relaton-iso 1.18.1 → 1.18.3

data/lib/relaton_iso/iso_bibliography.rb

@@ -6,205 +6,196 @@ require "relaton_iso/hit_collection"
 # require "relaton_iec"
 
 module RelatonIso
-  # Class methods for search ISO standards.
-  class IsoBibliography
-    class << self
-      # @param text [String]
-      # @return [RelatonIso::HitCollection]
-      def search(text)
-        HitCollection.new(text.gsub("\u2013", "-")).fetch
-      rescue SocketError, Timeout::Error, Errno::EINVAL, Errno::ECONNRESET,
-             EOFError, Net::HTTPBadResponse, Net::HTTPHeaderSyntaxError,
-             Net::ProtocolError, OpenSSL::SSL::SSLError, Errno::ETIMEDOUT,
-             Algolia::AlgoliaUnreachableHostError => e
-        raise RelatonBib::RequestError, e.message
-      end
+  # Methods for search ISO standards.
+  module IsoBibliography
+    extend self
+
+    # @param text [Pubid::Iso::Identifier, String]
+    # @return [RelatonIso::HitCollection]
+    def search(pubid, opts = {})
+      pubid = Pubid::Iso::Identifier.parse(pubid) if pubid.is_a? String
+      HitCollection.new(pubid, opts).fetch
+    rescue  SocketError, Timeout::Error, Errno::EINVAL, Errno::ECONNRESET,
+            EOFError, Net::HTTPBadResponse, Net::HTTPHeaderSyntaxError,
+            Net::ProtocolError, OpenSSL::SSL::SSLError, Errno::ETIMEDOUT => e
+      raise RelatonBib::RequestError, e.message
+    end
 
-      # @param ref [String] the ISO standard Code to look up (e..g "ISO 9000")
-      # @param year [String, NilClass] the year the standard was published
-      # @param opts [Hash] options; restricted to :all_parts if all-parts
-      # @option opts [Boolean] :all_parts if all-parts reference is required
-      # @option opts [Boolean] :keep_year if undated reference should return
-      #   actual reference with year
-      #
-      # @return [RelatonIsoBib::IsoBibliographicItem] Relaton XML serialisation of reference
-      def get(ref, year = nil, opts = {}) # rubocop:disable Metrics/CyclomaticComplexity,Metrics/MethodLength,Metrics/PerceivedComplexity,Metrics/AbcSize
-        code = ref.gsub("\u2013", "-")
-
-        # parse "all parts" request
-        code.sub! " (all parts)", ""
-        opts[:all_parts] ||= $~ && opts[:all_parts].nil?
-
-        query_pubid = Pubid::Iso::Identifier.parse(code)
-        query_pubid.year = year if year
-        query_pubid.part = nil if opts[:all_parts]
-        Util.warn "(#{query_pubid}) Fetching from iso.org ..."
-
-        hits, missed_year_ids = isobib_search_filter(query_pubid, opts)
-        tip_ids = look_up_with_any_types_stages(hits, ref, opts)
-
-        ret = if !opts[:all_parts] || hits.size == 1
-                hits.any? && hits.first.fetch(opts[:lang])
-              else
-                hits.to_all_parts(opts[:lang])
-              end
-
-        return fetch_ref_err(query_pubid, missed_year_ids, tip_ids) unless ret
-
-        response_docid = ret.docidentifier.first.id.sub(" (all parts)", "")
-        response_pubid = Pubid::Iso::Identifier.parse(response_docid)
-
-        Util.warn "(#{query_pubid}) Found: `#{response_pubid}`"
-
-        get_all = (
-          (query_pubid.year && opts[:keep_year].nil?) ||
-          opts[:keep_year] ||
-          opts[:all_parts]
-        )
-        return ret if get_all
-
-        ret.to_most_recent_reference
-      rescue Pubid::Core::Errors::ParseError
-        Util.warn "(#{code}) Is not recognized as a standards identifier."
-        nil
-      end
+    # @param ref [String] the ISO standard Code to look up (e..g "ISO 9000")
+    # @param year [String, NilClass] the year the standard was published
+    # @param opts [Hash] options; restricted to :all_parts if all-parts
+    # @option opts [Boolean] :all_parts if all-parts reference is required
+    # @option opts [Boolean] :keep_year if undated reference should return
+    #   actual reference with year
+    #
+    # @return [RelatonIsoBib::IsoBibliographicItem] Bibliographic item
+    def get(ref, year = nil, opts = {}) # rubocop:disable Metrics/CyclomaticComplexity,Metrics/MethodLength,Metrics/PerceivedComplexity,Metrics/AbcSize
+      code = ref.gsub("\u2013", "-")
+
+      # parse "all parts" request
+      code.sub! " (all parts)", ""
+      opts[:all_parts] ||= $~ && opts[:all_parts].nil?
+
+      query_pubid = Pubid::Iso::Identifier.parse(code)
+      query_pubid.root.year = year.to_i if year&.respond_to?(:to_i)
+      Util.warn "(#{query_pubid}) Fetching from Relaton repository ..."
+
+      hits, missed_year_ids = isobib_search_filter(query_pubid, opts)
+      tip_ids = look_up_with_any_types_stages(hits, ref, opts)
+      ret = hits.fetch_doc
+      return fetch_ref_err(query_pubid, missed_year_ids, tip_ids) unless ret
+
+      response_pubid = ret.docidentifier.first.id # .sub(" (all parts)", "")
+      Util.warn "(#{query_pubid}) Found: `#{response_pubid}`"
+      get_all = (query_pubid.root.year && opts[:keep_year].nil?) || opts[:keep_year] || opts[:all_parts]
+      return ret if get_all
+
+      ret.to_most_recent_reference
+    rescue Pubid::Core::Errors::ParseError
+      Util.warn "(#{code}) Is not recognized as a standards identifier."
+      nil
+    end
 
