relaton-iso 1.16.0 → 1.16.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b3ca89f9616730766487b295af824ea9b797aa863f0bbc248e21e8f26862e02c
-  data.tar.gz: 070dedc5d81dfeea1085da56872e960a021691193c0ad1b7b47c4a3c8704a2ed
+  metadata.gz: 479a728a58c56799448fd6d468e0d19fe245b731119f8dcd9ae6f19a7b624e07
+  data.tar.gz: ac89507180ca01978bfe98b68fbe02450f2c33015bd38d788752f3bf933911ad
 SHA512:
-  metadata.gz: a62229e67a1547c934babfb52d77d1a2c51f0982840bca9a889499479cbdc0b2e7bcbc0291bfa0812afaf25eb5d13f6c760b073a5b6742b588275a922fcbe210
-  data.tar.gz: 46ef00eeef8fa1b4f55281ae7566fbee4f5ac743c8d90bb0053fd12255121ff5e994e47b16e441c019137e8962a3d4ce298f954b29dd08b219537ba3d8d70edd
+  metadata.gz: 71cc49dc2afa8690f02f7035ec5cc13981eb620e2b8c3792456401c152a4ca8192b2ffbd7445c6c982886e61f679427a2d5afbf26e13c6ebcfffcc8d54f7e5c9
+  data.tar.gz: 853da0772a998533c5f461ff297bef978c75e1f58b2df1fec5eff0fea6d306807420453a6ba37b1348d44e58c3a71dcf743228c7703b73fd9b7d72c9d4309598

data/README.adoc

@@ -72,14 +72,14 @@ item.docidentifier.detect { |di| di.type == "URN" }.id
 [source,ruby]
 ----
 item = RelatonIso::IsoBibliography.get "ISO 19115:2003"
-[relaton-iso] ("ISO 19115:2003") fetching from ISO...
-[relaton-iso] ("ISO 19115:2003") Found exact match.
+[relaton-iso] (ISO 19115:2003) Fetching from iso.org ...
+[relaton-iso] (ISO 19115:2003) Found: `ISO 19115:2003`
 => #<RelatonIsoBib::IsoBibliographicItem:0x00007f8c83429e30
 ...
 
 item = RelatonIso::IsoBibliography.get "ISO 19115", "2003"
-[relaton-iso] ("ISO 19115:2003") Fetching from ISO...
-[relaton-iso] ("ISO 19115:2003") Found ("ISO 19115:2003").
+[relaton-iso] (ISO 19115:2003) Fetching from iso.org ...
+[relaton-iso] (ISO 19115:2003) Found: `ISO 19115:2003`
 => #<RelatonIsoBib::IsoBibliographicItem:0x0000000112c9ca80
 ...
 
@@ -92,8 +92,8 @@ item.docidentifier[0].id
 [source,ruby]
 ----
 item = RelatonIso::IsoBibliography.get "ISO 19115"
-[relaton-iso] ("ISO 19115") fetching from ISO...
-[relaton-iso] ("ISO 19115") Found ("ISO 19115:2003").
+[relaton-iso] (ISO 19115) Fetching from iso.org ...
+[relaton-iso] (ISO 19115) Found: `ISO 19115:2003`
 => #<RelatonIsoBib::IsoBibliographicItem:0x00007f8c830275a8
 ...
 
@@ -106,8 +106,8 @@ item.docidentifier[0].id
 [source,ruby]
 ----
 item = RelatonIso::IsoBibliography.get "ISO 19115-1"
-[relaton-iso] ("ISO 19115-1") fetching from ISO...
-[relaton-iso] ("ISO 19115-1") Found ("ISO 19115-1:2014").
+[relaton-iso] (ISO 19115-1) Fetching from iso.org ...
+[relaton-iso] (ISO 19115-1) Found: `ISO 19115-1:2014`
 => #<RelatonIsoBib::IsoBibliographicItem:0x00007f8c83408af0
 ...
 
