ddr-models 2.11.0 → 3.0.0.alpha.1

data/lib/ddr/models/version.rb

@@ -1,5 +1,5 @@
 module Ddr
   module Models
-    VERSION = "2.11.0"
+    VERSION = "3.0.0.alpha.1"
   end
 end

data/lib/ddr/models/year_facet.rb

@@ -1,95 +1,153 @@
 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'
 
-    # circa 1920, ca. 1920, c1920 => 1920
-    CIRCA = Regexp.new '\b(circa\s+|ca?\.\s*|c(?=\d{4}[^\d]*))'
+    # 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'
 
-    # 1935-1940 => 1935/1940
-    YEAR_RANGE = Regexp.new '(?<=\d{4})-(?=\d{4})'
+    # YYY0s (1920s)
+    # YYY0s? (1920s?)
+    DECADE = Regexp.new '\A(\d{3}0)s\??\z'
 
-    # 1920s, 1920s?, 192u, 192-, 192-?, 192? => 192x
-    DECADE = Regexp.new '(?<=\A\d{3})(-\??|0s\??|u|\?)\z'
+    # YYYY-MM (2010-01)
+    # YYYY/MM (2010/01)
+    YEAR_MONTH = Regexp.new '\A(\d{4})[/-](0[1-9]|1[0-2])\z'
 
-    # 2010/01 => 2010-01
-    MONTH = Regexp.new '(?<=\A\d{4})\/(?=\d{2}\z)'
+    # YYYY-YYYY (1935-2010)
+    # YYYY/YYYY (1935/2010)
+    YEAR_RANGE = Regexp.new '\A(\d{4})[/-](\d{4})\z'
 
-    # 193u/, 193x/ => 1930/
-    START_DECADE = Regexp.new '(?<=\d{3})[uxX](?=\/)'
+    # YYYY (1979)
+    YEAR = Regexp.new '\A\d{4}\z'
 
-    # /194x, /194u => /1949
-    END_DECADE = Regexp.new '(?<=\/\d{3})[uxX]'
+    SQUARE_BRACKETS = Regexp.new '[\[\]]'
 
-    # 19uu => 19xx
-    CENTURY = Regexp.new '(?<=\A\d{2})uu(?=\z)'
+    # c. 1920
+    # ca. 1920
+    # c1920
+    CIRCA = Regexp.new '\b(circa\s+|ca?\.\s*|c(?=\d{4}[^\d]*))'
 
-    def self.call(object)
-      new(object).call
+    class << self
+      def call(obj)
+        new(obj).values
+      end
     end
 
-    attr_reader :object
+    attr_reader :obj, :values
 
-    def initialize(object)
-      @object = object
+    def initialize(obj)
+      @obj = obj
+      @values = []
+      facet_values
     end
 
-    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)
+    def facet_values
+      obj.descMetadata.date.each do |date|
+        date.split(/;/).each do |value|
+          clean! value
+          years = extract_years(value)
+          validate! years
+          values.push *years
         end
       end
     end
 
-    private
+    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
+
+    def validate!(years)
+      years = years & valid_years.to_a
+    end
+
+    def parse_year(value)
+      Date.parse(value).year
+    rescue ArgumentError
+      nil
+    end
+
+    def valid_years
+      (EARLIEST_YEAR..latest_year)
+    end
+
+    def latest_year
+      Date.today.year + 100
+    end
+
+    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
+
+    def match_year_range(value)
+      if m = YEAR_RANGE.match(value)
+        m[1, 2]
+      end
+    end
+
+    def match_year_month(value)
+      if m = YEAR_MONTH.match(value)
+        m[1]
+      end
+    end
+
+    def match_year(value)
+      if m = YEAR.match(value)
+        value
+      end
+    end
+
+    def match_in_decade(value)
+      if m = IN_DECADE.match(value)
+        [ "#{m[1]}0", "#{m[1]}9" ]
+      end
+    end
 
-    def source_dates
-      object.descMetadata.date
+    def match_in_century(value)
+      if m = IN_CENTURY.match(value)
+        [ "#{m[1]}00", "#{m[1]}99" ]
+      end
     end
 
