relaton-bipm 1.12.3 → 1.13.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b6da8165d7298a3658db3d941f2e0540e61bc53ddd726a79383cc36dc77f1b9a
-  data.tar.gz: 6748ecb244655bbbe8ef3226c5f72572d89104f591c256fe3db9f59cd3730b83
+  metadata.gz: 8684961db2e3b70deffa9ba3a943ac4ce4990edd0b00a0e5311c0d92ca61f344
+  data.tar.gz: 4d585b8f37043d19e14bf2bf9f1a9ee32186978f86db61a9fa2aade824171133
 SHA512:
-  metadata.gz: c58dfe2d3407b0b0ec5bcc42a412eb740ff7834559a533e7f8677789590f78016ef58111936cc7021a76b263f6f4dfce5d636b559d0fd83a5484c822a2aa2bb4
-  data.tar.gz: b5e9ad9000dd8605eefc6cc06957d9841e9b283f3f471616ba681f64086ffc70004ca929c0d291dd66a3b46a56491cf96ca4dceaf2e36de599a876f84aef40d0
+  metadata.gz: f426fa79bd1c71061fe411ff764435817c49d0a9c9e56333304ae54390a6f20384dbdaf40fba7e19062c3618ca212fbd2d21dc3180ee461f75d0d2f5d6c6ffe3
+  data.tar.gz: 0b7b9707e8cd16955591edd14592bc59550cb0efb21cfcdceacd5733a471af0d4b8ee4ec39138e8ae6653f22e006144195e1ab8e1a685e95199f2f6408ebdee5

data/README.adoc

@@ -39,35 +39,34 @@ Or install it yourself as:
 
 Allowed document names are:
 
-- guide-ccem-ampere-realization
-- guide-ccl-metre-realization-1
-- guide-ccl-metre-realization-2
-- guide-ccl-metre-realization-3
-- guide-ccm-kilogram-calibration
-- guide-ccm-kilogram-dissemination
-- mep-ampere-definition
-- mep-candela-definition
-- mep-candela-rapport
-- mep-kelvin-absolute-uncertainty
-- mep-kelvin-absolute
-- mep-kelvin-definition
-- mep-kelvin-lt-johnson-noise
-- mep-kelvin-relative
-- mep-kilogram-definition
-- mep-metre-definition
-- mep-mole-definition
-- mep-second-definition
-- si-brochure
-- sib-a4
+- CCEM-GD-RSI-1
+- CCL-GD-MeP-1
+- CCL-GD-MeP-2
+- CCL-GD-MeP-3
+- CCM-GD-RSI-1
+- CCM-GD-RSI-2
+- SI MEP A1
+- SI MEP Cd1
+- Rapport BIPM-2019/05
+- SI MEP KUPRTM
+- SI MEP KAPRT
+- SI MEP K1
+- SI MEP KLJNT
+- SI MEP KRPRT
+- SI MEP Kg1
+- SI MEP M1
+- SI MEP Mol1
+- SI MEP S1
+- SI Brochure
 
 ==== Reference structire for Metrologia documents
 
-`BIPM Metrologia {JOURNAL} {VOLUME} {ISSUE} {ARTICLE}`
+`BIPM Metrologia {JOURNAL} {VOLUME} {ISSUE} {PAGE}`
 
 - `{JOURNAL}` - number of journal, required
 - `{VOLUME}` - number of volume, optional
 - `{ISSUE}` - number of issue, optional
-- `{ARTICLE}` - number of article, optional
+- `{PAGE}` - number of page, optional
 
 ==== Reference structures for CCTF (CCDS), CGMP, CIPM documents
 
@@ -94,13 +93,13 @@ require 'relaton_bipm'
 => true
 
 # get BIPM brochure
-item = RelatonBipm::BipmBibliography.get "BIPM si-brochure"
-[relaton-bipm] ("BIPM si-brochure") fetching...
-[relaton-bipm] ("BIPM si-brochure") found si-brochure
+item = RelatonBipm::BipmBibliography.get "BIPM SI Brochure"
+[relaton-bipm] ("BIPM SI Brochure") fetching...
+[relaton-bipm] ("BIPM SI Brochure") found BIPM SI Brochure
 => #<RelatonBipm::BipmBibliographicItem:0x007ffb83982fe8
 ...
 