@@ -120,14 +120,14 @@ item.docidentifier[0].id
 [source,ruby]
 ----
 item = RelatonIso::IsoBibliography.get "ISO 19115 (all parts)"
-[relaton-iso] ("ISO 19115") Fetching from ISO...
-[relaton-iso] ("ISO 19115") Found ("ISO 19115").
+[relaton-iso] (ISO 19115) Fetching from iso.org ...
+[relaton-iso] (ISO 19115) Found: `ISO 19115`
 => #<RelatonIsoBib::IsoBibliographicItem:0x00007f8ca216e118
 ...
 
 item = RelatonIso::IsoBibliography.get "ISO 19115", nil, all_parts: true
-[relaton-iso] ("ISO 19115") Fetching from ISO...
-[relaton-iso] ("ISO 19115") Found ("ISO 19115").
+[relaton-iso] (ISO 19115) Fetching from iso.org ...
+[relaton-iso] (ISO 19115) Found: `ISO 19115`
 => #<RelatonIsoBib::IsoBibliographicItem:0x00007f8c830f3d38
 ...
 
@@ -135,13 +135,13 @@ item.docidentifier[0].id
 => "ISO 19115 (all parts)"
 
 item = RelatonIso::IsoBibliography.get "ISO 19115-1 (all parts)"
-[relaton-iso] ("ISO 19115") Fetching from ISO...
-[relaton-iso] ("ISO 19115") Found ("ISO 19115").
+[relaton-iso] (ISO 19115) Fetching from iso.org ...
+[relaton-iso] (ISO 19115) Found: `ISO 19115`
 => #<RelatonIsoBib::IsoBibliographicItem:0x00007f8c8290e5a0
 
 item = RelatonIso::IsoBibliography.get "ISO 19115-1", nil, all_parts: true
-[relaton-iso] ("ISO 19115") Fetching from ISO...
-[relaton-iso] ("ISO 19115") Found ("ISO 19115").
+[relaton-iso] (ISO 19115) Fetching from iso.org ...
+[relaton-iso] (ISO 19115) Found: `ISO 19115`
 => #<RelatonIsoBib::IsoBibliographicItem:0x00007f8c925355b8
 ...
 
@@ -234,8 +234,8 @@ item.title lang: 'fr'
     @type="main">]>
 
 item = RelatonIso::IsoBibliography.get "ISO 19115:2003"
-[relaton-iso] ("ISO 19115:2003") Fetching from ISO...
-[relaton-iso] ("ISO 19115:2003") Found ("ISO 19115:2003").
+[relaton-iso] (ISO 19115:2003) Fetching from iso.org ...
+[relaton-iso] (ISO 19115:2003) Found: `ISO 19115:2003`
 => #<RelatonIsoBib::IsoBibliographicItem:0x00007fa8870b69e0
 
 item.abstract lang: 'en'

data/lib/relaton_iso/hit.rb

@@ -42,8 +42,9 @@ module RelatonIso
     # @return [Pubid::Iso::Identifier]
     def pubid
       @pubid ||= Pubid::Iso::Identifier.parse_from_title(hit[:title])
-    rescue Pubid::Iso::Errors::WrongTypeError, Pubid::Iso::Errors::ParseError => e
-      Util.warn "unable to find an identifier in `#{hit[:title]}`."
+    rescue Pubid::Iso::Errors::WrongTypeError,
+           Pubid::Iso::Errors::ParseError => e
+      Util.warn "Unable to find an identifier in: `#{hit[:title]}`."
       Util.warn e.message
     end
   end

data/lib/relaton_iso/iso_bibliography.rb

@@ -38,7 +38,7 @@ module RelatonIso
         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..."
+        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)
@@ -54,7 +54,7 @@ module RelatonIso
         response_docid = ret.docidentifier.first.id.sub(" (all parts)", "")
         response_pubid = Pubid::Iso::Identifier.parse(response_docid)
 
-        Util.warn "(#{query_pubid}) Found `#{response_pubid}`."
+        Util.warn "(#{query_pubid}) Found: `#{response_pubid}`"
 
         get_all = (
           (query_pubid.year && opts[:keep_year].nil?) ||
@@ -65,7 +65,7 @@ module RelatonIso
 
         ret.to_most_recent_reference
       rescue Pubid::Core::Errors::ParseError
-        Util.warn "(#{code}) is not recognized as a standards identifier."
+        Util.warn "(#{code}) Is not recognized as a standards identifier."
         nil
       end
 