-      # @param query_pubid [Pubid::Iso::Identifier]
-      # @param pubid [Pubid::Iso::Identifier]
-      # @param all_parts [Boolean] match with any parts when true
-      # @return [Boolean]
-      def matches_parts?(query_pubid, pubid, all_parts: false)
-        # match only with documents with part number
-        return !pubid.part.nil? if all_parts
+    # @param query_pubid [Pubid::Iso::Identifier]
+    # @param pubid [Pubid::Iso::Identifier]
+    # @param all_parts [Boolean] match with any parts when true
+    # @return [Boolean]
+    def matches_parts?(query_pubid, pubid, all_parts: false)
+      # match only with documents with part number
+      return !pubid.part.nil? if all_parts
 
-        query_pubid.part == pubid.part
-      end
+      query_pubid.part == pubid.part
+    end
 
-      #
-      # Matches base of query_pubid and pubid.
-      #
-      # @param [Pubid::Iso::Identifier] query_pubid pubid to match
-      # @param [Pubid::Iso::Identifier] pubid pubid to match
-      # @param [Boolean] any_types_stages match with any types and stages
-      #
-      # @return [<Type>] <description>
-      #
-      def matches_base?(query_pubid, pubid, any_types_stages: false) # rubocop:disable Metrics/AbcSize,Metrics/CyclomaticComplexity,Metrics?PerceivedComplexity
-        return false unless pubid.respond_to?(:publisher)
-
-        query_pubid.publisher == pubid.publisher &&
-          query_pubid.number == pubid.number &&
-          query_pubid.copublisher == pubid.copublisher &&
-          (any_types_stages || query_pubid.stage == pubid.stage) &&
-          (any_types_stages || query_pubid.is_a?(pubid.class))
+    #
+    # Matches base of query_pubid and pubid.
+    #
+    # @param [Pubid::Iso::Identifier] query_pubid pubid to match
+    # @param [Pubid::Iso::Identifier] pubid pubid to match
+    # @param [Boolean] any_types_stages match with any types and stages
+    #
+    # @return [<Type>] <description>
+    #
+    def matches_base?(query_pubid, pubid, any_types_stages: false) # rubocop:disable Metrics?PerceivedComplexity
+      return false unless pubid.respond_to?(:publisher)
+
+      query_pubid.publisher == pubid.publisher &&
+        query_pubid.number == pubid.number &&
+        query_pubid.copublisher == pubid.copublisher &&
+        (any_types_stages || query_pubid.stage == pubid.stage) &&
+        (any_types_stages || query_pubid.is_a?(pubid.class))
+    end
+
+    # @param hit_collection [RelatonIso::HitCollection]
+    # @param year [String]
+    # @return [Array<RelatonIso::HitCollection, Array<String>>] hits and missed year IDs
+    def filter_hits_by_year(hit_collection, year)
+      missed_year_ids = Set.new
+      return [hit_collection, missed_year_ids] if year.nil?
+
+      # filter by year
+      hits = hit_collection.select do |hit|
+        hit.pubid.year ||= hit.hit[:year]
+        next true if check_year(year, hit)
+
+        missed_year_ids << hit.pubid.to_s if hit.pubid.year
+        false
       end
 
