relaton-itu 1.19.1 → 1.19.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: fdf374d5a2f89b0e69329f74e3399be48aaac4d972cd4ff96126c9834ff43437
-  data.tar.gz: 493fb3ccd5540490290d6c6c480d02526552a69320cff6c1d4842a5a9d4c335c
+  metadata.gz: 6eb58814c2b6eb0575c22b0e73606a7e54b0011f412e7d797c8be92d150a31f9
+  data.tar.gz: 2cf52ac96e85e7f916c0c7efc452c81514483c52553e02171cc562b1c0557027
 SHA512:
-  metadata.gz: 7278fb014a216fd7755e88348c765905c3487d4e8fffd2a756b9957a1454af5cef73041db9e804e8764bc426eab7203928f1a161a10bb6468eb4746e10cd65e4
-  data.tar.gz: 5582e31fdf889dfdf651308837db2bebe5c7689f459da21cdefc1219c66f178f8e045488ebe18c496b7449a8af5c14a5ac0956054b73b0fd7c0f99366ad0e77d
+  metadata.gz: 8e869a3131ab07a3601eedb46e5d62a6a4ac92e638c1c5e362003193d24b699bf829ffa2eade9705a01a5073f63e22f311686500c8a21fde958f4c309b72dc35
+  data.tar.gz: 57531fcb780e7ec0950628efae847101c4daef24d68809e53ff3d7934818fe21ab794f027bf2dcc0ad73b50137693274ec960a8cc5368e03f6fe523f7fa436ad

data/lib/relaton_itu/hit_collection.rb

@@ -82,7 +82,7 @@ module RelatonItu
       {
         "Input" => input.to_s,
         "Start" => 0,
-        "Rows" => 10,
+        "Rows" => 20,
         "SortBy" => "RELEVANCE",
         "ExactPhrase" => false,
         "CollectionName" => "General",

data/lib/relaton_itu/pubid.rb

@@ -18,11 +18,15 @@ module RelatonItu
       rule(:month1) { num.repeat(2, 2).as(:month) }
       rule(:date1) { str(" (") >> (month1 >> str("/")).maybe >> year >> str(")") }
       rule(:month2) { match["IVX"].repeat(1, 3).as(:month) }
-      rule(:date2) { str(" - ") >> num.repeat(2, 2) >> dot >> month2 >> dot >> year }
+      rule(:date2) { str(" - ") >> num.repeat(2, 2).as(:day) >> dot >> month2 >> dot >> year }
       rule(:date) { date1 | date2 }
       rule(:date?) { date.maybe }
 
-      rule(:amd) { space >> (str("Amd") | str("Amendment")) >> dot? >> space >> num.repeat(1, 2).as(:amd) }
+      rule(:amd_month) { num.repeat(2, 2) }
+      rule(:amd_year) { num.repeat(4, 4) }
+      rule(:amd_date) { str(" (") >> (amd_month >> str("/") >> amd_year).as(:amd_date) >> str(")") }
+      rule(:amd_date?) { amd_date.maybe }
+      rule(:amd) { space >> (str("Amd") | str("Amendment")) >> dot? >> space >> num.repeat(1, 2).as(:amd) >> amd_date? }
       rule(:amd?) { amd.maybe }
 
       rule(:sup) { space >> str("Suppl") >> dot? >> space >> num.repeat(1, 2).as(:suppl) }
@@ -31,11 +35,14 @@ module RelatonItu
       rule(:annex) { space >> str("Annex") >> space >> match["[:alnum:]"].repeat(1, 2).as(:annex) }
       rule(:annex?) { annex.maybe }
 
-      rule(:itu_pubid) { prefix >> sector >> type? >> code >> sup? >> annex? >> date? >> amd? >> any.repeat }
+      rule(:ver) { space >> str("(V") >> num.repeat(1, 2).as(:version) >> str(")") }
+      rule(:ver?) { ver.maybe }
+
+      rule(:itu_pubid) { prefix >> sector >> type? >> code >> sup? >> annex? >> ver? >> date? >> amd? >> any.repeat }
       root(:itu_pubid)
     end
 
-    attr_accessor :prefix, :sector, :type, :code, :suppl, :annex, :year, :month, :amd
+    attr_accessor :prefix, :sector, :type, :code, :suppl, :annex, :version, :year, :month, :day, :amd, :amd_date
 