-# get BIPM Metrologia article
+# get BIPM Metrologia page
 bib = RelatonBipm::BipmBibliography.get "BIPM Metrologia 29 6 373"
 [relaton-bipm] ("BIPM Metrologia 29 6 373") fetching...
 [relaton-bipm] ("BIPM Metrologia 29 6 373") found Metrologia 29 6 373
@@ -172,11 +171,12 @@ RelatonBipm::BipmBibliography.get "CIPM Recommendation 1948"
 [source,ruby]
 ----
 item.to_xml
-=> "<bibitem id="si-brochure" type="standard">
+=> "<bibitem id="BIPMSIBrochure" type="standard">
       <fetched>2021-01-10</fetched>
       <title format="text/plain" language="en" script="Latn">The International System of Units (SI)</title>
       <title format="text/plain" language="fr" script="Latn">Le Système international d’unités (SI)</title>
-      <docidentifier type="BIPM">si-brochure</docidentifier>
+      <uri type="src">https://www.bipm.org/en/publications/si-brochure</uri>
+      <docidentifier type="BIPM">BIPM SI Brochure</docidentifier>
       <date type="updated">
         <on>2019-05-20</on>
       </date>
@@ -205,7 +205,7 @@ item.to_xml bibdata: true
       <title format="text/plain" language="en" script="Latn">The International System of Units (SI)</title>
       <title format="text/plain" language="fr" script="Latn">Le Système international d’unités (SI)</title>
       <uri type="src">https://www.bipm.org/en/publications/si-brochure</uri>
-      <docidentifier type="BIPM">si-brochure</docidentifier>
+      <docidentifier type="BIPM">BIPM SI Brochure</docidentifier>
       <date type="updated">
         <on>2019-05-20</on>
       </date>

data/grammars/biblio.rng

@@ -225,9 +225,9 @@
         <zeroOrMore>
           <ref name="forename"/>
         </zeroOrMore>
-        <zeroOrMore>
-          <ref name="initial"/>
-        </zeroOrMore>
+        <optional>
+          <ref name="formatted-initials"/>
+        </optional>
         <ref name="surname"/>
         <zeroOrMore>
           <ref name="addition"/>
@@ -247,8 +247,8 @@
       <ref name="LocalizedString"/>
     </element>
   </define>
-  <define name="initial">
-    <element name="initial">
+  <define name="formatted-initials">
+    <element name="formatted-initials">
       <ref name="LocalizedString"/>
     </element>
   </define>
@@ -264,6 +264,9 @@
   </define>
   <define name="forename">
     <element name="forename">
+      <optional>
+        <attribute name="initial"/>
+      </optional>
       <ref name="LocalizedString"/>
     </element>
   </define>

data/grammars/isodoc.rng

@@ -69,6 +69,17 @@
         </zeroOrMore>
       </element>
     </define>
+    <define name="AdmonitionType">
+      <choice>
+        <value>warning</value>
+        <value>note</value>
+        <value>tip</value>
+        <value>important</value>
+        <value>caution</value>
+        <value>statement</value>
+        <value>editorial</value>
+      </choice>
+    </define>
     <define name="index">
       <element name="index">
         <optional>
@@ -1274,7 +1285,12 @@
   </define>
   <define name="span">
     <element name="span">
-      <attribute name="class"/>
+      <optional>
+        <attribute name="class"/>
+      </optional>
+      <optional>
+        <attribute name="style"/>
+      </optional>
       <oneOrMore>
         <ref name="TextElement"/>
       </oneOrMore>
@@ -2493,6 +2509,16 @@
           <text/>
         </element>
       </optional>
+      <optional>
+        <element name="amendment">
+          <text/>
+        </element>
+      </optional>
+      <optional>
+        <element name="corrigendum">
+          <text/>
+        </element>
+      </optional>
       <optional>
         <element name="language">
           <text/>

data/lib/relaton_bipm/bipm_bibliography.rb

@@ -15,11 +15,14 @@ module RelatonBipm
     class << self
       # @param text [String]
       # @return [RelatonBipm::BipmBibliographicItem]
-      def search(text, _year = nil, _opts = {}) # rubocop:disable Metrics/AbcSize
+      def search(text, _year = nil, _opts = {}) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
         warn "[relaton-bipm] (\"#{text}\") fetching..."
         ref = text.sub(/^BIPM\s/, "")
         item = ref.match?(/^Metrologia/i) ? get_metrologia(ref, magent) : get_bipm(ref, magent)
-        return unless item
+        unless item
+          warn "[relaton-bipm] (\"#{text}\") not found."
+          return
+        end
 
