cocina_display 2.12.0 → 2.13.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 16e21603fafbb47fe2471e436a256e19eac1d3b69acb85c1bb246d3f39e0d06c
-  data.tar.gz: '0898b33cdbf5b6c0b9f1f4d9fcdea3b4492d8b265b67fa90a08e14394e476bd2'
+  metadata.gz: 8dfb0909c582a5d1e3109bf0c98983e0ebd523af7340722856cb9d4482faa925
+  data.tar.gz: c2143a971249f8e6262dced0f4a05ef0d197426d3e86173f808c6f8f9321727e
 SHA512:
-  metadata.gz: ddc270e2faa1baa9fbcbbbeaaf8f6fba3ae030e5b6be5741ed407c398a1e1211daaeefdddeeb60ae056748a963c6b1f7c0017b745cc2a2a83d7dcf97de062064
-  data.tar.gz: d56dc751d95514b82c2cd943a21ab3befee10adeddb69546796a30e28513a9ac4bf3bc6c4e212b5fc0d0d585a9f34ea339e4c6ccfd5224550e6df35eb575c531
+  metadata.gz: e83761f763f300b62a7da1624d65f21b89fea93ab4170052c66a324dcd7b306b534fc5b84dc726dbb9945d5b98746398d84d8a9dbc00f3bd467658e12b801241
+  data.tar.gz: c06a74d7e19525eb27c71e314f2676d79fb1c3422efa0a9e08f39404feaa001d0c7252986761447a0bee6dd5bad6bc818310dae7a2fa89cb760af836b4bfd33f

data/lib/cocina_display/events/event.rb

@@ -161,9 +161,17 @@ module CocinaDisplay
       end
 
       # Locations associated with this event as a single string.
+      # Renders every parallel value, so vernacular and transliterated forms
+      # of the same place both appear.
       # @return [String]
       def place_str
-        locations.map(&:to_s).compact_blank.uniq.join(", ")
+        location_values.map(&:to_s).compact_blank.uniq.join(", ")
+      end
+
+      # All locations associated with this event, in every language/script.
+      # @return [Array<CocinaDisplay::Events::LocationValue>]
+      def location_values
+        locations.flat_map(&:parallel_values)
       end
     end
   end

data/lib/cocina_display/events/imprint.rb

@@ -50,7 +50,7 @@ module CocinaDisplay
       # 3. Keep only unique locations after decoding
       # @return [Array<String>]
       def display_locations
-        unencoded_locs, encoded_locs = locations.partition { |loc| loc.unencoded_value? }
+        unencoded_locs, encoded_locs = location_values.partition { |loc| loc.unencoded_value? }
         locs_for_display = unencoded_locs.presence || encoded_locs
         locs_for_display.map(&:to_s).compact_blank.uniq
       end

data/lib/cocina_display/events/location.rb

@@ -1,10 +1,13 @@
 module CocinaDisplay
   module Events
-    # A single location represented in a Cocina event, like a publication place.
-    class Location
+    # A location represented in a Cocina event, like a publication place.
+    # Can be expressed as several {LocationValue}s in different languages/scripts.
+    class Location < Parallel::Parallel
       MARC_COUNTRIES_FILE_PATH = CocinaDisplay.root / "config" / "marc_countries.yml"
 
-      attr_reader :cocina
+      # Common display methods reference the main location value. For parallel
+      # values, see #translated_value and #transliterated_value.
+      delegate :to_s, :unencoded_value?, :country_name, to: :main_value
 
       # A hash mapping MARC country codes to their names.
       # @return [Hash{String => String}]
@@ -12,50 +15,12 @@ module CocinaDisplay
         @marc_countries ||= YAML.safe_load_file(MARC_COUNTRIES_FILE_PATH)
       end
 
-      # Initialize a Location object with Cocina structured data.
-      # @param cocina [Hash] The Cocina structured data for the location.
-      def initialize(cocina)
-        @cocina = cocina
-      end
-
-      # The name of the location.
-      # Decodes a MARC country code if present and no value was present.
-      # @return [String, nil]
-      def to_s
-        cocina["value"] || country_name
-      end
-
-      # Is there an unencoded value (name) for this location?
-      # @return [Boolean]
-      def unencoded_value?
-        cocina["value"].present?
-      end
-
-      # Decoded country name if the location is encoded with a MARC country code.
-      # @return [String, nil]
-      def country_name
-        Location.marc_countries[code] if marc_country? && valid_country_code?
-      end
-
       private
 
