ddr-models 3.0.0.rc1 → 3.0.0.rc2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 3e6b6dc994859b545193b3ab914a59a97d465a9a
-  data.tar.gz: 3bcd281d06aa80c0fa1eb970d809e55eae905d26
+  metadata.gz: 7de034ef99997f329a03eca2b79eee78a8a6ae12
+  data.tar.gz: 3d0f8694d937b5b6a6bfbf94f0889af0520c3ecb
 SHA512:
-  metadata.gz: d69f123b6661631477dd7d6722ef911bde491831d38ed0f4f14eabd7b33649622e517353d8445a101278018fad45ea014cb1bbe2134dba5c961f06fab5d86550
-  data.tar.gz: 73050a50d9221d5ec777183b319862a64b2a724b24caf4d22c72f6cca011d6f43b8fd8241479772b6d4abe027503f1658abf2aa00f6ea191f76b2fedeb2b4fe6
+  metadata.gz: 7d279fe99f40579c4409eaf63e89213c42c644f6836c50a56c2701602be81a86045ade82d993adb5bcc4f08511a1d66859d72dacfca41f6b1c62d7e5e5e6052b
+  data.tar.gz: db68f4040d5ac62d39a71bc7f7630ea297d2b07038812672a57273f7299caf3f05db2d31f5972362ed1cef74e006e079add3f166e9199b59fb7dd0dfa18dad8a

data/ddr-models.gemspec

@@ -34,6 +34,7 @@ Gem::Specification.new do |s|
   s.add_dependency "ddr-antivirus", "~> 2.1.1"
   s.add_dependency "virtus", "~> 1.0.5"
   s.add_dependency "hashie", "~> 3.4.3"
+  s.add_dependency "edtf", "~> 2.3"
 
   s.add_development_dependency "bundler", "~> 1.11", ">= 1.11.2"
   s.add_development_dependency "rake"

data/lib/ddr/models/version.rb

@@ -1,5 +1,5 @@
 module Ddr
   module Models
-    VERSION = "3.0.0.rc1"
+    VERSION = "3.0.0.rc2"
   end
 end

data/lib/ddr/models/year_facet.rb

@@ -1,153 +1,95 @@
 require "date"
+require "edtf"
 
 module Ddr::Models
   class YearFacet
 
     EARLIEST_YEAR = 1000
+    LATEST_YEAR = Date.today.year + 100
+    VALID_YEARS = (EARLIEST_YEAR..LATEST_YEAR)
+    VALUE_SEP = /;/
 
     # Between 1965 and 1968
     BETWEEN = Regexp.new '\A([Bb]etween\s+)(\d{4})(\s+and\s+)(\d{4})\??\z'
 
-    # YYYx (192x)
-    # YYYX (192X)
-    # YYY? (192?)
-    # YYY- (192-)
-    # YYY-? (192-?)
-    IN_DECADE = Regexp.new '\A(\d{3})([xX\-]\??|\?)\z'
-
-    # YYxx (19xx)
-    IN_CENTURY = Regexp.new '\A(\d{2})xx\z'
-
-    # YYY0s (1920s)
-    # YYY0s? (1920s?)
-    DECADE = Regexp.new '\A(\d{3}0)s\??\z'
-
-    # YYYY-MM (2010-01)
-    # YYYY/MM (2010/01)
-    YEAR_MONTH = Regexp.new '\A(\d{4})[/-](0[1-9]|1[0-2])\z'
-
-    # YYYY-YYYY (1935-2010)
-    # YYYY/YYYY (1935/2010)
-    YEAR_RANGE = Regexp.new '\A(\d{4})[/-](\d{4})\z'
-
-    # YYYY (1979)
-    YEAR = Regexp.new '\A\d{4}\z'
-
-    SQUARE_BRACKETS = Regexp.new '[\[\]]'
-
-    # c. 1920
-    # ca. 1920
-    # c1920
+    # circa 1920, ca. 1920, c1920 => 1920
     CIRCA = Regexp.new '\b(circa\s+|ca?\.\s*|c(?=\d{4}[^\d]*))'
 
