relaton-ietf 1.7.1 → 1.8.0

data/grammars/reqt.rng

@@ -30,15 +30,34 @@
         <data type="boolean"/>
       </attribute>
     </optional>
+    <optional>
+      <attribute name="number"/>
+    </optional>
     <optional>
       <attribute name="subsequence"/>
     </optional>
+    <optional>
+      <attribute name="keep-with-next">
+        <data type="boolean"/>
+      </attribute>
+    </optional>
+    <optional>
+      <attribute name="keep-lines-together">
+        <data type="boolean"/>
+      </attribute>
+    </optional>
     <attribute name="id">
       <data type="ID"/>
     </attribute>
     <optional>
       <attribute name="filename"/>
     </optional>
+    <optional>
+      <attribute name="model"/>
+    </optional>
+    <optional>
+      <attribute name="type"/>
+    </optional>
     <optional>
       <ref name="reqtitle"/>
     </optional>
@@ -48,9 +67,9 @@
     <optional>
       <ref name="subject"/>
     </optional>
-    <optional>
+    <zeroOrMore>
       <ref name="reqinherit"/>
-    </optional>
+    </zeroOrMore>
     <zeroOrMore>
       <ref name="classification"/>
     </zeroOrMore>
@@ -135,6 +154,16 @@
         <data type="boolean"/>
       </attribute>
     </optional>
+    <optional>
+      <attribute name="keep-with-next">
+        <data type="boolean"/>
+      </attribute>
+    </optional>
+    <optional>
+      <attribute name="keep-lines-together">
+        <data type="boolean"/>
+      </attribute>
+    </optional>
     <oneOrMore>
       <ref name="BasicBlock"/>
     </oneOrMore>

data/lib/relaton_ietf/hash_converter.rb

@@ -5,9 +5,9 @@ module RelatonIetf
       # Ovverides superclass's method
       #
       # @param item [Hash]
-      # @retirn [RelatonIec::IecBibliographicItem]
+      # @retirn [RelatonIetf::IetfBibliographicItem]
       def bib_item(item)
-        IecBibliographicItem.new(item)
+        IetfBibliographicItem.new(**item)
       end
     end
   end

data/lib/relaton_ietf/ietf_bibliographic_item.rb

@@ -17,6 +17,13 @@ module RelatonIetf
       super
     end
 
+    # @param hash [Hash]
+    # @return [RelatonIetf::IetfBibliographicItem]
+    def self.from_hash(hash)
+      item_hash = ::RelatonIetf::HashConverter.hash_to_bib(hash)
+      new **item_hash
+    end
+
     # @param opts [Hash]
     # @option opts [Nokogiri::XML::Builder] :builder XML builder
     # @option opts [Boolean] :bibdata

data/lib/relaton_ietf/ietf_bibliography.rb

@@ -20,7 +20,11 @@ module RelatonIetf
       def get(code, _year = nil, _opts = {})
         warn "[relaton-ietf] (\"#{code}\") fetching..."
         result = search code
-        warn "[relaton-ietf] (\"#{code}\") found #{result.docidentifier.first.id}"
+        if result
+          warn "[relaton-ietf] (\"#{code}\") found #{result.docidentifier.first.id}"
+        else
+          warn "[relaton-ietf] (\"#{code}\") not found"
+        end
         result
       end
     end

data/lib/relaton_ietf/processor.rb

@@ -27,8 +27,7 @@ module RelatonIetf
     # @param hash [Hash]
     # @return [RelatonIetf::IetfBibliographicItem]
     def hash_to_bib(hash)
-      item_hash = ::RelatonIetf::HashConverter.hash_to_bib(hash)
-      ::RelatonIetf::IetfBibliographicItem.new item_hash
+      ::RelatonIetf::IetfBibliographicItem.from_hash hash
     end
 
     # Returns hash of XML grammar

data/lib/relaton_ietf/scrapper.rb

@@ -1,7 +1,6 @@
 # frozen_string_literal: true
 
 require "net/http"
-require "nokogiri"
 require "relaton_bib"
 require "relaton_ietf/ietf_bibliographic_item"
 