-      # A code, like a MARC country code, representing the location.
-      # @return [String, nil]
-      def code
-        cocina["code"]
-      end
-
-      # Is this a decodable country code?
-      # Excludes blank values and "xx" (unknown) and "vp" (various places).
-      # @return [Boolean]
-      def valid_country_code?
-        code.present? && ["xx", "vp"].exclude?(code)
-      end
-
-      # Is this location encoded with a MARC country code?
-      # @return [Boolean]
-      def marc_country?
-        cocina.dig("source", "code") == "marccountry"
+      # The class to use for parallel values.
+      # @return [Class]
+      def parallel_value_class
+        LocationValue
       end
     end
   end

data/lib/cocina_display/events/location_value.rb

@@ -0,0 +1,46 @@
+module CocinaDisplay
+  module Events
+    # A location in a Cocina event in a single language/script.
+    class LocationValue < Parallel::ParallelValue
+      # The name of the location.
+      # Decodes a MARC country code if present and no value was present.
+      # @return [String, nil]
+      def to_s
+        cocina["value"] || country_name
+      end
+
+      # Is there an unencoded value (name) for this location?
+      # @return [Boolean]
+      def unencoded_value?
+        cocina["value"].present?
+      end
+
+      # Decoded country name if the location is encoded with a MARC country code.
+      # @return [String, nil]
+      def country_name
+        Location.marc_countries[code] if marc_country? && valid_country_code?
+      end
+
+      private
+
+      # A code, like a MARC country code, representing the location.
+      # @return [String, nil]
+      def code
+        cocina["code"]
+      end
+
+      # Is this a decodable country code?
+      # Excludes blank values and "xx" (unknown) and "vp" (various places).
+      # @return [Boolean]
+      def valid_country_code?
+        code.present? && ["xx", "vp"].exclude?(code)
+      end
+
+      # Is this location encoded with a MARC country code?
+      # @return [Boolean]
+      def marc_country?
+        cocina.dig("source", "code") == "marccountry"
+      end
+    end
+  end
+end

data/lib/cocina_display/geospatial.rb

@@ -335,6 +335,15 @@ module CocinaDisplay
       def self.supports?(input_str)
         input_str.match?(self::PATTERN)
       end
+
+      # Move a trailing hemisphere letter to the front, since that is where the
+      # decimal normalizer expects it.
+      # @example "121.5W" becomes "W121.5"
+      # @param [String] value
+      # @return [String]
+      def self.hemisphere_first(value)
+        value.sub(/\A(.+?)([NESW])\z/, '\2\1')
+      end
     end
 
     # Mixin that adds normalization for decimal degree coordinates.
@@ -345,17 +354,28 @@ module CocinaDisplay
 
       module Helpers
         # Convert hemispheres to plus/minus signs for parsing.
+        # @note The hemisphere can either lead or trail the degrees.
         # @param [String] coord_str
         # @return [String]
         def normalize_coord(coord_str)
-          coord_str.tr("EN", "+").tr("WS", "-")
+          hemisphere_first(coord_str).tr("EN", "+").tr("WS", "-")
         end
       end
     end
 
     # Mixin that adds normalization for DMS coordinates.
     module DMSParser
-      POINT_PATTERN = /(?<hem>[NESW])(?<deg>\d{1,3})[°⁰º]?(?:(?<min>\d{1,2})[ʹ′']?)?(?:(?<sec>\d{1,2})[ʺ"″]?)?/
+      # The degrees, minutes, and seconds of a coordinate, without a hemisphere.
+      DMS_PATTERN = /(?<deg>\d{1,3})[°⁰º]?(?:(?<min>\d{1,2})[ʹ′']?)?(?:(?<sec>\d{1,2})[ʺ"″]?)?/
+
+      # A single coordinate, with the hemisphere either leading or trailing the
+      # degrees. A hemisphere is required, so that a bare number is not read as
+      # a coordinate. Both spellings are common in MARC 034 and 255$c, and the
+      # trailing form is what {Coordinates#format_point} itself emits.
+      # @note A trailing hemisphere cannot be followed by digits, otherwise a
+      #   MARC 034 $b scale like "$b3100000W120°00′00″" would read the scale as
+      #   the degrees and steal the hemisphere from the coordinate after it.
+      POINT_PATTERN = /(?:(?<hem>[NESW])#{DMS_PATTERN}|#{DMS_PATTERN}(?<hem>[NESW])(?!\d))/
 
