pubid-nist 0.1.8

data/lib/pubid/nist/parsers/nbs_circ.rb

@@ -0,0 +1,21 @@
+module Pubid::Nist
+  module Parsers
+    class NbsCirc < Default
+      rule(:revision) do
+        ((str("rev") >> (letters >> year_digits).as(:revision)) |
+          (str("r") >> digits.as(:revision))
+        )
+      end
+
+      rule(:report_number) do
+        first_report_number >> edition.maybe >> (str("-") >> second_report_number).maybe
+      end
+
+      rule(:edition) do
+        (str("sup") >> str("").as(:supplement) >>
+          (letters.as(:edition_month) >> year_digits.as(:edition_year))) |
+          ((str("e") | str("-")) >> (digits.as(:edition) | letters.as(:edition_month) >> year_digits.as(:edition_year)))
+      end
+    end
+  end
+end

data/lib/pubid/nist/parsers/nbs_crpl.rb

@@ -0,0 +1,17 @@
+module Pubid::Nist
+  module Parsers
+    class NbsCrpl < Default
+      rule(:first_report_number) do
+        (digits >> str("-m").maybe).as(:first_report_number)
+      end
+
+      rule(:part) do
+        str("_") >> (digits >> str("-") >> digits).as(:part)
+      end
+
+      rule(:supplement) do
+        str("A").as(:supplement)
+      end
+    end
+  end
+end

data/lib/pubid/nist/parsers/nbs_csm.rb

@@ -0,0 +1,11 @@
+module Pubid::Nist
+  module Parsers
+    class NbsCsm < Default
+      rule(:identifier) do
+        (str(" ") | str(".")) >> report_number.maybe >> parts.repeat.as(:parts)
+      end
+
+      rule(:part_prefixes) { str("n") }
+    end
+  end
+end

data/lib/pubid/nist/parsers/nbs_fips.rb

@@ -0,0 +1,14 @@
+module Pubid::Nist
+  module Parsers
+    class NbsFips < Default
+      rule(:edition) do
+        str("-") >> (
+          (month_letters.as(:edition_month) >> year_digits.as(:edition_year)) |
+            year_digits.as(:edition_year) |
+          (month_letters.as(:edition_month) >> match('\d').repeat(2, 2).as(:edition_day) >>
+            str("/") >> year_digits.as(:edition_year))
+        )
+      end
+    end
+  end
+end

data/lib/pubid/nist/parsers/nbs_hb.rb

@@ -0,0 +1,31 @@
+module Pubid::Nist
+  module Parsers
+    class NbsHb < Default
+      # found patterns:
+      # 44e2-1955 -> 44e2
+      # 146v1-1991
+      # 105-1-1990 -> 105-1e1990
+      # 111r1977 / 146v1
+      # 130-1979 -> 130e1979
+      # 105-8 -> 105-8
+      # 28supp1957pt1
+      # 67suppFeb1965
+
+      rule(:edition) do
+        (str("supp") >> str("").as(:supplement) >>
+          (letters.as(:edition_month) >> year_digits.as(:edition_year)) |
+          str("e") >> year_digits.as(:edition_year)
+        )
+      end
+
+      rule(:report_number) do
+        digits.as(:first_report_number) >> volume.maybe >>
+          (str("e") >> digits.as(:edition) >> (str("-") >> digits).maybe |
+            str("-") >> year_digits.as(:edition_year) |
+            str("-") >> digits.as(:second_report_number) >> str("-") >>  year_digits.as(:edition_year) |
+            str("-") >> digits_with_suffix.as(:second_report_number)
+          ).maybe
+      end
+    end
+  end
+end

