referenceFrom")) || return
-      anchor = "#{ref.text}_#{document_suffix}"
-      return unless @files[docid][:anchors].inject([]) { |m, (_, x)| m+= x.values }.include?(anchor)
-      ref.content = anchor
-    end
-
-    # if there is a crossref to another document, with no anchor, retrieve the
-    # anchor given the locality, and insert it into the crossref
-    def update_anchor_create_loc(bib, e, docid)
-      ins = e.at(ns("./localityStack")) || return
-      type = ins&.at(ns("./locality/@type"))&.text
-      ref = ins&.at(ns("./locality/referenceFrom"))&.text
-      (anchor = @files[docid][:anchors][type][ref]) || return
-      ref_from = Nokogiri::XML::Node.new "referenceFrom", bib
-      ref_from.content = anchor.sub(/^_/, "")
-      locality = Nokogiri::XML::Node.new "locality", bib
-      locality[:type] = "anchor"
-      locality.add_child ref_from
-      ins << locality
-    end
-
     # compile and output individual file in collection
     def file_compile(f, filename, identifier)
       # warn "metanorma compile -x html #{f.path}"
       c = Compile.new
-      c.compile f.path, { format: :asciidoc, extension_keys: @format }.merge(@compile_options)
+      c.compile f.path, { format: :asciidoc,
+                          extension_keys: @format }.merge(@compile_options)
       @files[identifier][:outputs] = {}
       @format.each do |e|
         ext = c.processor.output_formats[e]
@@ -240,38 +93,11 @@ module Metanorma
       end
     end
 
-    # gather internal bibitem references
-    def gather_internal_refs
-      @files.each_with_object({}) do |(identifier, x), refs|
-        file, _ = targetfile(x, 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
-      end
-    end
-
-    # resolve file location for the target of each internal reference
-    def locate_internal_refs
-      refs = gather_internal_refs
-      @files.each do |identifier, x|
-        file, filename = targetfile(x, true)
-        docxml = Nokogiri::XML(file)
-        refs.each do |schema, ids|
-          ids.keys.each do |id|
-            n = docxml.at("//*[@id = '#{id}']") and n.at("./ancestor-or-self::*[@type = '#{schema}']") and
-              refs[schema][id] = identifier
-          end
-        end
-      end
-      refs.each do |schema, ids|
-        ids.each do |id, key|
-          key == true and refs[schema][id] = "Missing:#{schema}:#{id}"
-        end
-      end
-      refs
+    def copy_file_to_dest(fileref)
+      _file, filename = targetfile(fileref, true, false)
+      dest = File.join(@outdir, fileref[:rel_path])
+      FileUtils.mkdir_p(File.dirname(dest))
+      FileUtils.cp filename, dest
     end
 
     # process each file in the collection
@@ -279,12 +105,15 @@ module Metanorma
     def files # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
       internal_refs = locate_internal_refs
       @files.each do |identifier, x|
-        file, filename = targetfile(x, true)
-        file = update_xrefs(file, identifier, internal_refs)
-        Tempfile.open(["collection", ".xml"], encoding: "utf-8") do |f|
-          f.write(file)
-          f.close
-          file_compile(f, filename, identifier)
+        if x[:attachment] then copy_file_to_dest(x)
+        else
+          file, filename = targetfile(x, true)
+          file = update_xrefs(file, identifier, internal_refs)
+          Tempfile.open(["collection", ".xml"], encoding: "utf-8") do |f|
+            f.write(file)
+            f.close
+            file_compile(f, filename, identifier)
+          end
         end
       end
     end