relaton-nist 2.0.0.pre.alpha.2 → 2.0.0.pre.alpha.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 0e1709e47734509d547b5a112f57b77ff68df5dec8b3f4c62350346faa5c411c
-  data.tar.gz: 90d49b0829c2eaddca800b28fb2f82220d03b1c55935c2f80033511dc8d31dd9
+  metadata.gz: 8f241e09f50957db23bc91820f9bba18b14ed8fdf237646e0fdaa51f8ebc8f46
+  data.tar.gz: da8e0ac7ffc24a64f02624467c479369129125f6fdda3fe6da950e05fe2fa9bc
 SHA512:
-  metadata.gz: 9c0746c5faad3ff3742a92ec4cfad1f24f161f6908a3ae732653fe97384b32110016b95accbcb3af66aba7960f577bfec4fb8c0f4e63c6522f546fb0caec336a
-  data.tar.gz: 2c638a164a26ccd2e3aba38dab34ee2aa2f178e7427dbc2eb8d755077f16a7e0bc95787da9a2aaf063c48d2c2636aed099dd99d1dff4e295d0242218a2443767
+  metadata.gz: fd8257cfae0044309acc0b1818e903be60595e08722c961853f2f663f98395a18202749a89a96cbf587b9ed2b4cdf2a2d33a683fd65944dd5eefef06798e4fe3
+  data.tar.gz: 2087de5a12a59076d95eb2cc03f920d2dd9c39512ec8cb9e5392f7f0d60b19e7fb59949a4e2e3cca62d4d6e6abb22281098d32c83068646fb08c26b6b7da5c34

data/grammars/basicdoc.rng

@@ -187,6 +187,15 @@ Applicable to modify and delete</a:documentation>
         <a:documentation>Optional caption of this block</a:documentation>
       </attribute>
     </optional>
+    <optional>
+      <attribute name="position">
+        <a:documentation>For an "add" change, whether the change is added before or after the location</a:documentation>
+        <choice>
+          <value>before</value>
+          <value>after</value>
+        </choice>
+      </attribute>
+    </optional>
     <optional>
       <element name="location">
         <a:documentation>The location(s) in the original document which have undergone the change described in this block</a:documentation>
@@ -208,11 +217,15 @@ Applicable to modify and delete</a:documentation>
     </zeroOrMore>
     <optional>
       <element name="newcontent">
-        <a:documentation>New content to be added to the document; applicable to add and modify</a:documentation>
+        <a:documentation>New content to be added to the document; applicable to add and modify.
+Can be blocks and/or sections</a:documentation>
         <ref name="OptionalId"/>
         <zeroOrMore>
           <ref name="BasicBlock"/>
         </zeroOrMore>
+        <zeroOrMore>
+          <ref name="section"/>
+        </zeroOrMore>
       </element>
     </optional>
     <zeroOrMore>

data/grammars/biblio.rng

@@ -1142,11 +1142,11 @@ NOTE: This should preferably be encoded as  a URI or short identifier, rather th
         <a:documentation>Information about how long the current description of the bibliographic item is valid for</a:documentation>
       </ref>
     </optional>
-    <optional>
+    <zeroOrMore>
       <ref name="depiction">
         <a:documentation>Depiction of the bibliographic item, typically an image</a:documentation>
       </ref>
-    </optional>
+    </zeroOrMore>
   </define>
   <define name="ReducedBibliographicItem">
     <a:documentation>Reduced description of a bibliographic resource, without mandatory title and docidentifier, used for document relations
@@ -1939,10 +1939,10 @@ Detailed in https://www.relaton.org/model/relations/</a:documentation>
       <value>hasAnnotation</value>
       <value>draftOf</value>
       <value>hasDraft</value>
-      <value>preliminaryDraftOf</value>
-      <value>hasPreliminaryDraft</value>
-      <value>revisionDraftOf</value>
-      <value>hasRevisionDraft</value>
+      <value>predecessorDraftOf</value>
+      <value>hasPredecessorDraft</value>
+      <value>successorDraftOf</value>
+      <value>hasSuccessorDraft</value>
       <value>editionOf</value>
       <value>hasEdition</value>
       <value>updates</value>
@@ -2063,13 +2063,13 @@ provided that it is not the entire bibliographic item that is so related</a:docu
           <ref name="LocalizedString"/>
         </element>
       </optional>
-      <oneOrMore>
+      <zeroOrMore>
         <element name="taxon">
           <a:documentation>The keywords as a hierarchical taxonomy. For example, the sequence of `taxon` elements
 `pump`, `centrifugal pump`, `line shaft pump` represents a taxonomic classification</a:documentation>
           <ref name="LocalizedString"/>
         </element>
-      </oneOrMore>
+      </zeroOrMore>
       <zeroOrMore>
         <ref name="vocabid">
           <a:documentation>Identifiers for the keyword as a controlled vocabulary</a:documentation>