data/lib/pubid/nist/parsers/nbs_ir.rb

@@ -0,0 +1,29 @@
+module Pubid::Nist
+  module Parsers
+    class NbsIr < Default
+      rule(:second_report_number) do
+        (match('[\d.]').repeat(1) >> number_suffix.maybe).as(:second_report_number)
+      end
+
+      rule(:report_number) do
+        (first_report_number >>
+          (str("-") >> second_report_number).maybe
+        ).capture(:report_number) >>
+          # parse last part as volume for specific document numbers
+          dynamic do |_source, context|
+            report_number = context.captures[:report_number]
+              .values_at(:first_report_number, :second_report_number).join("-")
+            if %w[74-577 77-1420].include?(report_number)
+              str("-") >> match('\d').as(:volume)
+            else
+              self
+            end
+          end.maybe
+      end
+
+      rule(:part_prefixes) do
+        str("pt") | str("p") | str("-")
+      end
+    end
+  end
+end

data/lib/pubid/nist/parsers/nbs_lc.rb

@@ -0,0 +1,21 @@
+module Pubid::Nist
+  module Parsers
+    class NbsLc < Default
+      rule(:supplement) do
+        (str("supp") | str("sup")) >>
+          ((str("").as(:supplement) >> digits.as(:update_month) >> str("/") >> digits.as(:update_year)) |
+          match('\d').repeat.as(:supplement))
+      end
+
+      rule(:revision) do
+        str("r") >>
+          ((digits.as(:update_month) >> str("/") >> digits.as(:update_year)) |
+            digits.as(:revision))
+      end
+
+      # suffixes for LCIRC 378
+      rule(:number_suffix) { match("[abcdefghA-Z]") }
+    end
+  end
+end
+

data/lib/pubid/nist/parsers/nbs_mn.rb

@@ -0,0 +1,6 @@
+module Pubid::Nist
+  module Parsers
+    class NbsMn < Default
+    end
+  end
+end

data/lib/pubid/nist/parsers/nbs_mp.rb

@@ -0,0 +1,9 @@
+module Pubid::Nist
+  module Parsers
+    class NbsMp < Default
+      rule(:edition) do
+        (str("e") >> digits.as(:edition)) | (str("(") >> digits.as(:edition) >> str(")"))
+      end
+    end
+  end
+end

data/lib/pubid/nist/parsers/nbs_rpt.rb

@@ -0,0 +1,13 @@
+module Pubid::Nist
+  module Parsers
+    class NbsRpt < Default
+      rule(:report_number) do
+        (month_letters >>
+          str("-") >> (month_letters >> year_digits)).as(:report_number) |
+          (digits >> str("-") >> (digits | str("A"))).as(:report_number) |
+          (digits >> (str("a") | str("b")).maybe).as(:report_number) |
+          (str("ADHOC") | str("div9")).as(:report_number)
+      end
+    end
+  end
+end

data/lib/pubid/nist/parsers/nbs_sp.rb

@@ -0,0 +1,13 @@
+module Pubid::Nist
+  module Parsers
+    class NbsSp < Default
+      rule(:part_prefixes) do
+        str("p") | str("P")
+      end
+
+      rule(:volume) do
+        str("v") >> (match('[\da-z-]').repeat(1) >> match("[A-Z]").repeat).as(:volume)
+      end
+    end
+  end
+end

data/lib/pubid/nist/parsers/nbs_tn.rb

@@ -0,0 +1,9 @@
+module Pubid::Nist
+  module Parsers
+    class NbsTn < Default
+      rule(:second_report_number) do
+        (digits_with_suffix | str("A")).as(:second_report_number)
+      end
+    end
+  end
+end

data/lib/pubid/nist/parsers/nist_gcr.rb

@@ -0,0 +1,14 @@
+module Pubid::Nist
+  module Parsers
+    class NistGcr < Default
+      rule(:report_number) do
+        (digits >>
+          str("-") >> digits >> (str("-") >> digits).maybe).as(:report_number)
+      end
+
+      rule(:volume) do
+        str("v") >> (digits >> match('[A-Z]').repeat).as(:volume)
+      end
+    end
+  end
+end

data/lib/pubid/nist/parsers/nist_hb.rb

@@ -0,0 +1,8 @@
+require_relative "nbs_hb"
+
+module Pubid::Nist
+  module Parsers
+    class NistHb < NbsHb
+    end
+  end
+end