@@ -11,26 +10,22 @@ module RelatonIetf
   # Scrapper module
   module Scrapper
     GH_URL = "https://raw.githubusercontent.com/relaton/relaton-data-ietf/master/data/reference."
-    RFC_URI_PATTERN = "https://xml2rfc.tools.ietf.org/public/rfc/bibxml"
-    # ID_URI_PATTERN = "https://xml2rfc.tools.ietf.org/public/rfc/bibxml-ids/reference.CODE"
     BCP_URI_PATTERN = "https://www.rfc-editor.org/info/CODE"
 
     class << self
-      # rubocop:disable Metrics/MethodLength
-
       # @param text [String]
       # @param is_relation [TrueClass, FalseClass]
       # @return [RelatonIetf::IetfBibliographicItem]
-      def scrape_page(text, is_relation = false)
+      def scrape_page(text, is_relation: false)
         # Remove initial "IETF " string if specified
         ref = text.gsub(/^IETF /, "")
-        if ref.match? /^BCP/ then bcp_item BCP_URI_PATTERN.dup, ref
-        else rfc_item ref, is_relation
-        end
+        /^(?:RFC|BCP|FYI|STD)\s(?<num>\d+)/ =~ ref
+        ref.sub! /(?<=^(?:RFC|BCP|FYI|STD)\s)(\d+)/, num.rjust(4, "0") if num
+        rfc_item ref, is_relation
       rescue Timeout::Error, Errno::EINVAL, Errno::ECONNRESET, EOFError,
              Net::HTTPBadResponse, Net::HTTPHeaderSyntaxError,
              Net::ProtocolError, SocketError
-        raise RelatonBib::RequestError, "No document found for #{ref} reference."
+        raise RelatonBib::RequestError, "No document found for #{ref} reference"
       end
 
       # @param reference [Nokogiri::XML::Element, nil]
@@ -38,7 +33,7 @@ module RelatonIetf
       # @param url [String, NilClass]
       # @param ver [String, NilClass] Internet Draft version
       # @return [RelatonIetf::IetfBibliographicItem]
-      def fetch_rfc(reference, is_relation = false, url = nil, ver = nil) # rubocop:disable Metrics/AbcSize
+      def fetch_rfc(reference, is_relation: false, url: nil, ver: nil) # rubocop:disable Metrics/AbcSize,Metrics/MethodLength
         return unless reference
 
         ietf_item(
@@ -50,16 +45,17 @@ module RelatonIetf
           language: [language(reference)],
           link: link(reference, url, ver),
           title: titles(reference),
+          formattedref: formattedref(reference),
           abstract: abstracts(reference),
           contributor: contributors(reference),
+          relation: relations(reference),
           date: dates(reference),
           series: series(reference),
           place: ["Fremont, CA"],
           keyword: reference.xpath("front/keyword").map(&:text),
-          doctype: doctype(reference[:anchor])
+          doctype: doctype(reference[:anchor]),
         )
       end
-      # rubocop:enable Metrics/MethodLength
 
       private
 
@@ -95,6 +91,7 @@ module RelatonIetf
       end
 
       # @param ref [String]
+      # @param is_relation [Boolen, nil]
       # @return [RelatonIetf::IetfBibliographicItem]
       def rfc_item(ref, is_relation)
         /(?<=-)(?<ver>\d{2})$/ =~ ref
@@ -105,55 +102,58 @@ module RelatonIetf
 
         uri = "#{GH_URL}#{ref.sub(/\s|\u00a0/, '.')}.xml"
         doc = Nokogiri::XML get_page(uri)
-        fetch_rfc doc.at("//reference"), is_relation, uri, ver
-      end
-
-      # @param uri_template [String]
-      # @param reference [String]
-      # @return [RelatonIetf::IetfBibliographicItem]
-      def bcp_item(uri_template, reference) # rubocop:disable Metrics/MethodLength
-        uri = uri_template.sub "CODE", reference.sub(" ", "").downcase
-        doc = Nokogiri::HTML get_page(uri)
-        ietf_item(
-          id: reference,
-          title: [content: ""],
-          docid: [RelatonBib::DocumentIdentifier.new(type: "IETF", id: reference)],
-          language: ["en"],
-          link: [{ type: "src", content: uri }],
-          relation: fetch_relations(doc),
-          doctype: "rfc"
-        )
+        r = doc.at("/referencegroup", "/reference")
+        fetch_rfc r, is_relation: is_relation, url: uri, ver: ver
       end
 