       def self.included(base)
         base.const_set(:POINT_PATTERN, POINT_PATTERN)
@@ -418,14 +438,6 @@ module CocinaDisplay
     # found in Cocina structured values, so that Geo::Coord can parse it.
     # Subclasses define a PATTERN and mix in a parser module for normalize_coord.
     class CoordinateNormalizer < CoordinatesParser
-      # Move a trailing hemisphere letter to the front, since that is where the
-      # parser normalizers expect it.
-      # @example "121.5W" becomes "W121.5"
-      # @param [String] value
-      # @return [String]
-      def self.hemisphere_first(value)
-        value.sub(/\A(.+?)([NESW])\z/, '\2\1')
-      end
     end
 
     # Normalizes DMS values, including the packed form used in MARC 034 subfields.
@@ -436,12 +448,6 @@ module CocinaDisplay
 
       # Either DMS punctuation, or a hemisphere paired with packed digits.
       PATTERN = /[°⁰º′ʹ'″ʺ"]|\A[NESW]\d{4,}\z|\A\d{4,}[NESW]\z/
-
-      # @param [String] value
-      # @return [String, nil]
-      def self.normalize_coord(value)
-        super(hemisphere_first(value))
-      end
     end
 
     # Normalizes decimal degree values, either signed or paired with a hemisphere.
@@ -452,12 +458,6 @@ module CocinaDisplay
       include DecimalParser
 
       PATTERN = /\A[NESW+-]?\d{1,3}(?:\.\d+)?[NESW]?\z/
-
-      # @param [String] value
-      # @return [String]
-      def self.normalize_coord(value)
-        super(hemisphere_first(value))
-      end
     end
 
     # Parse for decimal degree points, like "41.891797, 12.486419".
@@ -468,10 +468,12 @@ module CocinaDisplay
     end
 
     # Parser for DMS-format points, like "N34°03′08″ W118°14′37″".
+    # @note The hemisphere can either lead or trail the degrees in each half.
+    # @example 34°03′08″N 118°14′37″W
     class DMSPointParser < PointParser
       include DMSParser
 
-      PATTERN = /(?<lat>[^EW]+)(?<lng>[^NS]+)/
+      PATTERN = /(?<lat>[NS][^NS]+|[^NS]+[NS])(?<lng>[EW][^EW]+|[^EW]+[EW])/
     end
 
     # DMS-format bounding boxes with varying punctuation, delimited by -- and /.

data/lib/cocina_display/version.rb

@@ -2,5 +2,5 @@
 
 # :nodoc:
 module CocinaDisplay
-  VERSION = "2.12.0" # :nodoc:
+  VERSION = "2.13.0" # :nodoc:
 end

data/script/find_records.rb

@@ -3,28 +3,29 @@
 # This script is a simple, brute-force method for finding records that
 # exhibit certain characteristics in the public Cocina JSON for testing.
 #
-# It queries purl-fetcher for all DRUIDs released to a specific target and
-# then fetches each corresponding public Cocina record from PURL and examines it.
-#
-# You need to be on VPN to do this, as the purl-fetcher API is only accessible
-# from within the Stanford network.
+# It reads the public PURL sitemap to enumerate all released DRUIDs and then
+# fetches each corresponding public Cocina record from PURL and examines it.
 #
 # To use, modify any of the noted items below, then run:
 # $ bundle exec ruby script/find_records.rb
 #
 # You can exit early with Ctrl-C, and it will report how many records were
-# checked before exiting. Running through an entire target will take awhile,
+# checked before exiting. Running through the entire sitemap will take awhile,
 # on the order of 30 minutes or more.
 
 require "benchmark"
+require "net/http"
 require "pp"
-require "purl_fetcher/client"
+require "rexml/document"
+require "stringio"
+require "uri"
+require "zlib"
 require "cocina_display"
 require "cocina_display/utils"
 
-# This should correspond to one of the release targets available in purl-fetcher,
-# i.e. "Searchworks", "Earthworks", etc.
-RELEASE_TARGET = "Searchworks"
+# The PURL sitemap index. This points to one or more gzipped child sitemaps,
+# each of which lists PURL URLs (one per released DRUID).
+SITEMAP_URL = "https://purl.stanford.edu/system/sitemap/sitemap.xml.gz"
 
 # Modify this expression to match the JSON path you want to search, or just
 # modify the `examine_record` method directly.
@@ -36,8 +37,33 @@ def examine_record(record)
   record.path(PATH_EXPR).map { |value, _node, _key, path| [path, CocinaDisplay::Utils.deep_compact_blank(value)] }
 end
 