@@ -179,7 +179,8 @@ module RelatonIso
         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
+        filter_hits hit_collection, query_pubid, opts[:all_parts],
+                    any_types_stages
       end
 
       #
@@ -196,7 +197,8 @@ module RelatonIso
         # 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_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

data/lib/relaton_iso/scrapper.rb

@@ -270,11 +270,10 @@ module RelatonIso
       # @param doc [Nokogiri::HTML::Document]
       # @return [Hash]
       def fetch_workgroup(doc) # rubocop:disable Metrics/MethodLength,Metrics/AbcSize,Metrics/CyclomaticComplexity,Metrics/PerceivedComplexity
-        wg = doc.at("//div[@class='clearfix']")
-        wg_link = wg.at "span/a"
-        return unless wg_link
+        wg = doc.at("////div[contains(., 'Technical Committe')]/following-sibling::span/a")
+        return unless wg
 
-        workgroup = wg_link.text.split "/"
+        workgroup = wg.text.split "/"
         type = workgroup[1]&.match(/^[A-Z]+/)&.to_s || "TC"
         # {
         #   name: "International Organization for Standardization",
@@ -282,44 +281,48 @@ module RelatonIso
         #   url: "www.iso.org",
         # }
         tc_numb = workgroup[1]&.match(/\d+/)&.to_s&.to_i
-        tc_name = wg.at("span[@class='entry-title']").text
-        tc = RelatonBib::WorkGroup.new(name: tc_name, identifier: wg_link.text,
+        tc_name = wg[:title]
+        tc = RelatonBib::WorkGroup.new(name: tc_name, identifier: wg.text,
                                        type: type, number: tc_numb)
         RelatonIsoBib::EditorialGroup.new(technical_committee: [tc])
       end
 
-      # rubocop:disable Metrics/MethodLength
-
       # Fetch relations.
       # @param doc [Nokogiri::HTML::Document]
       # @return [Array<Hash>]
-      def fetch_relations(doc) # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity
+      def fetch_relations(doc)
         types = ["Now", "Now under review"]
         doc.xpath("//ul[@class='steps']/li", "//div[@class='sub-step']").reduce([]) do |a, r|
-          r_type = r.at("h4", "h5").text
-          date = []
-          type = case r_type.strip
-                 when "Previously", "Will be replaced by" then "obsoletes"
-                 when "Corrigenda / Amendments", "Revised by", "Now confirmed"
-                   on = doc.xpath('//span[@class="stage-date"][contains(., "-")]').last
-                   date << { type: "circulated", on: on.text } if on
-                   "updates"
-                 else r_type
-                 end
-          if types.include?(type) then a
-          else
-            a + r.css("a").map do |id|
-              docid = RelatonBib::DocumentIdentifier.new(type: "ISO", id: id.text, primary: true)
-              fref = RelatonBib::FormattedRef.new(content: id.text, format: "text/plain")
-              bibitem = RelatonIsoBib::IsoBibliographicItem.new(
-                docid: [docid], formattedref: fref, date: date,
-              )
-              { type: type, bibitem: bibitem }
+          type, date = relation_type(r.at("h4", "h5").text.strip, doc)
+          next a if types.include?(type)
+
+          a + create_relations(r, type, date)
+        end
+      end
+
+      def relation_type(type, doc)
+        date = []
+        t = case type.strip
+            when "Previously", "Will be replaced by" then "obsoletes"
+            when "Corrigenda / Amendments", "Revised by", "Now confirmed"
+              on = doc.xpath('//span[@class="stage-date"][contains(., "-")]').last
+              date << { type: "circulated", on: on.text } if on
+              "updates"
+            else type
             end
-          end
+        [t, date]
+      end
+
+      def create_relations(rel, type, date)
+        rel.css("a").map do |id|
+          docid = RelatonBib::DocumentIdentifier.new(type: "ISO", id: id.text, primary: true)
+          fref = RelatonBib::FormattedRef.new(content: id.text, format: "text/plain")
+          bibitem = RelatonIsoBib::IsoBibliographicItem.new(
+            docid: [docid], formattedref: fref, date: date,
+          )
+          { type: type, bibitem: bibitem }
         end
       end
-      # rubocop:enable Metrics/MethodLength
 