         warn("[relaton-bipm] (\"#{text}\") found #{item.docidentifier[0].id}")
         item
@@ -216,11 +219,19 @@ module RelatonBipm
       # @param art [String]
       # @param agent [Mechanize]
       # @return [RelatonBipm::BipmBibliographicItem]
-      def get_article_from_issue(vol, ish, art, agent)
+      def get_article_from_issue(vol, ish, art, agent) # rubocop:disable Metrics/MethodLength
         url = issue_url vol, ish
         rsp = agent.get url
         check_response rsp
-        get_article rsp.at("//div[@class='indexer'][.='#{art}']/../div/a")[:href], vol, ish, agent
+        link = rsp.at("//div[@class='indexer'][.='#{art}']/../div/a")
+        unless link
+          arts = rsp.xpath("//div[@class='indexer']").map(&:text)
+          warn "[relaton-bipm] Page #{art} not found in \"BIPM Metrologia #{vol} #{ish}\" issue."
+          warn "[relaton-bipm] Availabe pages in the issue are: (#{arts.join(', ')})"
+          return
+        end
+
+        get_article link[:href], vol, ish, agent
       end
 
       # @param path [String]
@@ -329,11 +340,13 @@ module RelatonBipm
       #
       def check_response(rsp) # rubocop:disable Metrics/AbcSize
         if rsp.code == "302"
+          warn "[relaton-bipm] This source employs anti-DDoS measures that unfortunately affects automated requests."
+          warn "[relaton-bipm] Please visit this link in your browser to resolve the CAPTCHA, then retry: #{rsp.uri}"
           warn "[relaton-bipm] #{rsp.uri} is redirected to #{rsp.header['location']}"
-          raise RelatonBib::RequestError, "redirected to #{rsp.header['location']}"
+          raise RelatonBib::RequestError, "cannot access #{rsp.uri}"
         elsif rsp.code != "200"
           warn "[read_bipm] can't acces #{rsp.uri} #{rsp.code}"
-          raise RelatonBib::RequestError, "can't acces #{rsp.uri} #{rsp.code}"
+          raise RelatonBib::RequestError, "cannot acces #{rsp.uri} #{rsp.code}"
         end
       end
     end

data/lib/relaton_bipm/data_fetcher.rb

@@ -57,17 +57,24 @@ module RelatonBipm
     # Parse SI brochure and write them to YAML files
     #
     def parse_si_brochure # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
+      # puts "Parsing SI brochure..."
+      # puts "Ls #{Dir['*']}"
+      # puts "Ls #{Dir['bipm-si-brochure/*']}"
+      # puts "Ls #{Dir['bipm-si-brochure/site/*']}"
+      # puts "Ls #{Dir['bipm-si-brochure/site/documents/*']}"
       Dir["bipm-si-brochure/site/documents/*.rxl"].each do |f|
+        puts "Parsing #{f}"
         docstd = Nokogiri::XML File.read f
         doc = docstd.at "/bibdata"
         hash1 = RelatonBipm::XMLParser.from_xml(doc.to_xml).to_hash
+        fix_si_brochure_id hash1
         hash1["fetched"] = Date.today.to_s
-        hash1["docid"].detect { |id| id["type"] == "BIPM" }["primary"] = true
         outfile = File.join @output, File.basename(f).sub(/(?:-(?:en|fr))?\.rxl$/, ".yaml")
         @index[[hash1["docnumber"] || File.basename(outfile, ".yaml")]] = outfile
         hash = if File.exist? outfile
                  warn_duplicate = false
                  hash2 = YAML.load_file outfile
+                 fix_si_brochure_id hash2
                  deep_merge hash1, hash2
                else
                  warn_duplicate = true
@@ -75,9 +82,33 @@ module RelatonBipm
                end
         item = RelatonBipm::BipmBibliographicItem.from_hash(**hash)
         write_file outfile, item, warn_duplicate: warn_duplicate
+        puts "Saved to #{outfile}"
       end
     end
 