-    class << self
-      def call(obj)
-        new(obj).values
-      end
-    end
-
-    attr_reader :obj, :values
-
-    def initialize(obj)
-      @obj = obj
-      @values = []
-      facet_values
-    end
-
-    def facet_values
-      obj.desc_metadata.date.each do |date|
-        date.split(/;/).each do |value|
-          clean! value
-          years = extract_years(value)
-          validate! years
-          values.push *years
-        end
-      end
-    end
-
-    def extract_years(value)
-      years = match_years(value) || parse_year(value)
-      Array(years)
-    end
-
-    def clean!(value)
-      value.strip!
-      value.gsub! SQUARE_BRACKETS, ""
-      value.gsub! CIRCA, ""
-    end
+    # 1935-1940 => 1935/1940
+    YEAR_RANGE = Regexp.new '(?<=\d{4})-(?=\d{4})'
 
-    def validate!(years)
-      years = years & valid_years.to_a
-    end
+    # 1920s, 1920s?, 192u, 192-, 192-?, 192? => 192x
+    DECADE = Regexp.new '(?<=\A\d{3})(-\??|0s\??|u|\?)\z'
 
-    def parse_year(value)
-      Date.parse(value).year
-    rescue ArgumentError
-      nil
-    end
+    # 2010/01 => 2010-01
+    MONTH = Regexp.new '(?<=\A\d{4})\/(?=\d{2}\z)'
 
-    def valid_years
-      (EARLIEST_YEAR..latest_year)
-    end
+    # 193u/, 193x/ => 1930/
+    START_DECADE = Regexp.new '(?<=\d{3})[uxX](?=\/)'
 
-    def latest_year
-      Date.today.year + 100
-    end
+    # /194x, /194u => /1949
+    END_DECADE = Regexp.new '(?<=\/\d{3})[uxX]'
 
-    def match_years(value)
-      result = match_year_range(value) ||
-               match_year_month(value) ||
-               match_year(value) ||
-               match_in_decade(value) ||
-               match_in_century(value) ||
-               match_decade(value) ||
-               match_between(value)
-      first_year, last_year = Array(result).map(&:to_i)
-      last_year ? (first_year..last_year) : first_year
-    end
+    # 19uu => 19xx
+    CENTURY = Regexp.new '(?<=\A\d{2})uu(?=\z)'
 
-    def match_year_range(value)
-      if m = YEAR_RANGE.match(value)
-        m[1, 2]
-      end
+    def self.call(object)
+      new(object).call
     end
 
-    def match_year_month(value)
-      if m = YEAR_MONTH.match(value)
-        m[1]
-      end
-    end
+    attr_reader :object
 
-    def match_year(value)
-      if m = YEAR.match(value)
-        value
-      end
+    def initialize(object)
+      @object = object
     end
 
-    def match_in_decade(value)
-      if m = IN_DECADE.match(value)
-        [ "#{m[1]}0", "#{m[1]}9" ]
+    def call
+      source_dates.each_with_object([]) do |date, facet_values|
+        date.split(VALUE_SEP).each do |value|
+          value.strip!
+          edtf_date = convert_to_edtf(value)
+          years = Array(edtf_years(edtf_date))
+          years.select! { |year| VALID_YEARS.include?(year) }
+          facet_values.push(*years)
+        end
       end
     end
 
-    def match_in_century(value)
-      if m = IN_CENTURY.match(value)
-        [ "#{m[1]}00", "#{m[1]}99" ]
-      end
-    end
+    private
 
-    def match_decade(value)
-      if m = DECADE.match(value)
-        [ m[1], m[1].sub(/0\z/, "9") ]
-      end
+    def source_dates
+      object.desc_metadata.date
     end
 
-    def match_between(value)
+    def convert_to_edtf(value)
       if m = BETWEEN.match(value)
         value.sub! m[1], ""  # [Bb]etween
-        value.sub! m[3], "-" # and
-        match_year_range(value)
+        value.sub! m[3], "/" # and
       end
+      substitutions.reduce(value) { |memo, (regexp, repl)| memo.gsub(regexp, repl) }
+    end
+
+    def substitutions
+      [
+        [ CIRCA,         "" ],
+        [ YEAR_RANGE,   "/" ],
+        [ DECADE,       "x" ],
+        [ MONTH,        "-" ],
+        [ START_DECADE, "0" ],
+        [ END_DECADE,   "9" ],
+        [ CENTURY,     "xx" ],
+      ]
+    end
+
+    def edtf_years(value)
+      case parsed = EDTF.parse!(value)
+      when Date, EDTF::Season
+        parsed.year
+      when EDTF::Set, EDTF::Interval, EDTF::Epoch
+        parsed.map(&:year).uniq
+      end
+    rescue ArgumentError # EDTF cannot parse
+      nil
     end
 