-      def fetch_relations(doc)
-        doc.xpath("//table/tr/td/a[contains(., 'RFC')]").map do |r|
-          RelatonBib::DocumentRelation.new(
-            type: "merges",
-            bibitem: scrape_page(r.text, true)
-          )
+      # @param reference [Nokogiri::XML::Element]
+      # @return [Hash]
+      def relations(reference)
+        reference.xpath("reference").map do |ref|
+          { type: "includes", bibitem: fetch_rfc(ref, is_relation: true) }
         end
       end
 
+      # @param uri [String]
+      # @return [String] HTTP response body
       def get_page(uri)
         res = Net::HTTP.get_response(URI(uri))
-        if res.code != "200"
-          raise RelatonBib::RequestError, "No document found at #{uri}"
-        end
+        return unless res.code == "200"
+
+        #   raise RelatonBib::RequestError, "No document found at #{uri}"
+        # end
 
         res.body
       end
 
+      # @param reference [Nokogiri::XML::Element]
       # @return [String]
       def language(reference)
         reference[:lang] || "en"
       end
 
+      # @param reference [Nokogiri::XML::Element]
       # @return [Array<Hash>]
       def titles(reference)
-        title = reference.at("./front/title")
-        [{ content: title.text, language: language(reference), script: "Latn" }]
+        reference.xpath("./front/title").map do |title|
+          { content: title.text, language: language(reference), script: "Latn" }
+        end
       end
 
+      # @param reference [Nokogiri::XML::Element]
+      # @return [RelatonBib::FormattedRef, nil]
+      def formattedref(reference)
+        return if reference.at "./fornt/title"
+
+        cont = (reference[:anchor] || reference[:docName] || reference[:number])
+        if cont
+          RelatonBib::FormattedRef.new(
+            content: cont, language: language(reference), script: "Latn",
+          )
+        end
+      end
+
+      # @param reference [Nokogiri::XML::Element]
       # @return [Array<RelatonBib::FormattedString>]
       def abstracts(ref)
         ref.xpath("./front/abstract").map do |a|
@@ -164,11 +164,13 @@ module RelatonIetf
         end
       end
 
+      # @param reference [Nokogiri::XML::Element]
       # @return [Array<Hash>]
       def contributors(reference)
         persons(reference) + organizations(reference)
       end
 
+      # @param reference [Nokogiri::XML::Element]
       # @return [Array<Hash{Symbol=>RelatonBib::Person,Symbol=>Array<String>}>]
       def persons(reference)
         reference.xpath("./front/author[@surname]|./front/author[@fullname]")
@@ -176,13 +178,15 @@ module RelatonIetf
           entity = RelatonBib::Person.new(
             name: full_name(author, reference),
             affiliation: [affiliation(author)],
-            contact: contacts(author.at("./address"))
+            contact: contacts(author.at("./address")),
           )
           { entity: entity, role: [contributor_role(author)] }
         end
       end
 
-      # @return [Array<Hash{Symbol=>RelatonBib::Organization,Symbol=>Array<String>}>]
+      # @param reference [Nokogiri::XML::Element]
+      # @return [Array<Hash{Symbol=>RelatonBib::Organization,
+      #   Symbol=>Array<String>}>]
       def organizations(reference)
         publisher = { entity: new_org, role: [type: "publisher"] }
         orgs = reference.xpath("./seriesinfo").reduce([publisher]) do |mem, si|
@@ -191,21 +195,21 @@ module RelatonIetf
           mem << { entity: new_org(si[:stream], nil), role: [type: "author"] }
         end
         orgs + reference.xpath(
-          "front/author[not(@surname)][not(@fullname)]/organization"
+          "front/author[not(@surname)][not(@fullname)]/organization",
         ).map do |org|
           { entity: new_org(org.text, nil), role: [type: "author"] }
         end
       end
 
-      # @param author [Nokogiri::XML::Document]
-      # @param ref [Nokogiri::XML::Document]
+      # @param author [Nokogiri::XML::Element]
+      # @param ref [Nokogiri::XML::Element]
       # @return [RelatonBib::FullName]
       def full_name(author, ref)
         lang = language ref
         RelatonBib::FullName.new(
           completename: localized_string(author[:fullname], lang),
           initial: [localized_string(author[:initials], lang)].compact,
-          surname: localized_string(author[:surname], lang)
+          surname: localized_string(author[:surname], lang),
         )
       end
 