+    #
+    # Update ID of SI brochure
+    #
+    # @param [Hash] hash hash of bibitem
+    #
+    # @return [void]
+    #
+    def fix_si_brochure_id(hash) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
+      did = hash["docid"].detect { |id| id["type"] == "BIPM" }
+      did["primary"] = true
+      return unless did["id"] == "BIPM Brochure"
+
+      isbn = hash["docid"].detect { |id| id["type"] == "ISBN" }
+      num = if isbn && isbn["id"] == "978-92-822-2272-0"
+              "SI Brochure"
+            else
+              "SI Brochure, Appendix 4"
+            end
+      hash["id"] = hash["id"].sub(/(?<=^BIPM)Brochure$/i, num.gsub(/[,\s]/, ""))
+      hash["docnumber"] = hash["docnumber"].sub(/^Brochure$/i, num)
+      did["id"] = did["id"].sub(/(?<=^BIPM\s)Brochure$/i, num)
+    end
+
     #
     # Deep merge two hashes
     #
@@ -117,9 +148,9 @@ module RelatonBipm
     def fetch_type(dir, body) # rubocop:disable Metrics/AbcSize
       type = dir.split("/").last.split("-").first.sub(/s$/, "")
       body_dir = File.join @output, body.downcase
-      Dir.mkdir body_dir unless Dir.exist? body_dir
+      FileUtils.mkdir_p body_dir
       outdir = File.join body_dir, type.downcase
-      Dir.mkdir outdir unless Dir.exist? outdir
+      FileUtils.mkdir_p outdir
       Dir[File.join(dir, "*.{yml,yaml}")].each { |en_file| fetch_meeting en_file, body, type, outdir }
     end
 
@@ -228,7 +259,7 @@ module RelatonBipm
         file += "-#{num_justed}" if num.size < 4
         file += ".yaml"
         out_dir = File.join args[:dir], r["type"].downcase
-        Dir.mkdir out_dir unless Dir.exist? out_dir
+        FileUtils.mkdir_p out_dir
         path = File.join out_dir, file
         write_file path, item
         @index[["#{args[:body]} #{type} #{year}-#{num_justed}", "#{args[:body]} #{type} #{args[:num]}-#{num_justed}"]] = path

data/lib/relaton_bipm/index.rb

@@ -28,7 +28,7 @@ module RelatonBipm
     def path
       @path ||= begin
         dir = File.join Dir.home, ".relaton", "bipm"
-        FileUtils.mkdir_p dir unless Dir.exist? dir
+        FileUtils.mkdir_p dir
         File.join dir, "index.yaml"
       end
     end

data/lib/relaton_bipm/version.rb

@@ -1,3 +1,3 @@
 module RelatonBipm
-  VERSION = "1.12.3".freeze
+  VERSION = "1.13.2".freeze
 end

data/lib/relaton_bipm/xml_parser.rb

@@ -23,17 +23,6 @@ module RelatonBipm
         BipmBibliographicItem.new(**item_hash)
       end
 
-      # @param item [Nokogiri::XML::Element]
-      # @return [Array<RelatonBib::FormattedString>]
-      def fetch_titles(item)
-        item.xpath("./title").map do |t|
-          RelatonBib::TypedTitleString.new(
-            content: t.text, language: t[:language], script: t[:script],
-            format: t[:format]
-          )
-        end
-      end
-
       def fetch_dates(item) # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity,Metrics/PerceivedComplexity
         item.xpath("./date").reduce([]) do |a, d|
           type = d[:type].to_s.empty? ? "published" : d[:type]

data/relaton_bipm.gemspec

@@ -42,7 +42,7 @@ Gem::Specification.new do |spec| # rubocop:disable Metrics/BlockLength
 
   spec.add_dependency "faraday", "~> 1.0"
   spec.add_dependency "mechanize", "~> 2.8.0"
-  spec.add_dependency "relaton-bib", "~> 1.12.0"
+  spec.add_dependency "relaton-bib", "~> 1.13.0"
   spec.add_dependency "rubyzip", "~> 2.3.0"
   spec.add_dependency "serrano", "~> 1.0"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: relaton-bipm
 version: !ruby/object:Gem::Version
-  version: 1.12.3
+  version: 1.13.2
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2022-08-02 00:00:00.000000000 Z
+date: 2022-08-31 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: byebug
@@ -142,14 +142,14 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 1.12.0
+        version: 1.13.0
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 1.12.0
+        version: 1.13.0
 - !ruby/object:Gem::Dependency
   name: rubyzip
   requirement: !ruby/object:Gem::Requirement