     #
     # Create a new ITU publication identifier.
@@ -45,20 +52,26 @@ module RelatonItu
     # @param [String, nil] type
     # @param [String] code
     # @param [String, nil] suppl number
+    # @param [String, nil] version
     # @param [String, nil] year
     # @param [String, nil] month
+    # @param [String, nil] day
     # @param [String, nil] amd amendment number
+    # @param [String, nil] amd_date amendment
     #
     def initialize(prefix:, sector:, code:, **args)
       @prefix = prefix
       @sector = sector
       @type = args[:type]
+      @day = args[:day]
       @code, year, month = date_from_code code
       @suppl = args[:suppl]
       @annex = args[:annex]
+      @version = args[:version]
       @year = args[:year] || year
       @month = roman_to_2digit args[:month] || month
       @amd = args[:amd]
+      @amd_date = args[:amd_date]
     end
 
     def self.parse(id)
@@ -75,9 +88,12 @@ module RelatonItu
       hash[:type] = type if type && with_type
       hash[:suppl] = suppl if suppl
       hash[:annex] = annex if annex
+      hash[:version] = version if version
       hash[:year] = year if year
       hash[:month] = month if month
+      hash[:day] = day if day
       hash[:amd] = amd if amd
+      hash[:amd_date] = amd_date if amd_date
       hash
     end
 
@@ -91,18 +107,26 @@ module RelatonItu
       s << " #{code}"
       s << " Suppl. #{suppl}" if suppl
       s << " Annex #{annex}" if annex
+      s << " (V#{version})" if version
       s << date_to_s
       s << " Amd #{amd}" if amd
+      s << " (#{amd_date})" if amd_date
       s
     end
 
     def ===(other, ignore_args = [])
       hash = to_h with_type: false
       other_hash = other.to_h with_type: false
+      hash.delete(:version) if ignore_args.include?(:version)
+      other_hash.delete(:version) unless hash[:version]
+      hash.delete(:day)
+      other_hash.delete(:day)
       hash.delete(:month)
       other_hash.delete(:month)
       hash.delete(:year) if ignore_args.include?(:year)
       other_hash.delete(:year) unless hash[:year]
+      hash.delete(:amd_date) if ignore_args.include?(:amd_date)
+      other_hash.delete(:amd_date) unless hash[:amd_date]
       hash == other_hash
     end
 
@@ -133,7 +157,22 @@ module RelatonItu
       end.to_s.rjust(2, "0")
     end
 
+    def month_to_roman
+      int = month.to_i
+      return month unless int.between? 1, 12
+
+      roman_tens = ["", "X"]
+      roman_units = ["", "I", "II", "III", "IV", "V", "VI", "VII", "VIII", "IX"]
+
+      tens = int / 10
+      units = int % 10
+
+      roman_tens[tens] + roman_units[units]
+    end
+
     def date_to_s
+      # if code.match?(/^OB\./) && day && month
+      #   " - #{day}.#{month_to_roman}.#{year}"
       if month && year then " (#{month}/#{year})"
       elsif year then " (#{year})"
       else ""

data/lib/relaton_itu/scrapper.rb

@@ -39,12 +39,14 @@ module RelatonItu
         end
 
         # Fetch edition.
-        edition = doc.at("//table/tr/td/span[contains(@id, 'Label8')]/b")&.text
+        edition = doc.at("//table/tr/td[contains(@style,'color: white')]/span[contains(@id, 'Label8')]/b")&.text
+        docid = fetch_docid(doc, hit)
 
         ItuBibliographicItem.new(
+          id: fetch_id(docid),
           fetched: Date.today.to_s,
           type: "standard",
-          docid: fetch_docid(doc, hit.hit[:title]),
+          docid: docid,
           edition: edition,
           language: ["en"],
           script: ["Latn"],
@@ -65,20 +67,24 @@ module RelatonItu
 
       private
 
+      def fetch_id(docid)
+        docid.find(&:primary).id.gsub(/[.\s()\/-]/, "")
+      end
+
       # Fetch abstracts.
       # @param doc [Mechanize::Page]
       # @param hit [RelatonItu::Hit]
       # @return [Array<Hash>]
       def fetch_abstract(doc, hit) # rubocop:disable Metrics/AbcSize,Metrics/MethodLength
-        abstract_url = doc.at '//table/tr[td/span[.="In force"]]/td/span[contains(@id, "lbl_dms")]/div'
-        content = if abstract_url
-                    url = abstract_url[:onclick].match(/https?[^']+/).to_s
-                    rsp = hit.hit_collection.agent.get url
-                    d = Nokogiri::HTML rsp.body.encode(undef: :replace, replace: "")
-                    d.css("p.MsoNormal").text.gsub("\r\n", "").squeeze(" ").gsub("\u00a0", "")
-                  elsif a = doc.at('//table/tr/td/span[contains(@class, "observation")]/text()')
-                    a.text.strip
-                  end
+        abstract_url = doc.at '//table/tr/td[contains(@style,"color: white")]/span[contains(@id, "lbl_dms")]/div'
+        if abstract_url
+          url = abstract_url[:onclick].match(/https?[^']+/).to_s
+          rsp = hit.hit_collection.agent.get url
+          d = Nokogiri::HTML rsp.body.encode(undef: :replace, replace: "")
+          d.css("p.MsoNormal").text.gsub("\r\n", "").squeeze(" ").gsub("\u00a0", "")
+        elsif a = doc.at('//table/tr/td/span[contains(@class, "observation")]/text()')
+          a.text.strip
+        end => content
         return [] unless content
 