-      # @param hit_collection [RelatonIso::HitCollection]
-      # @param year [String]
-      # @return [Array<RelatonIso::HitCollection, Array<String>>] hits and missed year IDs
-      def filter_hits_by_year(hit_collection, year)
-        missed_year_ids = Set.new
-        return [hit_collection, missed_year_ids] if year.nil?
+      [hits, missed_year_ids]
+    end
+
+    private
 
-        # filter by year
-        hits = hit_collection.select do |hit|
-          hit.pubid.year ||= hit.hit[:year]
-          next true if check_year(year, hit)
+    def check_year(year, hit) # rubocop:disable Metrics/AbcSize
+      (hit.pubid.base.nil? && hit.pubid.year.to_s == year.to_s) ||
+        (!hit.pubid.base.nil? && hit.pubid.base.year.to_s == year.to_s) ||
+        (!hit.pubid.base.nil? && hit.pubid.year.to_s == year.to_s)
+    end
 
-          missed_year_ids << hit.pubid.to_s if hit.pubid.year
-          false
-        end
+    # @param pubid [Pubid::Iso::Identifier] PubID with no results
+    def fetch_ref_err(pubid, missed_year_ids, tip_ids) # rubocop:disable Metrics/MethodLength,Metrics/AbcSize
+      Util.warn "(#{pubid}) Not found."
 
-        [hits, missed_year_ids]
+      if missed_year_ids.any?
+        ids = missed_year_ids.map { |i| "`#{i}`" }.join(", ")
+        Util.warn "(#{pubid}) TIP: No match for edition year #{pubid.year}, but matches exist for #{ids}."
       end
 
-      private
-
-      def check_year(year, hit) # rubocop:disable Metrics/AbcSize
-        (hit.pubid.base.nil? && hit.pubid.year.to_s == year.to_s) ||
-          (!hit.pubid.base.nil? && hit.pubid.base.year.to_s == year.to_s) ||
-          (!hit.pubid.base.nil? && hit.pubid.year.to_s == year.to_s)
+      if tip_ids.any?
+        ids = tip_ids.map { |i| "`#{i}`" }.join(", ")
+        Util.warn "(#{pubid}) TIP: Matches exist for #{ids}."
       end
 
-      # @param pubid [Pubid::Iso::Identifier] PubID with no results
-      def fetch_ref_err(pubid, missed_year_ids, tip_ids) # rubocop:disable Metrics/MethodLength,Metrics/AbcSize
-        Util.warn "(#{pubid}) Not found."
-
-        if missed_year_ids.any?
-          ids = missed_year_ids.map { |i| "`#{i}`" }.join(", ")
-          Util.warn "(#{pubid}) TIP: No match for edition year " \
-                    "#{pubid.year}, but matches exist for #{ids}."
-        end
-
-        if tip_ids.any?
-          ids = tip_ids.map { |i| "`#{i}`" }.join(", ")
-          Util.warn "(#{pubid}) TIP: Matches exist for #{ids}."
-        end
-
-        if pubid.part
-          Util.warn "(#{pubid}) TIP: If it cannot be found, " \
-                    "the document may no longer be published in parts."
-        else
-          Util.warn "(#{pubid}) TIP: If you wish to cite " \
-                    "all document parts for the reference, use " \
-                    "`#{pubid.to_s(format: :ref_undated)} (all parts)`."
-        end
-
-        nil
+      if pubid.part
+        Util.warn "(#{pubid}) TIP: If it cannot be found, the document may no longer be published in parts."
+      else
+        Util.warn "(#{pubid}) TIP: If you wish to cite all document parts for the reference, " \
+                  "use `#{pubid.to_s(format: :ref_undated)} (all parts)`."
       end
 
