relaton-bsi 1.7.pre1

data/grammars/reqt.rng

@@ -0,0 +1,165 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<grammar xmlns="http://relaxng.org/ns/structure/1.0" datatypeLibrary="http://www.w3.org/2001/XMLSchema-datatypes">
+  <!--
+    Presupposes isodoc.rnc, is included in it
+    include "isodoc.rnc" { }
+  -->
+  <define name="requirement">
+    <element name="requirement">
+      <ref name="RequirementType"/>
+    </element>
+  </define>
+  <define name="recommendation">
+    <element name="recommendation">
+      <ref name="RequirementType"/>
+    </element>
+  </define>
+  <define name="permission">
+    <element name="permission">
+      <ref name="RequirementType"/>
+    </element>
+  </define>
+  <define name="RequirementType">
+    <optional>
+      <attribute name="obligation">
+        <ref name="ObligationType"/>
+      </attribute>
+    </optional>
+    <optional>
+      <attribute name="unnumbered">
+        <data type="boolean"/>
+      </attribute>
+    </optional>
+    <optional>
+      <attribute name="subsequence"/>
+    </optional>
+    <attribute name="id">
+      <data type="ID"/>
+    </attribute>
+    <optional>
+      <attribute name="filename"/>
+    </optional>
+    <optional>
+      <ref name="reqtitle"/>
+    </optional>
+    <optional>
+      <ref name="label"/>
+    </optional>
+    <optional>
+      <ref name="subject"/>
+    </optional>
+    <optional>
+      <ref name="reqinherit"/>
+    </optional>
+    <zeroOrMore>
+      <ref name="classification"/>
+    </zeroOrMore>
+    <zeroOrMore>
+      <choice>
+        <ref name="measurementtarget"/>
+        <ref name="specification"/>
+        <ref name="verification"/>
+        <ref name="import"/>
+        <ref name="description"/>
+      </choice>
+    </zeroOrMore>
+    <optional>
+      <ref name="reqt_references"/>
+    </optional>
+    <zeroOrMore>
+      <choice>
+        <ref name="requirement"/>
+        <ref name="recommendation"/>
+        <ref name="permission"/>
+      </choice>
+    </zeroOrMore>
+  </define>
+  <define name="reqtitle">
+    <element name="title">
+      <ref name="FormattedString"/>
+    </element>
+  </define>
+  <define name="label">
+    <element name="label">
+      <text/>
+    </element>
+  </define>
+  <define name="subject">
+    <element name="subject">
+      <text/>
+    </element>
+  </define>
+  <define name="reqinherit">
+    <element name="inherit">
+      <text/>
+    </element>
+  </define>
+  <define name="measurementtarget">
+    <element name="measurement-target">
+      <ref name="RequirementSubpart"/>
+    </element>
+  </define>
+  <define name="specification">
+    <element name="specification">
+      <ref name="RequirementSubpart"/>
+    </element>
+  </define>
+  <define name="verification">
+    <element name="verification">
+      <ref name="RequirementSubpart"/>
+    </element>
+  </define>
+  <define name="import">
+    <element name="import">
+      <ref name="RequirementSubpart"/>
+    </element>
+  </define>
+  <define name="description">
+    <element name="description">
+      <ref name="RequirementSubpart"/>
+    </element>
+  </define>
+  <define name="reqt_references">
+    <element name="references">
+      <oneOrMore>
+        <ref name="bibitem"/>
+      </oneOrMore>
+    </element>
+  </define>
+  <define name="RequirementSubpart">
+    <optional>
+      <attribute name="type"/>
+    </optional>
+    <optional>
+      <attribute name="exclude">
+        <data type="boolean"/>
+      </attribute>
+    </optional>
+    <oneOrMore>
+      <ref name="BasicBlock"/>
+    </oneOrMore>
+  </define>
+  <define name="ObligationType">
+    <choice>
+      <value>requirement</value>
+      <value>recommendation</value>
+      <value>permission</value>
+    </choice>
+  </define>
+  <define name="classification">
+    <element name="classification">
+      <ref name="classification_tag"/>
+      <ref name="classification_value"/>
+    </element>
+  </define>
+  <define name="classification_tag">
+    <element name="tag">
+      <text/>
+    </element>
+  </define>
+  <define name="classification_value">
+    <element name="value">
+      <text/>
+    </element>
+  </define>
+</grammar>

data/lib/relaton_bsi.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+require_relative "relaton_bsi/version"
+require "relaton_bsi/bsi_bibliography"
+
+module RelatonBsi
+ # Returns hash of XML reammar
+  # @return [String]
+  def self.grammar_hash
+    gem_path = File.expand_path "..", __dir__
+    grammars_path = File.join gem_path, "grammars", "*"
+    grammars = Dir[grammars_path].sort.map { |gp| File.read gp }.join
+    Digest::MD5.hexdigest grammars
+  end
+end