data/lib/relaton/nist/data_fetcher.rb

@@ -16,12 +16,13 @@ module Relaton
         fetch_tech_pubs
         # add_static_files
         index.save
+        report_errors
       end
 
       def fetch_tech_pubs
         xml_data = Mechanize.new.get(URL).body
         docs = LocMods::Collection.from_xml xml_data
-        docs.mods.each { |doc| write_file ModsParser.new(doc, series).parse }
+        docs.mods.each { |doc| write_file ModsParser.new(doc, series, @errors).parse }
       end
 
       def write_file(bib)
@@ -54,10 +55,6 @@ module Relaton
         bib.to_rfcxml
       end
 
-      def gh_issue_channel
-        ["relaton/relaton-data-nist", "Error fetching NIST documents"]
-      end
-
       def log_error(msg)
         Util.error msg
       end

data/lib/relaton/nist/mods_parser.rb

@@ -15,9 +15,10 @@ module Relaton
       ATTRS = %i[type docidentifier title source abstract date contributor
                   relation place series].freeze
 
-      def initialize(doc, series)
+      def initialize(doc, series, errors = {})
         @doc = doc
         @series = series
+        @errors = errors
       end
 
       # @return [Bib::ItemData]
@@ -29,24 +30,24 @@ module Relaton
         ItemData.new(**args)
       end
 
-      def parse_type
-        "standard"
-      end
+      def parse_type = "standard"
 
       # @return [Array<Bib::Docidentifier>]
       def parse_docidentifier
-        [
+        ids = [
           { type: "NIST", content: pub_id, primary: true },
           { type: "DOI", content: parse_doi },
-        ].map { |id| Bib::Docidentifier.new(**id) }
+        ].reject { |id| id[:content].nil? || id[:content].empty? }
+        @errors[:docidentifier] &&= ids.empty?
+        ids.map { |id| Bib::Docidentifier.new(**id) }
       end
 
       # @return [String]
-      def pub_id
-        get_id_from_str parse_doi
-      end
+      def pub_id = get_id_from_str parse_doi
 
       def get_id_from_str(str)
+        return if str.nil? || str.empty?
+
         ::Pubid::Nist::Identifier.parse(str).to_s
       rescue ::Pubid::Core::Errors::ParseError
         str.gsub(".", " ").sub(/^[\D]+/, &:upcase)
@@ -66,13 +67,15 @@ module Relaton
 
       def parse_doi
         url = @doc.location.reduce(nil) { |m, l| m || l.url.detect { |u| u.usage == "primary display" } }
+        return if url.nil?
+
         id = remove_doi_prefix(url.content)
+        return if id.nil?
+
         replace_wrong_doi(id)
       end
 
-      def remove_doi_prefix(id)
-        id.match(/10\.6028\/(.+)/)[1]
-      end
+      def remove_doi_prefix(id) = id.match(/10\.6028\/(.+)/)&.send(:[], 1)
 
       # @return [Array<Bib::Title>]
       def parse_title
@@ -90,6 +93,7 @@ module Relaton
         elsif title.size == 1
           title[0].instance_variable_set :@type, "main"
         end
+        @errors[:title] &&= title.empty?
         title
       end
 