-    def convert_to_edtf(value)
+    def match_decade(value)
+      if m = DECADE.match(value)
+        [ m[1], m[1].sub(/0\z/, "9") ]
+      end
+    end
+
+    def match_between(value)
       if m = BETWEEN.match(value)
         value.sub! m[1], ""  # [Bb]etween
-        value.sub! m[3], "/" # and
+        value.sub! m[3], "-" # and
+        match_year_range(value)
       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/lib/ddr/notifications.rb

@@ -1,14 +1,16 @@
-module Ddr::Notifications
+module Ddr
+  module Notifications
 
-  FIXITY_CHECK = "fixity_check.events.ddr"
-  VIRUS_CHECK = "virus_check.events.ddr"
-  CREATION = "creation.events.ddr"
-  UPDATE = "update.events.ddr"
-  DELETION = "deletion.events.ddr"
+    FIXITY_CHECK = "fixity_check.events.ddr"
+    VIRUS_CHECK = "virus_check.events.ddr"
+    CREATION = "creation.events.ddr"
+    UPDATE = "update.events.ddr"
+    DELETION = "deletion.events.ddr"
 
-  def self.notify_event(type, args={})
-    name = "#{type}.events.ddr"
-    ActiveSupport::Notifications.instrument(name, args)
-  end
+    def self.notify_event(type, args={})
+      name = "#{type}.events.ddr"
+      ActiveSupport::Notifications.instrument(name, args)
+    end
 
+  end
 end

data/lib/ddr/utils.rb

@@ -1,9 +1,10 @@
 require 'openssl'
 
 module Ddr::Utils
-  extend Deprecation
 
-  def self.digest content, algorithm
+  DEFAULT_MIME_TYPE = "application/octet-stream"
+
+  def self.digest(content, algorithm)
     raise TypeError, "Algorithm must be a string: #{algorithm.inspect}" unless algorithm.is_a?(String)
     digest_class = OpenSSL::Digest.const_get(algorithm.sub("-", "").to_sym)
     digest_class.new(content).to_s
@@ -13,11 +14,12 @@ module Ddr::Utils
 
   # Return a mime type for the file, using the file_name if necessary
   # file can be a File object or file path (String)
-  # @return [String] the mime type or default
+  # Return default mime type if unable to determine otherwise
   def self.mime_type_for(file, file_name=nil)
-    Deprecation.warn(self, "`Ddr::Utils.mime_type_for` is deprecated and will be removed in ddr-models 3.0." \
-                           " Use `Ddr::Models::MediaType.call(file)` instead. (called from #{caller.first})")
-    Ddr::Models::MediaType.call(file_name || file)
+    return file.content_type if file.respond_to?(:content_type) # E.g., Rails uploaded file
+    path = file_name || file_path(file) rescue nil
+    mime_types = MIME::Types.of(path) rescue []                 # MIME::Types.of blows up on nil
+    mime_types.empty? ? DEFAULT_MIME_TYPE : mime_types.first.content_type
   end
 
   def self.file_or_path?(file)
@@ -28,7 +30,7 @@ module Ddr::Utils
 
   def self.file_path?(file)
     # length is a sanity check
-    file.is_a?(String) && (file.length < 1024) && File.exist?(file)
+    file.is_a?(String) && (file.length < 1024) && File.exists?(file)
   end
 
   def self.file_path(file)
@@ -37,16 +39,13 @@ module Ddr::Utils
     elsif file_path?(file)
       file
     else
-      raise ArgumentError, "Argument is neither a File nor a path to an existing file."
+      raise ArgumentError, "File argument is neither a File nor a path to an existing file."
     end
   end
 
   def self.file_name_for(file)
-    if file.respond_to?(:original_filename) && file.original_filename.present?
-      file.original_filename
-    else
-      File.basename file_path(file)
-    end
+    return file.original_filename if file.respond_to?(:original_filename) && file.original_filename.present?
+    File.basename(file_path(file)) rescue nil
   end
 
   def self.file_uri?(uri)
@@ -64,23 +63,15 @@ module Ddr::Utils
   # Return file path for URI string
   # Should reverse .path_to_uri
   # "file:/path/to/file" => "/path/to/file"
-  # @param uri [String] The URI string to pathify
-  # @return [String] the file path
-  def self.path_from_uri(uri_string)
-    uri = URI.parse(uri_string)
-    unless uri.scheme == "file"
-      raise ArgumentError, "URI does not have the file: scheme."
-    end
-    URI.unescape(uri.path)
+  def self.path_from_uri(uri)
+    URI.unescape(URI.parse(uri).path)
   end
 