data/lib/pubid/nist/parsers/nist_ir.rb

@@ -0,0 +1,29 @@
+module Pubid::Nist
+  module Parsers
+    class NistIr < Default
+      rule(:number_suffix) { match("[abcA-Z]") }
+
+      rule(:revision) do
+        str("r") >> (digits | month_letters >> year_digits).as(:revision)
+      end
+
+      rule(:report_number) do
+        (year_digits.as(:first_report_number) >>
+          str("-") >> year_digits.as(:edition_year)) |
+          first_report_number >> (str("-") >>
+            ((digits | match("[aAB]") | str("CAS") | str("FRA")) >>
+              # for extra number for NIST IR 85-3273-10
+              (str("-") >> digits).maybe).as(:second_report_number)).maybe
+      end
+
+      rule(:revision) do
+        str("r") >>
+          ((digits.as(:update_month) >> str("/") >> digits.as(:update_year)) |
+            (month_letters.as(:update_month) >> year_digits.as(:update_year)) |
+            digits.as(:revision) |
+            str("").as(:revision))
+      end
+    end
+  end
+end
+

data/lib/pubid/nist/parsers/nist_ncstar.rb

@@ -0,0 +1,7 @@
+module Pubid::Nist
+  module Parsers
+    class NistNcstar < Default
+      rule(:number_suffix) { match("[abcdefghijA-Z]") }
+    end
+  end
+end

data/lib/pubid/nist/parsers/nist_owmwp.rb

@@ -0,0 +1,11 @@
+module Pubid::Nist
+  module Parsers
+    class NistOwmwp < Default
+      rule(:report_number) do
+        digits_with_suffix.as(:first_report_number) >>
+          (str("-") >> (digits_with_suffix >> (str("-") >> digits_with_suffix).maybe)
+                         .as(:second_report_number)).maybe
+      end
+    end
+  end
+end

data/lib/pubid/nist/parsers/nist_sp.rb

@@ -0,0 +1,53 @@
+module Pubid::Nist
+  module Parsers
+    class NistSp < Default
+      rule(:version) do
+        ((str("ver") >> (digits >> (str(".") >> digits).maybe).as(:version)) |
+          (str("v") >>
+            (match('\d') >> str(".") >> match('\d') >> (str(".") >> match('\d')).maybe).as(:version)))
+      end
+
+      rule(:number_suffix) { match["[A-Zabcd]"] }
+
+      rule(:first_report_number) do
+        digits >> (str("GB") | str("a")).maybe
+      end
+
+      rule(:report_number) do
+        # (\d+-\d{1,3}).as(:report_number)
+        # (\d+-\d{4}).as(:report_number) when first number is 250
+        # (\d+).as(:report_number)-(\d{4}).as(:edition)
+        # or \d-\d-(\d).as(:revision)
+        (first_report_number.capture(:first_number).as(:first_report_number) >>
+          (str("-") >>
+            dynamic do |_source, context|
+              # consume 4 numbers or any amount of numbers
+              # for document ids starting from 250
+              (if context.captures[:first_number] == "250"
+                 digits
+               else
+                 # do not consume edition numbers (parse only if have not 4 numbers)
+                 year_digits.absent? >> digits
+               end >> number_suffix.maybe
+                # parse last number as edition if have 4 numbers
+              ) | (str("NCNR") | str("PERMIS") | str("BFRL"))
+            end.as(:second_report_number)
+            # parse A-Z and abcd as part of report number
+          ).maybe
+        )
+      end
+
+      rule(:edition) do
+        ((str("e") >> year_digits.as(:edition_year)) | (str("-") >> year_digits.as(:edition_year)) |
+          (str("e") >> digits.as(:edition)))
+      end
+
+      rule(:revision) do
+        ((str("rev") | str("r")) >> (digits >> match("[a-z]").maybe).as(:revision)) |
+          (str("-") >> digits.as(:revision)) |
+          (str("r") >> match("[a-z]").as(:revision)) |
+          (str("r") >> str("").as(:revision))
+      end
+    end
+  end
+end