   end

data/spec/models/year_facet_spec.rb

@@ -1,64 +1,91 @@
 module Ddr::Models
   RSpec.describe YearFacet do
 
-    subject { described_class.new(obj) }
+    subject { described_class.call(obj) }
     let(:obj) { Item.new }
     before { obj.dc_date = [ date ] }
 
     describe "splitting on semicolons" do
       let(:date) { "1935; 1936; 1937; 1938" }
-      its(:values) { is_expected.to eq([1935, 1936, 1937, 1938]) }
+      it { is_expected.to eq([1935, 1936, 1937, 1938]) }
     end
 
     describe "year range" do
-      %w( 1935-1940 1935/1940 ).each do |value|
+      %w( 1935-1940 1935/1940 1935?/1940? 1935~/1940~ ).each do |value|
         describe value do
           let(:date) { value }
-          its(:values) { is_expected.to eq((1935..1940).to_a) }
+          it { is_expected.to eq((1935..1940).to_a) }
         end
       end
     end
 
-    describe "in decade" do
-      %w( 192x 192X 192? 192- 192-? ).each do |value|
+    describe "decade" do
+      %w( 192u 192x 192X 192? 192- 192-? 1920s 1920s? ).each do |value|
         describe value do
           let(:date) { value }
-          its(:values) { is_expected.to eq((1920..1929).to_a) }
+          it { is_expected.to eq((1920..1929).to_a) }
         end
       end
     end
 
-    describe "in century -- YYxx (19xx)" do
-      let(:date) { "19xx" }
-      its(:values) { is_expected.to eq((1900..1999).to_a) }
+    describe "century" do
+      %w( 19xx 19uu ).each do |value|
+        describe value do
+          let(:date) { value }
+          it { is_expected.to eq((1900..1999).to_a) }
+        end
+      end
     end
 
-    describe "decade" do
-      %w( 1920s 1920s? ).each do |value|
+    describe "uncertain interval" do
+      let(:date) { "199u/200u" }
+      it { is_expected.to eq((1990..2009).to_a) }
+    end
+
+    describe "set" do
+      let(:date) { "[1999,2000,2003]" }
+      it { is_expected.to eq([1999, 2000, 2003]) }
+    end
+
+    describe "year" do
+      %w( 2010 2010? 2010~ ).each do |value|
         describe value do
           let(:date) { value }
-          its(:values) { is_expected.to eq((1920..1929).to_a) }
+          it { is_expected.to eq([2010]) }
         end
       end
     end
 
-    describe "year + month" do
-      %w( 2010-01 2010/01 ).each do |value|
+    describe "month" do
+      %w( 2010-01 2010/01 2010-12? 2010-11/2010-12 ).each do |value|
         describe value do
           let(:date) { value }
-          its(:values) { is_expected.to eq([2010]) }
+          it { is_expected.to eq([2010]) }
         end
       end
     end
 
-    describe "between" do
-      let(:date) { "Between 1965 and 1968" }
-      its(:values) { is_expected.to eq((1965..1968).to_a) }
+    describe "day" do
+      %w( 2010-12-10 2010-12-10? 2010-12-10/2010-12-20 ).each do |value|
+        describe value do
+          let(:date) { value }
+          it { is_expected.to eq([2010]) }
+        end
+      end
     end
 
-    describe "year" do
-      let(:date) { "1965" }
-      its(:values) { is_expected.to eq([1965]) }
+    describe "season" do
+      %w( 2010-21 2010-22 2010-23 2010-24 ).each do |value|
+        describe value do
+          let(:date) { value }
+          it { is_expected.to eq([2010]) }
+        end
+      end
+    end
+
+    describe "between" do
+      let(:date) { "Between 1965 and 1968" }
+      it { is_expected.to eq((1965..1968).to_a) }
     end
 
   end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: ddr-models
 version: !ruby/object:Gem::Version
-  version: 3.0.0.rc1
+  version: 3.0.0.rc2
 platform: ruby
 authors:
 - Jim Coble
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-04-22 00:00:00.000000000 Z
+date: 2016-05-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rails
@@ -235,6 +235,20 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: 3.4.3
+- !ruby/object:Gem::Dependency
+  name: edtf
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '2.3'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '2.3'
 - !ruby/object:Gem::Dependency
   name: bundler
   requirement: !ruby/object:Gem::Requirement