         [{
@@ -106,33 +112,48 @@ module RelatonItu
 
       # Fetch docid.
       # @param doc [Mechanize::Page]
-      # @param title [String]
+      # @param hit [RelatonItu::Hit]
       # @return [Hash]
-      def fetch_docid(doc, title)
-        docids = doc.xpath(
+      def fetch_docid(doc, hit)
+        docids = hit.hit[:code].to_s.split(" | ").map { |c| createdocid(c) }
+        docids += parse_id(doc).map { |c| createdocid c.text } if docids.empty?
+        docids << createdocid(title) unless docids.any?
+        docids
+      end
+
+      def parse_id(doc)
+        doc.xpath(
           "//span[@id='ctl00_content_main_uc_rec_main_info1_rpt_main_ctl00_lbl_rec']",
           "//td[.='Identical standard:']/following-sibling::td",
           "//div/table[1]/tr[4]/td/strong",
-        ).map { |c| createdocid c.text }
-        docids << createdocid(title) unless docids.any?
-        docids
+        )
       end
 
       # @param text [String]
       # @return [RelatonBib::DocumentIdentifier]
       def createdocid(text) # rubocop:disable Metrics/MethodLength
-        %r{
-          ^(?<code>(?:(?:ITU-\w|ISO/IEC)\s)?[^(:]+)
-          (?:\((?:(?<_month>\d{2})/)?(?<_year>\d{4})\))?
-          (?::[^(]+\((?<buldate>\d{2}\.\w{1,4}\.\d{4})\))?
-          (?:\s(?<corr>(?:Amd|Cor)\.\s?\d+))?
-          # (\s\(((?<_cormonth>\d{2})\/)?(?<_coryear>\d{4})\))?
-        }x =~ text.squeeze(" ")
-        corr&.sub!(/\.\s?/, " ")
-        id = [code.sub(/[[:space:]]$/, ""), corr].compact.join " "
-        id += " - #{buldate}" if buldate
-        type = id.match(%r{^\w+}).to_s
-        type = "ITU" if type == "G"
+        # %r{
+        #   ^(?<code>(?:(?:ITU-\w|ISO/IEC)\s)?[^(:]*)
+        #   (?:\s\(V(?<version>\d+)\))?
+        #   (?:\s\((?:(?<_month>\d{2})/)?(?<_year>\d{4})\))?
+        #   (?::[^(]+\((?<buldate>\d{2}\.\w{1,4}\.\d{4})\))?
+        #   (?:\s(?<corr>(?:Amd|Cor)\.\s?\d+))?
+        #   # (\s\(((?<_cormonth>\d{2})\/)?(?<_coryear>\d{4})\))?
+        # }x =~ text.squeeze(" ")
+        # corr&.sub!(/\.\s?/, " ")
+        # id = [code.sub(/[[:space:]]$/, ""), corr].compact.join " "
+        # id += " (V#{version})" if version
+        # id += " - #{buldate}" if buldate
+        # type = id.match(%r{^\w+}).to_s
+        # type = "ITU" if type == "G"
+        if text.match?(/^(?:ISO|ETSI)/)
+          type = "ISO"
+          text.match(/[^(]+/).to_s.strip.squeeze(" ")
+        else
+          pubid = Pubid.parse(text)
+          type = pubid.prefix # == "G" ? "ITU" : pubid.prefix
+          pubid.to_s
+        end => id
         RelatonBib::DocumentIdentifier.new(type: type, id: id, primary: true)
       end
 

data/lib/relaton_itu/version.rb

@@ -1,3 +1,3 @@
 module RelatonItu
-  VERSION = "1.19.1".freeze
+  VERSION = "1.19.3".freeze
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: relaton-itu
 version: !ruby/object:Gem::Version
-  version: 1.19.1
+  version: 1.19.3
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2024-07-31 00:00:00.000000000 Z
+date: 2024-10-01 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: mechanize
@@ -109,7 +109,7 @@ files:
 - lib/relaton_itu/util.rb
 - lib/relaton_itu/version.rb
 - lib/relaton_itu/xml_parser.rb
-- relaton-itu.gemspec
+- relaton_itu.gemspec
 homepage: https://github.com/metanorma/relaton-itu
 licenses:
 - MIT