data/lib/pubid/nist/parsers/nist_tn.rb

@@ -0,0 +1,13 @@
+module Pubid::Nist
+  module Parsers
+    class NistTn < Default
+      rule(:report_number) do
+        first_report_number
+      end
+
+      rule(:edition_prefixes) do
+        str("-") | str("e")
+      end
+    end
+  end
+end

data/lib/pubid/nist/publisher.rb

@@ -0,0 +1,43 @@
+PUBLISHERS = YAML.load_file(File.join(File.dirname(__FILE__),
+                                      "../../../publishers.yaml"))
+
+module Pubid::Nist
+  class Publisher
+    attr_accessor :publisher
+
+    def initialize(publisher:)
+      @publisher = publisher
+    end
+
+    def to_s(format = :short)
+      return @publisher if %i[short mr].include?(format)
+
+      PUBLISHERS[format.to_s][@publisher]
+    end
+
+    def ==(other)
+      other.publisher == @publisher
+    end
+
+    def self.publishers_keys
+      PUBLISHERS["long"].keys
+    end
+
+    def self.parse(code)
+      publisher = /(#{PUBLISHERS["long"].keys.join('|')})/.match(code)
+      return new(publisher: publisher.to_s) if publisher
+
+      publisher = /(#{PUBLISHERS["long"].values.join('|')})(?=\.|\s)/.match(code)
+      return new(publisher: PUBLISHERS["long"].key(publisher.to_s)) if publisher
+
+      publisher = /(#{PUBLISHERS["abbrev"].values.join('|')})(?=\.|\s)/.match(code)
+      return new(publisher: PUBLISHERS["abbrev"].key(publisher.to_s)) if publisher
+
+      new(publisher: "NIST")
+    end
+
+    def self.regexp
+      /(#{PUBLISHERS["long"].keys.join('|')})(?=\.|\s)/
+    end
+  end
+end

data/lib/pubid/nist/serie.rb

@@ -0,0 +1,27 @@
+SERIES = YAML.load_file(File.join(File.dirname(__FILE__), "../../../series.yaml"))
+
+module Pubid::Nist
+  class Serie
+    attr_accessor :serie, :parsed
+
+    def initialize(serie:, parsed: nil)
+      @serie = serie
+      @parsed = parsed
+    end
+
+    def to_s(format = :short)
+      return @serie if format == :short
+
+      result = SERIES[format.to_s][@serie]
+      return @serie if result.nil? && format == :mr
+
+      return SERIES["long"][@serie] if result.nil?
+
+      result
+    end
+
+    def ==(other)
+      other.serie == @serie
+    end
+  end
+end

data/lib/pubid/nist/stage.rb

@@ -0,0 +1,28 @@
+STAGES = YAML.load_file(File.join(File.dirname(__FILE__), "../../../stages.yaml"))
+
+module Pubid::Nist
+  class Stage
+    attr_accessor :stage
+
+    def initialize(stage)
+      @stage = stage
+    end
+
+    def to_s(format = :short)
+      return "" if nil?
+
+      case format
+      when :short
+        "(#{@stage})"
+      when :mr
+        @stage
+      else
+        STAGES[@stage]
+      end
+    end
+
+    def nil?
+      @stage.nil?
+    end
+  end
+end

data/lib/pubid/nist/version.rb

@@ -0,0 +1,5 @@
+module Pubid
+  module Nist
+    VERSION = "0.1.8".freeze
+  end
+end