data/lib/relaton_bsi/bsi_bibliographic_item.rb

@@ -0,0 +1,46 @@
+module RelatonBsi
+  class BsiBibliographicItem < RelatonIsoBib::IsoBibliographicItem
+    TYPES = %w[
+      specification management-systems-standard code-of-practice guide
+      method-of-test method-of-specifying vocabulary classification
+    ].freeze
+
+    # @return [String, nil]
+    attr_reader :price_code
+
+    # @return [Boolean, nil]
+    attr_reader :cen_processing
+
+    # @params price_code [String, nil]
+    # @param cen_processing [Boolean, nil]
+    def initialize(**args)
+      # if args[:doctype] && !TYPES.include?(args[:doctype])
+      #   warn "[relaton-bsi] WARNING: invalid doctype: #{args[:doctype]}"
+      # end
+      @price_code = args.delete :price_code
+      @cen_processing = args.delete :cen_processing
+      super
+    end
+
+    # @param opts [Hash]
+    # @option opts [Nokogiri::XML::Builder] :builder XML builder
+    # @option opts [Boolean] :bibdata
+    # @option opts [String] :lang language
+    # @return [String] XML
+    def to_xml(**opts)
+      super **opts do |b|
+        if opts[:bibdata]
+          b.send "price-code", price_code if price_code
+          b.send "cen-processing", cen_processing unless cen_processing.nil?
+        end
+      end
+    end
+
+    # @param hash [Hash]
+    # @return [RelatonBsi::BsiBibliographicItem]
+    def self.from_hash(hash)
+      item_hash = ::RelatonBsi::HashConverter.hash_to_bib(hash)
+      new **item_hash
+    end
+  end
+end

data/lib/relaton_bsi/bsi_bibliography.rb

@@ -0,0 +1,113 @@
+# frozen_string_literal: true
+
+require "mechanize"
+require "relaton_iso_bib"
+require "relaton_bsi/bsi_bibliographic_item"
+require "relaton_bsi/scrapper"
+require "relaton_bsi/hit_collection"
+require "relaton_bsi/hit"
+require "relaton_bsi/xml_parser"
+require "relaton_bsi/hash_converter"
+
+module RelatonBsi
+  # Class methods for search ISO standards.
+  class BsiBibliography
+    class << self
+      # @param text [String]
+      # @return [RelatonBsi::HitCollection]
+      def search(text, year = nil)
+        code = text.sub(/^BSI\s/, "")
+        HitCollection.new code, year
+      rescue SocketError, Timeout::Error, Errno::EINVAL, Errno::ECONNRESET,
+             EOFError, Net::HTTPBadResponse, Net::HTTPHeaderSyntaxError,
+             Net::ProtocolError
+        raise RelatonBib::RequestError, "Could not access #{HitCollection::DOMAIN}"
+      end
+
+      # @param code [String] the BSI standard Code to look up
+      # @param year [String] the year the standard was published (optional)
+      # @param opts [Hash] options; restricted to :all_parts if all-parts reference is required
+      # @return [String] Relaton XML serialisation of reference
+      def get(code, year = nil, opts = {}) # rubocop:disable Metrics/CyclomaticComplexity,Metrics/MethodLength,Metrics/PerceivedComplexity
+        # if year.nil?
+        #   /^(?<code1>[^\s]+\s[^\s]+)\s\((\d{2}\/)?(?<year1>\d+)\)$/ =~ code
+        #   unless code1.nil?
+        #     code = code1
+        #     year = year1
+        #   end
+        # end
+
+        ret = bib_get1(code, year, opts)
+        return nil if ret.nil?
+
+        # ret = ret.to_most_recent_reference unless year || opts[:keep_year]
+        # ret = ret.to_all_parts if opts[:all_parts]
+        ret
+      end
+
+      private
+
+      def fetch_ref_err(code, year, missed_years) # rubocop:disable Metrics/MethodLength
+        id = year ? "#{code}:#{year}" : code
+        warn "[relaton-bsi] WARNING: no match found online for #{id}. "\
+          "The code must be exactly like it is on the standards website."
+        unless missed_years.empty?
+          warn "[relaton-bsi] (There was no match for #{year}, though there were matches "\
+            "found for #{missed_years.join(', ')}.)"
+        end
+        # if /\d-\d/.match? code
+        #   warn "[relaton-bsi] The provided document part may not exist, or the document "\
+        #     "may no longer be published in parts."
+        # else
+        #   warn "[relaton-bsi] If you wanted to cite all document parts for the reference, "\
+        #     "use \"#{code} (all parts)\".\nIf the document is not a standard, "\
+        #     "use its document type abbreviation (TS, TR, PAS, Guide)."
+        # end
+        nil
+      end
+
+      def search_filter(code) # rubocop:disable Metrics/AbcSize,Metrics/CyclomaticComplexity,Metrics/MethodLength,Metrics/PerceivedComplexity
+        %r{^BSI\s(?<code1>[^:]+)} =~ code
+        warn "[relaton-bsi] (\"#{code}\") fetching..."
+        result = search(code)
+        result.select do |i|
+          next true unless i.hit[:code]
+
+          %r{^(?<code2>[^:]+)} =~ i.hit[:code]
+          code2.include?(code1)
+        end
+      end
+
+      # Sort through the results from Isobib, fetching them three at a time,
+      # and return the first result that matches the code,
+      # matches the year (if provided), and which # has a title (amendments do not).
+      # Only expects the first page of results to be populated.
+      # Does not match corrigenda etc (e.g. ISO 3166-1:2006/Cor 1:2007)
+      # If no match, returns any years which caused mismatch, for error reporting
+      def isobib_results_filter(result, year)
+        missed_years = []
+        result.each do |r|
+          /:(?<pyear>\d{4})/ =~ r.hit[:code]
+          if !year || year == pyear
+            ret = r.fetch
+            return { ret: ret } if ret
+          end
+
+          missed_years << pyear
+        end
+        { years: missed_years }
+      end
+
+      def bib_get1(code, year, _opts)
+        result = search_filter(code) || return
+        ret = isobib_results_filter(result, year)
+        if ret[:ret]
+          warn "[relaton-bsi] (\"#{code}\") found #{ret[:ret].docidentifier.first&.id}"
+          ret[:ret]
+        else
+          fetch_ref_err(code, year, ret[:years])
+        end
+      end
+    end
+  end
+end