-      def look_up_with_any_types_stages(hits, ref, opts) # rubocop:disable Metrics/MethodLength
-        found_ids = []
-        return found_ids if hits.from_gh || hits.any? || !ref.match?(/^ISO[\/\s][A-Z]/)
+      nil
+    end
+
+    def look_up_with_any_types_stages(hits, ref, opts)
+      return [] if hits.any? || !ref.match?(/^ISO[\/\s][A-Z]/)
 
-        ref_no_type_stage = ref.sub(/^ISO[\/\s][A-Z]+/, "ISO")
-        pubid = Pubid::Iso::Identifier.parse(ref_no_type_stage)
-        resp, = isobib_search_filter(pubid, opts, any_types_stages: true)
-        resp.map &:pubid
-      end
+      ref_no_type_stage = ref.sub(/^ISO[\/\s][A-Z]+/, "ISO")
+      pubid = Pubid::Iso::Identifier.parse(ref_no_type_stage)
+      resp, = isobib_search_filter(pubid, opts, any_types_stages: true)
+      resp.map &:pubid
+    end
 
-      #
-      # Search for hits. If no found then trying missed stages.
-      #
-      # @param query_pubid [Pubid::Iso::Identifier] reference without correction
-      # @param opts [Hash]
-      # @param any_types_stages [Boolean] match with any stages
-      #
-      # @return [Array<RelatonIso::HitCollection, Array<String>>] hits and missed years
-      #
-      def isobib_search_filter(query_pubid, opts, any_types_stages: false) # rubocop:disable Metrics/MethodLength,Metrics/AbcSize,Metrics/CyclomaticComplexity,Metrics/PerceivedComplexity
-        query_pubid_without_year = query_pubid.dup
-        # remove year for query
-        query_pubid_without_year.year = nil
-        hit_collection = search(query_pubid_without_year.to_s)
-
-        # filter only matching hits
-        filter_hits hit_collection, query_pubid, opts[:all_parts],
-                    any_types_stages
+    #
+    # Search for hits. If no found then trying missed stages.
+    #
+    # @param query_pubid [Pubid::Iso::Identifier] reference without correction
+    # @param opts [Hash]
+    # @param any_types_stages [Boolean] match with any stages
+    #
+    # @return [Array<RelatonIso::HitCollection, Array<String>>] hits and missed years
+    #
+    def isobib_search_filter(query_pubid, opts, any_types_stages: false)
+      hit_collection = search(query_pubid, opts)
+
+      # filter only matching hits
+      filter_hits hit_collection, query_pubid, opts[:all_parts], any_types_stages
+    end
+
+    #
+    # Filter hits by query_pubid.
+    #
+    # @param hit_collection [RelatonIso::HitCollection]
+    # @param query_pubid [Pubid::Iso::Identifier]
+    # @param all_parts [Boolean]
+    # @param any_types_stages [Boolean]
+    #
+    # @return [Array<RelatonIso::HitCollection, Array<String>>] hits and missed year IDs
+    #
+    def filter_hits(hit_collection, query_pubid, all_parts, any_types_stages)
+      # filter out
+      excludings = build_excludings(all_parts, any_types_stages)
+      no_year_ref = hit_collection.ref_pubid_no_year.exclude(*excludings)
+      result = hit_collection.select do |i|
+        pubid_match?(i.pubid, query_pubid, excludings, no_year_ref) && !(all_parts && i.pubid.part.nil?)
       end
 