data/lib/pubid/nist.rb

@@ -0,0 +1,33 @@
+# frozen_string_literal: true
+
+require "yaml"
+require "parslet"
+
+module Pubid
+  module Nist
+
+  end
+end
+
+require_relative "nist/serie"
+require_relative "nist/parsers/default"
+require_relative "nist/document_transform"
+require_relative "nist/document_parser"
+
+Dir[File.join(__dir__, 'nist/parsers', '*.rb')].each do |file|
+  require file
+end
+
+PARSERS_CLASSES = Pubid::Nist::Parsers.constants.select do |c|
+  Pubid::Nist::Parsers.const_get(c).is_a?(Class)
+end.map do |parser_class|
+  parser = Pubid::Nist::Parsers.const_get(parser_class)
+  [parser.name.split("::").last.gsub(/(.)([A-Z])/, '\1 \2').upcase, parser]
+end.to_h
+
+require_relative "nist/document"
+require_relative "nist/publisher"
+require_relative "nist/stage"
+require_relative "nist/errors"
+require_relative "nist/nist_tech_pubs"
+require_relative "nist/edition"

data/lib/pubid-nist.rb

@@ -0,0 +1,3 @@
+# frozen_string_literal: true
+
+require_relative "pubid/nist"

data/publishers.yaml

@@ -0,0 +1,6 @@
+long:
+  NIST: National Institute of Standards and Technology
+  NBS: National Bureau of Standards
+abbrev:
+  NIST: Natl. Inst. Stand. Technol.
+  NBS: Natl. Bur. Stand.