-# Track total records in target and how many we've seen
-released_to_target = []
+# Fetch a URL and return the response body, transparently decompressing it if
+# it was served (or named) as gzip.
+def fetch_gzipped(url)
+  body = Net::HTTP.get(URI(url))
+  Zlib::GzipReader.new(StringIO.new(body)).read
+rescue Zlib::GzipFile::Error
+  body
+end
+
+# Extract every <loc> value from a sitemap or sitemap index document.
+def sitemap_locs(xml)
+  REXML::Document.new(xml).get_elements("//loc").map { |loc| loc.text.strip }
+end
+
+# Walk the sitemap index and yield the DRUID for every URL in each child sitemap.
+def each_druid_in_sitemap(sitemap_url)
+  return enum_for(:each_druid_in_sitemap, sitemap_url) unless block_given?
+
+  sitemap_locs(fetch_gzipped(sitemap_url)).each do |child_sitemap_url|
+    sitemap_locs(fetch_gzipped(child_sitemap_url)).each do |purl_url|
+      yield File.basename(URI(purl_url).path)
+    end
+  end
+end
+
+# Track total records and how many we've seen
+druids = []
 processed_records = 0
 
 # Handle Ctrl-C gracefully
@@ -46,24 +72,21 @@ Signal.trap("INT") do
   exit
 end
 
-# Fetch everything from purl-fetcher; note that this is one single HTTP request
-# that returns a massive JSON response – it can be quite slow
-puts "Finding records released to #{RELEASE_TARGET}..."
-client = PurlFetcher::Client::Reader.new
+# Read the sitemap; this involves a handful of HTTP requests (the index plus
+# each child sitemap) that are relatively quick compared to purl-fetcher.
+puts "Finding released records from the PURL sitemap..."
 query_time = Benchmark.realtime do
-  client.released_to(RELEASE_TARGET).each do |record|
-    released_to_target << record["druid"].delete_prefix("druid:")
-  end
-rescue Faraday::ConnectionFailed => e
-  puts "Connection failed: #{e.message}; are you on VPN?"
+  each_druid_in_sitemap(SITEMAP_URL) { |druid| druids << druid }
+rescue => e
+  puts "Failed to read sitemap: #{e.message}"
   exit 1
 end
-puts "Found #{released_to_target.size} records released to #{RELEASE_TARGET} in #{query_time.round(2)} seconds"
+puts "Found #{druids.size} records in the sitemap in #{query_time.round(2)} seconds"
 
 # Iterate through the list of DRUIDs and fetch each one from PURL, creating a
 # CocinaRecord object. Then call our examine_record method on it and if
 # anything was returned, print the DRUID and the results.
-released_to_target.each do |druid|
+druids.each do |druid|
   begin
     cocina_record = CocinaDisplay::CocinaRecord.fetch(druid)
     processed_records += 1

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification
 name: cocina_display
 version: !ruby/object:Gem::Version
-  version: 2.12.0
+  version: 2.13.0
 platform: ruby
 authors:
 - Nick Budak
 bindir: exe
 cert_chain: []
-date: 2026-08-11 00:00:00.000000000 Z
+date: 1980-01-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: janeway-jsonpath
@@ -217,20 +217,6 @@ dependencies:
     - - ">="
       - !ruby/object:Gem::Version
         version: 1.9.1
-- !ruby/object:Gem::Dependency
-  name: purl_fetcher-client
-  requirement: !ruby/object:Gem::Requirement
-    requirements:
-    - - "~>"
-      - !ruby/object:Gem::Version
-        version: '3.1'
-  type: :development
-  prerelease: false
-  version_requirements: !ruby/object:Gem::Requirement
-    requirements:
-    - - "~>"
-      - !ruby/object:Gem::Version
-        version: '3.1'
 email:
 - budak@stanford.edu
 executables: []
@@ -279,6 +265,7 @@ files:
 - lib/cocina_display/events/event.rb
 - lib/cocina_display/events/imprint.rb
 - lib/cocina_display/events/location.rb
+- lib/cocina_display/events/location_value.rb
 - lib/cocina_display/events/note.rb
 - lib/cocina_display/forms/form.rb
 - lib/cocina_display/forms/genre.rb
@@ -326,7 +313,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.6.2
+rubygems_version: 4.0.15
 specification_version: 4
 summary: Helpers for rendering Cocina metadata
 test_files: []