       # Fetch type.
       # @param ref [String]
@@ -343,14 +346,20 @@ module RelatonIso
       # @param doc [Nokogiri::HTML::Document]
       # @param lang [String]
       # @return [Array<RelatonBib::TypedTitleString>]
-      def fetch_title(doc, lang)
-        content = doc.at(
-          "//nav[contains(@class,'heading-condensed')]/h2 | "\
-          "//nav[contains(@class,'heading-condensed')]/h3",
-        )&.text&.gsub(/\u2014/, "-")
-        return RelatonBib::TypedTitleStringCollection.new unless content
-
-        RelatonBib::TypedTitleString.from_string content, lang, script(lang)
+      def fetch_title(doc, lang) # rubocop:disable Metrics/MethodLength,Metrics/AbcSize
+        head = doc.at "//nav[contains(@class,'heading-condensed')]"
+        types = { "h2" => "title-intro", "h3" => "title-main", "h4" => "title-part" }
+        title_types = head.xpath("h2 | h3 | h4").each_with_object({}) do |t, h|
+          h[types[t.name]] = t.text
+        end
+        title = RelatonBib::TypedTitleStringCollection.new
+        title_types.each do |type, content|
+          title << RelatonBib::TypedTitleString.new(
+            type: type, content: content, language: lang, script: script(lang),
+          )
+        end
+        main = title.map { |t| t.title.content }.join " - "
+        title << RelatonBib::TypedTitleString.new(type: "main", content: main, language: lang, script: script(lang))
       end
 
       # Return ISO script code.
@@ -363,12 +372,11 @@ module RelatonIso
         end
       end
 
-      # rubocop:disable Metrics/MethodLength
       # Fetch dates
       # @param doc [Nokogiri::HTML::Document]
       # @param ref [String]
       # @return [Array<Hash>]
-      def fetch_dates(doc, ref) # rubocop:disable Metrics/AbcSize, Metrics/PerceivedComplexity
+      def fetch_dates(doc, ref) # rubocop:disable Metrics/AbcSize, Metrics/PerceivedComplexity, Metrics/MethodLength
         dates = []
         %r{^[^\s]+\s[\d-]+:(?<ref_date_str>\d{4})} =~ ref
         pub_date_str = doc.xpath("//span[@itemprop='releaseDate']").text
@@ -400,25 +408,28 @@ module RelatonIso
           mem << { entity: publisher, role: [type: "publisher"] }
         end
       end
-      # rubocop:enable Metrics/MethodLength
 
       # Fetch ICS.
       # @param doc [Nokogiri::HTML::Document]
       # @return [Array<Hash>]
       def fetch_ics(doc)
-        doc.xpath("//dl[dt/strong[.='ICS']]/dd/span/a").map do |i|
+        doc.xpath("//div[contains(., 'ICS')]/following-sibling::span/a").map do |i|
           code = i.text.match(/[\d.]+/).to_s.split "."
           { field: code[0], group: code[1], subgroup: code[2] }
         end
       end
 
+      #
       # Fetch links.
-      # @param doc [Nokogiri::HTML::Document]
-      # @param url [String]
+      #
+      # @param doc [Nokogiri::HTML::Document] document to parse
+      # @param url [String] document url
+      #
       # @return [Array<Hash>]
+      #
       def fetch_link(doc, url)
         links = [{ type: "src", content: url }]
-        obp = doc.at_css("a#obp-preview")
+        obp = doc.at("//h4[contains(@class, 'h5')]/a")
         links << { type: "obp", content: obp[:href] } if obp
         rss = doc.at("//a[contains(@href, 'rss')]")
         links << { type: "rss", content: DOMAIN + rss[:href] } if rss

data/lib/relaton_iso/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module RelatonIso
-  VERSION = "1.16.0"
+  VERSION = "1.16.2"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: relaton-iso
 version: !ruby/object:Gem::Version
-  version: 1.16.0
+  version: 1.16.2
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2023-09-03 00:00:00.000000000 Z
+date: 2023-10-20 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: algolia