data/series.yaml

@@ -0,0 +1,139 @@
+long:
+  NIST AMS: Advanced Manufacturing Standard
+  NIST BSS: Building Science Series
+  NBS BSS: Building Science Series
+  NBS BMS: Building Material Structures Report
+  NBS BRPD-CRPL-D: Basic Radio Propagation Predictions Series
+  NBS BH: Building and Housing Reports
+  NBS CRPL: Central Radio Propagation Laboratory Reports
+  NBS CRPL-F-A: CRPL Ionospheric Data
+  NBS CRPL-F-B: CRPL Solar-Geophysical Data
+  NBS IP: CRPL Ionospheric Predictions
+  NBS CIRC: Circulars
+  NBS CIS: Consumer Information Series
+  NBS CS: Commercial Standards
+  NBS CS-E: Commercial Standards (emergency)
+  NBS CSM: Commercial Standards Monthly
+  NBS FIPS: Federal Information Processing Standards Publication
+  FIPS PUB: Federal Information Processing Standards Publication
+  NIST GCR: Grant/Contract Reports
+  NBS GCR: Grant/Contract Reports
+  NIST HB: Handbook
+  NBS HB: Handbook
+  NBS HR: Hydraulic Research in the United States
+  NBS IRPL: Interservice Radio Propagation Laboratory
+  ITL Bulletin: ITL Bulletin
+  NIST LC: Letter Circular
+  NBS LC: Letter Circular
+  NIST MN: Monograph
+  NBS MN: Monograph
+  NBS MP: Miscellaneous Publications
+  NIST NCSTAR: National Construction Safety Team Report
+  NIST NSRDS: National Standard Reference Data Series
+  NSRDS-NBS: National Standard Reference Data Series
+  NIST IR: Interagency or Internal Report
+  NBS IR: Interagency or Internal Report
+  NIST OWMWP: Office of Weights and Measures White Papers
+  NBS PC: Photographic Circulars
+  NBS RPT: Reports
+  NIST PS: Voluntary Product Standards
+  NBS SIBS: Special Interior Ballistics Studies
+  NBS PS: Voluntary Product Standards
+  NIST SP: Special Publication
+  NBS SP: Special Publication
+  NIST TN: Technical Note
+  NBS TN: Technical Note
+  NBS TIBM: Technical Information on Building Materials
+  NIST TTB: Technology Transfer Brief
+  NIST DCI: Data Collection Instruments
+  NIST EAB: Economic Analysis Brief
+  NIST Other: Other
+  CSRC White Paper: Cybersecurity Resource Center White Paper
+  CSRC Book: Cybersecurity Resource Center Book
+  CSRC Use Case: Cybersecurity Resource Center Use Case
+  CSRC Building Block: Cybersecurity Resource Center Building Block
+  JPCRD: Journal of Physical and Chemical Reference Data
+  JRES: Journal of Research of NIST
+abbrev:
+  NIST AMS: Adv. Man. Ser
+  NIST BSS: Bldg. Sci. Ser.
+  NBS BSS: Bldg. Sci. Ser.
+  NBS FIPS: Federal Inf. Process. Stds.
+  FIPS PUB: Federal Inf. Process. Stds.
+  NIST HB: Handb.
+  NBS HB: Handb.
+  NIST MN: Monogr.
+  NBS MN: Monogr.
+  NIST NCSTAR: Natl. Constr. Tm. Act Rpt.
+  NIST NSRDS: Natl. Stand. Ret. Data Ser.
+  NSRDS-NBS: Natl. Stand. Ret. Data Ser.
+  NIST PS: Prod. Stand.
+  NBS PS: Prod. Stand.
+  NIST SP: Spec. Publ.
+  NBS SP: Spec. Publ.
+  NIST TN: Tech. Note
+  NBS TN: Tech. Note
+  NIST DCI: Data Collect. Instr.
+  NIST Other: Other
+  CSRC White Paper: CSWP
+  CSRC Book: CSRC Book
+  CSRC Use Case: CSRC Use Case
+  CSRC Building Block: CSRC Building Block
+  JPCRD: J. Phys. & Chem. Ref. Data
+  JRES: J. Res. Natl. Inst. Stan.
+mr:
+  NIST AMS: NIST.AMS
+  NIST BSS: NIST.BSS
+  NBS BSS: NBS.BSS
+  NBS BMS: NBS.BMS
+  NBS BRPD-CRPL-D: NBS.BRPD-CRPL-D
+  NBS BH: NBS.BH
+  NBS CRPL: NBS.CRPL
+  NBS CRPL-F-A: NBS.CRPL-F-A
+  NBS CRPL-F-B: NBS CRPL-F-B
+  NBS IP: NBS.IP
+  NBS CIRC: NBS.CIRC
+  NBS CIS: NBS.CIS
+  NBS CS: NBS.CS
+  NBS CS-E: NBS.CS-E
+  NBS CSM: NBS.CSM
+  NBS FIPS: NBS.FIPS
+  FIPS PUB: NIST.FIPS
+  NIST GCR: NIST.GCR
+  NBS GCR: NBS.GCR
+  NIST HB: NIST.HB
+  NBS HB: NBS.HB
+  NBS HR: NBS.HR
+  NBS IRPL: NBS.IRPL
+  ITL Bulletin: NIST.ITLB
+  NIST LC: NIST.LC
+  NBS LC: NBS.LC
+  NIST MN: NIST.MN
+  NBS MN: NBS.MN
+  NBS MP: NBS.MP
+  NIST NCSTAR: NIST.NCSTAR
+  NIST NSRDS: NIST.NSRDS
+  NSRDS-NBS: NBS.NSRDS
+  NIST IR: NIST.IR
+  NBS IR: NBS.IR
+  NIST OWMWP: NIST.OWMWP
+  NBS PC: NBS.PC
+  NBS RPT: NBS.RPT
+  NIST PS: NIST.PS
+  NBS SIBS: NBS.SIBS
+  NBS PS: NBS.PS
+  NIST SP: NIST.SP
+  NBS SP: NBS.SP
+  NIST TN: NIST.TN
+  NBS TN: NBS.TN
+  NBS TIBM: NBS.TIBM
+  NIST TTB: NIST.TTB
+  NIST DCI: NIST.DCI
+  NIST EAB: NIST.EAB
+  NIST Other: NIST.O
+  CSRC White Paper: NIST.CSWP
+  CSRC Book: NIST.CSB
+  CSRC Use Case: NIST.CSUC
+  CSRC Building Block: NIST.CSBB
+  JPCRD: JPCRD
+  JRES: NIST.JRES

data/stages.yaml

@@ -0,0 +1,5 @@
+IPD: Initial Public Draft
+2PD: Second Public Draft
+FPD: Final Public Draft
+WD: Work-in-Progress Draft
+PRD: Preliminary Draft