relaton-itu 1.7.1 → 1.7.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 6abecc5b89b11a2e53d64fa08fec21a51861df0799a24ab46193f0563e53ca9e
-  data.tar.gz: 53f0d275a4bfab0d597f8ae91fcd505f70d5be44df743b5781a6a341a6fac23a
+  metadata.gz: 994e4bcf33070a5581017d2572048c0d0c85cebc9818d44b4cb679051c3edb55
+  data.tar.gz: 367230b05352067540dce9b8dbbc8e977d31358aca4f39aa96ac529742e553e6
 SHA512:
-  metadata.gz: aa42bec1e3c06656671893b4e08dfa523171a58ef9734d2c551413350e5572d60173d0e276f31a2630f22b6cfc2f6d8a3bea11105057fc903835a98cdf666eac
-  data.tar.gz: 2f997f7732fa074aa30b03c490504a501361310973c8c1e9e8e9b16843033a34dc6393c51d635f321f1009a49ef55b261bc9edb6a423b29feea227b02c5cd75b
+  metadata.gz: 679daec50ec3f9b60e1cefafff737ed644cc659fedb805550cc56c84670b3c3cb30fa327627c5728e4aa95e4b158fa7ede13612df4ef87d99c026722c2b2a5cb
+  data.tar.gz: '08f71b448a574f3088d635ef274f0316faddfec4d3097337144f46f3ec1918e61c056402bc26c660227376c9dc680fc11d3bb5903aadc433db796a4ff88a7098'

data/README.adoc

@@ -87,6 +87,16 @@ RelatonItu::ItuBibliography.get("ITU-T L.163", "2018", {})
 ...
 ----
 
+=== Get amendment
+[source,ruby]
+----
+RelatonItu::ItuBibliography.get "ITU-T G.989.2/Amd 1"
+[relaton-itu] ("ITU-T G.989.2/Amd 1") fetching...
+[relaton-itu] ("ITU-T G.989.2/Amd 1") found ITU-T G.989.2/Amd 1
+=> #<RelatonItu::ItuBibliographicItem:0x007fe5409e5840
+...
+----
+
 === Create bibliographic item form YAML
 [source,ruby]
 ----

data/lib/relaton_itu/itu_bibliography.rb

@@ -67,14 +67,35 @@ module RelatonItu
         nil
       end
 
-      def search_filter(code)
-        docidrx = %r{\w+\.\d+|\w\sSuppl\.\s\d+} # %r{^ITU-T\s[^\s]+}
-        c = code.sub(/Imp\s?/, "").match(docidrx).to_s
+      def search_filter(code, year)
+        %r{
+          ^(?<pref1>ITU)?(-(?<type1>\w))?\s?(?<code1>[^\s\/]+)
+          (\s\(((?<month1>\d{2})\/)?(?<year1>\d{4})\))?
+          (\s-\s(?<buldate1>\d{2}\.\w{1,4}\.\d{4}))?
+          (\/(?<corr1>(Amd|Cor)\s?\d+))?
+          (\s\(((?<cormonth1>\d{2})\/)?(?<coryear1>\d{4})\))?
+        }x =~ code
+        year ||= year1
+        # docidrx = %r{\w+\.\d+|\w\sSuppl\.\s\d+} # %r{^ITU-T\s[^\s]+}
+        # c = code.sub(/Imp\s?/, "").match(docidrx).to_s
         warn "[relaton-itu] (\"#{code}\") fetching..."
         result = search(code)
+        code1.sub! /(?<=\.)Imp(?=\d)/, "" if result.gi_imp
         result.select do |i|
-          i.hit[:code] &&
-            i.hit[:code].match(docidrx).to_s == c
+          %r{
+            ^(?<pref2>ITU)?(-(?<type2>\w))?\s?(?<code2>[\S]+)
+            (\s\(((?<month2>\d{2})\/)?(?<year2>\d{4})\))?
+            (\s(?<corr2>(Amd|Cor)\.\s?\d+))?
+            (\s\(((?<cormonth2>\d{2})\/)?(?<coryear2>\d{4})\))?
+          }x =~ i.hit[:code]
+          /:[^\(]+\((?<buldate2>\d{2}\.\w{1,4}\.\d{4})\)/ =~ i.hit[:title]
+          corr2&.sub! /\.\s?/, " "
+          pref1 == pref2 && (!type1 || type1 == type2) && code1 == code2 &&
+            (!year || year == year2) && (!month1 || month1 == month2) &&
+            corr1 == corr2 && (!coryear1 || coryear1 == coryear2) &&
+            buldate1 == buldate2 && (!cormonth1 || cormonth1 == cormonth2)
+          # i.hit[:code] &&
+          #   i.hit[:code].match(docidrx).to_s == c
         end
       end
 
@@ -86,24 +107,19 @@ module RelatonItu
       # If no match, returns any years which caused mismatch, for error reporting
       def isobib_results_filter(result, year)
         missed_years = []
-        # result.each_slice(3) do |s| # ISO website only allows 3 connections
-        #   fetch_pages(s, 3).each do |r|
         result.each do |r|
           return { ret: r.fetch } if !year
 
           /\(\d{2}\/(?<pyear>\d{4})\)/ =~ r.hit[:code]
-          # r.date.select { |d| d.type == "published" }.each do |d|
           return { ret: r.fetch } if year == pyear
 
           missed_years << pyear
-          # end
-          # end
         end
         { years: missed_years }
       end
 
       def itubib_get1(code, year, _opts)
-        result = search_filter(code) || return
+        result = search_filter(code, year) || return
         ret = isobib_results_filter(result, year)
         if ret[:ret]
           warn "[relaton-itu] (\"#{code}\") found #{ret[:ret].docidentifier.first&.id}"

data/lib/relaton_itu/scrapper.rb

@@ -44,7 +44,7 @@ module RelatonItu
         ItuBibliographicItem.new(
           fetched: Date.today.to_s,
           type: "standard",
-          docid: fetch_docid(doc),
+          docid: fetch_docid(doc, hit_data[:title]),
           edition: edition,
           language: ["en"],
           script: ["Latn"],
@@ -105,18 +105,32 @@ module RelatonItu
 
       # Fetch docid.
       # @param doc [Nokogiri::HTML::Document]
+      # @param title [String]
       # @return [Hash]
-      def fetch_docid(doc)
-        doc.xpath(
+      def fetch_docid(doc, title)
+        docids = 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 do |code|
-          id = code.text.match(%r{^.*?(?= \()|\w\.Imp\s?\d+}).to_s.squeeze(" ")
-          type = id.match(%r{^\w+}).to_s
-          type = "ITU" if type == "G"
-          RelatonBib::DocumentIdentifier.new(type: type, id: id)
-        end
+          "//div/table[1]/tr[4]/td/strong"
+        ).map { |c| createdocid c.text }
+        docids << createdocid(title) unless docids.any?
+        docids
+      end
+
+      def createdocid(text)
+        %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"
+        RelatonBib::DocumentIdentifier.new(type: type, id: id)
       end
 
       # Fetch status.

data/lib/relaton_itu/version.rb

@@ -1,3 +1,3 @@
 module RelatonItu
-  VERSION = "1.7.1".freeze
+  VERSION = "1.7.2".freeze
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: relaton-itu
 version: !ruby/object:Gem::Version
-  version: 1.7.1
+  version: 1.7.2
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2020-12-03 00:00:00.000000000 Z
+date: 2020-12-08 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: debase