data/lib/relaton_bsi/hash_converter.rb

@@ -0,0 +1,16 @@
+module RelatonBsi
+  class HashConverter < RelatonIsoBib::HashConverter
+    class << self
+      private
+
+      #
+      # Ovverides superclass's method
+      #
+      # @param item [Hash]
+      # @retirn [RelatonBsi::BsiBibliographicItem]
+      def bib_item(item)
+        BsiBibliographicItem.new(**item)
+      end
+    end
+  end
+end

data/lib/relaton_bsi/hit.rb

@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+
+module RelatonBsi
+  # Hit.
+  class Hit < RelatonBib::Hit
+    attr_writer :fetch
+
+    # Parse page.
+    # @return [RelatonBsi::BsiBibliographicItem]
+    def fetch
+      @fetch ||= Scrapper.parse_page self
+    end
+  end
+end

data/lib/relaton_bsi/hit_collection.rb

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+require "relaton_bsi/hit"
+
+module RelatonBsi
+  # Page of hit collection.
+  class HitCollection < RelatonBib::HitCollection
+    DOMAIN = "https://shop.bsigroup.com"
+
+    # @return [Mechanize]
+    attr_reader :agent
+
+    # @param ref [String]
+    # @param year [String]
+    def initialize(ref, year = nil)
+      super ref, year
+      @agent = Mechanize.new
+      resp = agent.get "#{DOMAIN}/SearchResults/?q=#{ref}"
+      @array = hits resp
+    end
+
+    private
+
+    # @param resp [Mechanize::Page]
+    # @return [Array<RelatonBsi::Hit>]
+    def hits(resp)
+      resp.xpath("//div[@class='resultsInd']").map do |h|
+        ref = h.at('div/h2/a')
+        code = ref.text.strip
+        url = ref[:href]
+        d = h.at("//div/strong[.='Published Date:']/following-sibling::strong").text
+        date = Date.parse(d)
+        Hit.new({ code: code, url: url, date: date }, self)
+      end
+    end
+  end
+end

data/lib/relaton_bsi/processor.rb

@@ -0,0 +1,38 @@
+require "relaton/processor"
+
+module RelatonBsi
+  class Processor < Relaton::Processor
+    def initialize
+      @short = :relaton_bsi
+      @prefix = "BSI"
+      @defaultprefix = %r{^BSI\s}
+      @idtype = "BSI"
+    end
+
+    # @param code [String]
+    # @param date [String, NilClass] year
+    # @param opts [Hash]
+    # @return [RelatonBsi::BsiBibliographicItem]
+    def get(code, date, opts)
+      ::RelatonBsi::BsiBibliography.get(code, date, opts)
+    end
+
+    # @param xml [String]
+    # @return [RelatonBsi::BsiBibliographicItem]
+    def from_xml(xml)
+      ::RelatonBsi::XMLParser.from_xml xml
+    end
+
+    # @param hash [Hash]
+    # @return [RelatonBsi::BsiBibliographicItem]
+    def hash_to_bib(hash)
+      ::RelatonBsi::BsiBibliographicItem.from_hash hash
+    end
+
+    # Returns hash of XML grammar
+    # @return [String]
+    def grammar_hash
+      @grammar_hash ||= ::RelatonBsi.grammar_hash
+    end
+  end
+end