-      #
-      # Filter hits by query_pubid.
-      #
-      # @param hit_collection [RelatonIso::HitCollection]
-      # @param query_pubid [Pubid::Iso::Identifier]
-      # @param all_parts [Boolean]
-      # @param any_stypes_tages [Boolean]
-      #
-      # @return [Array<RelatonIso::HitCollection, Array<String>>] hits and missed year IDs
-      #
-      def filter_hits(hit_collection, query_pubid, all_parts, any_stypes_tages) # rubocop:disable Metrics/AbcSize,Metrics/CyclomaticComplexity,Metrics/PerceivedComplexity
-        # filter out
-        result = hit_collection.select do |i|
-          hit_pubid = i.pubid
-          matches_base?(query_pubid, hit_pubid,
-                        any_types_stages: any_stypes_tages) &&
-            matches_parts?(query_pubid, hit_pubid, all_parts: all_parts) &&
-            query_pubid.corrigendums == hit_pubid.corrigendums &&
-            query_pubid.amendments == hit_pubid.amendments
-        end
-
-        filter_hits_by_year(result, query_pubid.year)
+      filter_hits_by_year(result, query_pubid.year)
+    end
+
+    def build_excludings(all_parts, any_types_stages)
+      excludings = %i[year edition]
+      excludings += %i[type stage iteration] if any_types_stages
+      excludings << :part if all_parts
+      excludings
+    end
+
+    def pubid_match?(pubid, query_pubid, excludings, no_year_ref)
+      if pubid.is_a? String then pubid == query_pubid.to_s
+      else
+        pubid = pubid.dup
+        pubid.base = pubid.base.exclude(:year, :edition) if pubid.base
+        pubid.exclude(*excludings) == no_year_ref
       end
     end
   end

data/lib/relaton_iso/processor.rb

@@ -9,16 +9,29 @@ module RelatonIso
       @prefix = "ISO"
       @defaultprefix = %r{^ISO(/IEC)?\s}
       @idtype = "ISO"
+      @datasets = %w[iso-ics]
     end
 
     # @param code [String]
-    # @param date [String, NilClass] year
+    # @param date [String, nil] year
     # @param opts [Hash]
     # @return [RelatonIsoBib::IsoBibliographicItem]
     def get(code, date, opts)
       ::RelatonIso::IsoBibliography.get(code, date, opts)
     end
 
+    #
+    # Fetch all the documents from https://www.iso.org/standards-catalogue/browse-by-ics.html
+    #
+    # @param [String] source source name (iso-rss, iso-rss-all)
+    # @param [Hash] opts
+    # @option opts [String] :output directory to output documents
+    # @option opts [String] :format output format (xml, yaml, bibxml)
+    #
+    def fetch_data(_source, opts)
+      DataFetcher.fetch(**opts)
+    end
+
     # @param xml [String]
     # @return [RelatonIsoBib::IsoBibliographicItem]
     def from_xml(xml)
@@ -28,7 +41,7 @@ module RelatonIso
     # @param hash [Hash]
     # @return [RelatonIsoBib::IsoBibliographicItem]
     def hash_to_bib(hash)
-      item_hash = ::RelatonIsoBib::HashConverter.hash_to_bib(hash)
+      item_hash = HashConverter.hash_to_bib(hash)
       ::RelatonIsoBib::IsoBibliographicItem.new(**item_hash)
     end
 
@@ -43,5 +56,12 @@ module RelatonIso
     def threads
       3
     end
+
+    #
+    # Remove index file
+    #
+    def remove_index_file
+      Relaton::Index.find_or_create(:iso, url: true, file: HitCollection::INDEXFILE).remove_file
+    end
   end
 end

data/lib/relaton_iso/queue.rb

@@ -0,0 +1,61 @@
+module RelatonIso
+  #
+  # Queue of links to fetch.
+  #
+  class Queue
+    extend Forwardable
+    def_delegator :queue, :[]
+
+    FILE = "iso-queue.txt".freeze
+
+    #
+    # Open queue file if exist. If not, create new empty queue.
+    #
+    # @return [Array<String>] queue
+    #
+    def queue
+      @queue ||= File.exist?(FILE) ? File.read(FILE).split("\n") : []
+    end
+
+    #
+    # Add item to queue at first position if it is not already there.
+    #
+    # @param [String] item item to add
+    #
+    # @return [void]
+    #
+    def add_first(item)
+      queue.unshift item unless queue.include? item
+    end
+
+    #
+    # Move or add item to the end of the queue.
+    #
+    # @param [String] item item to move or add
+    #
+    # @return [void]
+    #
+    def move_last(item)
+      queue.delete item
+      queue << item
+    end
+
+    #
+    # Take first item from the queue.
+    #
+    # @return [String] an item
+    #
+    # def take_first
+    #   queue.shift
+    # end
+
+    #
+    # Save queue to file.
+    #
+    # @return [void]
+    #
+    def save
+      File.write FILE, queue.to_a.join("\n")
+    end
+  end
+end