@@ -100,26 +104,31 @@ module Relaton
       end
 
       def parse_source
-        @doc.location.map do |location|
+        source = @doc.location.map do |location|
           url = location.url.first
           type = url.usage == "primary display" ? "doi" : "src"
           Bib::Uri.new content: url.content, type: type
         end
+        @errors[:source] &&= source.empty?
+        source
       end
 
       def parse_abstract
-        Array(@doc.abstract).map do |a|
+        abstract = Array(@doc.abstract).map do |a|
           content = a.content.gsub("\n", " ").squeeze(" ").strip
-          Bib::LocalizedMarkedUpString.new content: content, language: "en",
+          Bib::Abstract.new content: content, language: "en",
                                            script: "Latn"
         end
+        @errors[:abstract] &&= abstract.empty?
+        abstract
       end
 
       def parse_date
         date = @doc.origin_info[0].date_issued.map do |di|
           create_date(di, "issued")
-        end
-        date.compact
+        end.compact
+        @errors[:date] &&= date.empty?
+        date
       end
 
       def create_date(date, type)
@@ -136,13 +145,11 @@ module Relaton
         end
       end
 
-      def parse_doctype
-        Doctype.new(content: "standard")
-      end
+      def parse_doctype = Doctype.new(content: "standard")
 
       def parse_contributor
         # exclude primary contributors to avoid duplication
-        @doc.name.reject { |n| n.usage == "primary" }.map do |name|
+        contributor = @doc.name.reject { |n| n.usage == "primary" }.map do |name|
           entity, default_role = create_entity(name)
           next unless entity
 
@@ -152,6 +159,8 @@ module Relaton
           role << Bib::Contributor::Role.new(type: default_role) if role.empty?
           create_contributor(entity, role)
         end.compact
+        @errors[:contributor] &&= contributor.empty?
+        contributor
       end
 
       def create_contributor(entity, role)
@@ -200,16 +209,21 @@ module Relaton
       end
 
       def parse_relation
-        Array(@doc.related_item).reject { |ri| ri.type == "series" }.map do |ri|
+        relations = Array(@doc.related_item).reject { |ri| ri.type == "series" }.filter_map do |ri|
           type = RELATION_TYPES[ri.type]
-          Relation.new(type: type, bibitem: create_related_item(ri))
+          bibitem = create_related_item(ri)
+          Relation.new(type: type, bibitem: bibitem) if bibitem
         end
+        @errors[:relation] &&= relations.empty?
+        relations
       end
 
       def create_related_item(item)
         item_id = get_id_from_str related_item_id(item)
+        return if item_id.nil? || item_id.empty?
+
         docid = Bib::Docidentifier.new(type: "NIST", content: item_id)
-        fref = Bib::LocalizedMarkedUpString.new(content: item_id)
+        fref = Bib::Formattedref.new(content: item_id)
         ItemData.new(docidentifier: [docid], formattedref: fref)
       end
 
@@ -219,15 +233,20 @@ module Relaton
         else
           item.name[0].name_part[0].content
         end => id
-        replace_wrong_doi remove_doi_prefix(id)
+        doi = remove_doi_prefix(id)
+        return if doi.nil?
+
+        replace_wrong_doi(doi)
       end
 
       def parse_place
-        @doc.origin_info.select { |p| p.event_type == "publisher" }.map do |p|
-          place = p.place[0].place_term[0].content
-          /(?<city>\w+), (?<state>\w+)/ =~ place
+        place = @doc.origin_info.select { |p| p.event_type == "publisher" }.map do |p|
+          pl = p.place[0].place_term[0].content
+          /(?<city>\w+), (?<state>\w+)/ =~ pl
           Bib::Place.new(city: city, region: create_region(state))
         end
+        @errors[:place] &&= place.empty?
+        place
       end
 