   # Return URI string for file path
   # Should reverse .path_from_uri
   # "/path/to/file" => "file:/path/to/file"
-  # @param path [String] the file path
-  # @return [String] the file: URI string
   def self.path_to_uri(path)
-    uri = URI.parse URI.escape(path)
+    uri = URI.parse(URI.escape(path))
     uri.scheme = "file"
     uri.to_s
   end
@@ -178,8 +169,4 @@ module Ddr::Utils
     dts.map { |dt| solr_date(dt) }
   end
 
-  class << self
-    alias_method :file_name, :file_name_for
-  end
-
 end

data/lib/ddr/vocab.rb

@@ -1,22 +1,24 @@
 require "rdf/vocab"
 
-module Ddr::Vocab
-  extend ActiveSupport::Autoload
+module Ddr
+  module Vocab
+    extend ActiveSupport::Autoload
 
-  BASE_URI = "http://repository.lib.duke.edu/vocab"
+    BASE_URI = "http://repository.lib.duke.edu/vocab"
 
-  PREMIS = begin
-             RDF::Vocab::PREMIS::V1
-           rescue NameError
-             RDF::Vocab::PREMIS
-           end
+    PREMIS = begin
+               RDF::Vocab::PREMIS::V1
+             rescue NameError
+               RDF::Vocab::PREMIS
+             end
 
-  autoload :Asset
-  autoload :Contact
-  autoload :Display
-  autoload :DukeTerms
-  autoload :RDFVocabularyParser
-  autoload :Roles
-  autoload :Vocabulary
+    autoload :Asset
+    autoload :Contact
+    autoload :Display
+    autoload :DukeTerms
+    autoload :RDFVocabularyParser
+    autoload :Roles
+    autoload :Vocabulary
 
+  end
 end

data/lib/ddr/vocab/asset.rb

@@ -1,51 +1,26 @@
-module Ddr::Vocab
-  class Asset < RDF::StrictVocabulary("http://repository.lib.duke.edu/vocab/asset/")
+ module Ddr
+  module Vocab
+    class Asset < RDF::StrictVocabulary("http://repository.lib.duke.edu/vocab/asset/")
 
-    property "permanentId",
-             label: "Permanent Identifier"
+      property "permanentId",
+        label: "Permanent Identifier"
 
-    property "permanentUrl",
-             label: "Permanent URL"
+      property "permanentUrl",
+        label: "Permanent URL"
 
-    property "workflowState",
-             label: "Workflow State"
+      property "workflowState",
+        label: "Workflow State"
 
-    property "adminSet",
-             label: "Administrative Set",
-             comment: "A name under which objects (principally collections) are grouped for administrative purposes."
+      property "adminSet",
+        label: "Administrative Set",
+        comment: "A name under which objects (principally collections) are grouped for administrative purposes."
 
-    property "eadId",
-             label: "EAD ID"
+      property "eadId",
+        label: "EAD ID"
 
-    property "archivesSpaceId",
-             label: "ArchivesSpace Identifier"
-
-    property "isLocked",
-             label: "Is Locked?"
-
-    property "ingestedBy",
-             label: "Ingested By",
-             comment: "The agent (person or software) that initiated or performed the ingestion of the object."
-
-    property "ingestionDate",
-             label: "Ingestion Date",
-             comment: "The date/time at which the object was originally ingested into the repository."
-
-    property "rightsNote",
-             label: "Rights Note",
-             comment: "Free-text statement about the rights status of the resource."
-
-    property "alephId",
-             label: "Aleph ID",
-             comment: "The Aleph identifier for a catalog record corresponding to the object."
-
-    property "affiliation",
-             label: "Affiliation",
-             comment: "An organizational entity associated with the object content."
-
-    property "nestedPath",
-             label: "Nested Path",
-             comment: "The nested/tree path to the object to be reflected in structural metadata."
+      property "multiresImageFilePath",
+        label: "Multires Image File Path"
 
+    end
   end
 end