@@ -218,7 +222,7 @@ module RelatonIetf
         RelatonBib::LocalizedString.new(content, lang)
       end
 
-      # @param postal [Nokogiri::XML::Document]
+      # @param postal [Nokogiri::XML::Element]
       # @return [Array<RelatonBib::Address, RelatonBib::Phone>]
       def contacts(addr)
         contacts = []
@@ -232,7 +236,7 @@ module RelatonIetf
         contacts
       end
 
-      # @param postal [Nokogiri::XML::Document]
+      # @param postal [Nokogiri::XML::Element]
       # @rerurn [RelatonBib::Address]
       def address(postal) # rubocop:disable Metrics/CyclomaticComplexity
         RelatonBib::Address.new(
@@ -240,7 +244,7 @@ module RelatonIetf
           city: postal.at("./city")&.text,
           postcode: postal.at("./code")&.text,
           country: postal.at("./country")&.text,
-          state: postal.at("./region")&.text
+          state: postal.at("./region")&.text,
         )
       end
 
@@ -252,7 +256,7 @@ module RelatonIetf
         contacts << RelatonBib::Contact.new(type: type, value: value.text)
       end
 
-      # @param author [Nokogiri::XML::Document]
+      # @param author [Nokogiri::XML::Element]
       # @return [RelatonBib::Affiliation]
       def affiliation(author)
         organization = author.at("./organization")
@@ -287,6 +291,7 @@ module RelatonIetf
       #
       # Extract date from reference.
       #
+      # @param reference [Nokogiri::XML::Element]
       # @return [Array<RelatonBib::BibliographicDate>] published data.
       #
       def dates(reference)
@@ -298,8 +303,6 @@ module RelatonIetf
         [RelatonBib::BibliographicDate.new(type: "published", on: date)]
       end
 
-      # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
-
       #
       # Extract document identifiers from reference
       #
@@ -308,12 +311,12 @@ module RelatonIetf
       #
       # @return [Array<RelatonBib::DocumentIdentifier>]
       #
-      def docids(reference, ver) # rubocop:disable Metrics/MethodLength,Metrics/CyclomaticComplexity,Metrics/PerceivedComplexity
+      def docids(reference, ver) # rubocop:disable Metrics/MethodLength,Metrics/CyclomaticComplexity,Metrics/PerceivedComplexity,Metrics/AbcSize
         id = (reference[:anchor] || reference[:docName] || reference[:number])
         ret = []
         if id
           ret << RelatonBib::DocumentIdentifier.new(
-            type: "IETF", id: id.sub(/^(RFC)/, "\\1 ")
+            type: "IETF", id: id.sub(/^(RFC)/, "\\1 "),
           )
         end
         if (id = reference[:anchor])
@@ -327,11 +330,10 @@ module RelatonIetf
           RelatonBib::DocumentIdentifier.new(id: id, type: si[:name])
         end.compact
       end
-      # enable Metrics/MethodLength, Metrics/AbcSize
 
       #
       # Extract series form reference
-      # @param reference [Nokogiri::XML::Document]
+      # @param reference [Nokogiri::XML::Element]
       #
       # @return [Array<RelatonBib::Series>]
       #
@@ -341,17 +343,17 @@ module RelatonIetf
 
           RelatonBib::Series.new(
             title: RelatonBib::TypedTitleString.new(
-              content: si[:name], language: language(reference), script: "Latn"
+              content: si[:name], language: language(reference), script: "Latn",
             ),
             number: si[:value],
-            type: "main"
+            type: "main",
           )
         end.compact
       end
 
       #
       # extract status
-      # @param reference [Nokogiri::XML::Document]
+      # @param reference [Nokogiri::XML::Element]
       #
       # @return [RelatonBib::DocumentStatus]
       #

data/lib/relaton_ietf/version.rb

@@ -1,3 +1,3 @@
 module RelatonIetf
-  VERSION = "1.7.1".freeze
+  VERSION = "1.8.0".freeze
 end

data/lib/relaton_ietf/xml_parser.rb

@@ -9,7 +9,7 @@ module RelatonIetf
       # @param item_hash [Hash]
       # @return [RelatonIetf::IetfBibliographicItem]
       def bib_item(item_hash)
-        IetfBibliographicItem.new item_hash
+        IetfBibliographicItem.new **item_hash
       end
     end
   end