       def create_region(state)
@@ -237,12 +256,14 @@ module Relaton
       end
 
       def parse_series
-        Array(@doc.related_item).select { |ri| ri.type == "series" }.map do |ri|
+        series = Array(@doc.related_item).select { |ri| ri.type == "series" }.map do |ri|
           tinfo = ri.title_info[0]
           tcontent = tinfo.title[0].strip
           title = Bib::Title.new(content: tcontent)
           Bib::Series.new(title: [title], number: tinfo.part_number&.first)
         end
+        @errors[:series] &&= series.empty?
+        series
       end
     end
   end

data/lib/relaton/nist/scraper.rb

@@ -264,7 +264,7 @@ module Relaton
           end
           ids = [Bib::Docidentifier.new(content: ref, type: "NIST", primary: true)]
           link = [Bib::Uri.new(type: "src", content: uri)]
-          bib = ItemData.new(formattedref: ref, source: link, docidentifier: ids)
+          bib = ItemData.new(formattedref: Bib::Formattedref.new(content: ref), source: link, docidentifier: ids)
           Relation.new(type: t, description: descr, bibitem: bib)
         end
 
@@ -273,7 +273,7 @@ module Relaton
         def fetch_keywords(doc)
           doc["keywords"].map do |kw|
             text = kw.is_a?(String) ? kw : kw.text
-            Bib::Keyword.new(taxon: [Bib::LocalizedString.new(content: text)])
+            Bib::Keyword.new(vocab: Bib::LocalizedString.new(content: text))
           end
         end
 

data/lib/relaton/nist/tech_pubs_parser.rb

@@ -113,7 +113,7 @@ module Relaton
         ).each_with_object([]) do |a, m|
           next if a.text.empty?
 
-          m << Bib::LocalizedMarkedUpString.new(
+          m << Bib::Abstract.new(
             content: a.text, language: @doc["language"], script: "Latn",
           )
         end
@@ -267,7 +267,7 @@ module Relaton
         @doc.xpath("./ns:program/ns:related_item", ns: NS).map do |rel|
           rdoi = rel.at_xpath("ns:intra_work_relation|ns:inter_work_relation", ns: NS)
           id = rdoi.text.split("/")[1..].join("/").gsub(".", " ")
-          fref = Bib::LocalizedMarkedUpString.new(content: id)
+          fref = Bib::Formattedref.new(content: id)
           docid = Bib::Docidentifier.new(type: "NIST", content: id, primary: true)
           bibitem = ItemData.new(formattedref: fref, docidentifier: [docid])
           type = RELATION_TYPES[rdoi["relationship-type"]]

data/lib/relaton/nist/version.rb

@@ -1,5 +1,5 @@
 module Relaton
   module Nist
-    VERSION = "2.0.0-alpha.2".freeze
+    VERSION = "2.0.0-alpha.3".freeze
   end
 end

data/relaton_nist.gemspec

@@ -27,8 +27,8 @@ Gem::Specification.new do |spec|
   spec.add_dependency "mechanize", "~> 2.0"
   spec.add_dependency "loc_mods", "~> 0.2.0"
   spec.add_dependency "pubid", "~> 1.15.6"
-  spec.add_dependency "relaton-bib", "~> 2.0.0-alpha.4"
-  spec.add_dependency "relaton-core", "~> 0.0.9"
+  spec.add_dependency "relaton-bib", "~> 2.0.0-alpha.7"
+  spec.add_dependency "relaton-core", "~> 0.0.13"
   spec.add_dependency "relaton-index", "~> 0.2.0"
   spec.add_dependency "rubyzip"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: relaton-nist
 version: !ruby/object:Gem::Version
-  version: 2.0.0.pre.alpha.2
+  version: 2.0.0.pre.alpha.3
 platform: ruby
 authors:
 - Ribose Inc.
@@ -71,28 +71,28 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 2.0.0.pre.alpha.4
+        version: 2.0.0.pre.alpha.7
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 2.0.0.pre.alpha.4
+        version: 2.0.0.pre.alpha.7
 - !ruby/object:Gem::Dependency
   name: relaton-core
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.0.9
+        version: 0.0.13
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.0.9
+        version: 0.0.13
 - !ruby/object:Gem::Dependency
   name: relaton-index
   requirement: !ruby/object:Gem::Requirement