stanford-mods 1.4.0 → 1.5.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 77f36520f38dfd83cd8cc79480fc5fbda45b2df6
-  data.tar.gz: b1e7a4db9be39cdbeeca3124a62e3c0403ba31a2
+  metadata.gz: fda4f5ab13e13c81db05b27fd3ca5e5e17be0e1d
+  data.tar.gz: 13a2ee552756a2e700ee102ada1b56b55d5cda82
 SHA512:
-  metadata.gz: fd0ab891b9f578611b928231dcf45debcffc209068032fdbcd98d93ada56e95fc5afa82be97dfc163287297e190ff1ef258abef27dcd58170ec18152381da398
-  data.tar.gz: f604d45d319fbce30215ff436ad62ebc23a0bab11d995d6599bef137dae9a5465344d69d2492d84b29436711106ae7e80ce9090cb3005c3be4fe1fb004b58b7e
+  metadata.gz: 63c3efbf88ea9d76b80182c490cb01c3d6d404858ce7d91f46b1d98c3df284cf8dad19cf6b55eb2185a1779bafd8c750dde4e04cf39e64f6c810cf030cc80359
+  data.tar.gz: 275a160cf3603693e5801f0d5ff530995463c00a5f77625391fc09efaca8b13f0621f5a2e5034de5e1ae9cbb27d7a59fa0dcaf53aac877d057e65e004ecc3627

data/lib/stanford-mods.rb

@@ -1,5 +1,6 @@
 require 'mods'
 require 'stanford-mods/date_parsing'
+require 'stanford-mods/coordinate'
 require 'stanford-mods/geo_spatial'
 require 'stanford-mods/name'
 require 'stanford-mods/origin_info'

data/lib/stanford-mods/coordinate.rb

@@ -0,0 +1,81 @@
+# encoding: utf-8
+module Stanford
+  module Mods
+    ##
+    # Geospatial coordinate parsing
+    class Coordinate
+      attr_reader :value
+
+      def initialize(value)
+        @value = value
+      end
+
+      # Convert the coordinate to a WKT/CQL ENVELOPE representation
+      def as_envelope
+        return unless valid?
+
+        "ENVELOPE(#{bounds[:min_x]}, #{bounds[:max_x]}, #{bounds[:min_y]}, #{bounds[:max_y]})"
+      end
+
+      # Convert the coordinate to a Solr 4.x bbox-format representation
+      def as_bbox
+        return unless valid?
+
+        "#{bounds[:min_x]} #{bounds[:min_y]} #{bounds[:max_x]} #{bounds[:max_y]}"
+      end
+
+      def valid?
+        return false if bounds.empty?
+
+        range_x = -180.0..180.0
+        range_y = -90.0..90.0
+
+        range_x.include?(bounds[:min_x]) &&
+          range_x.include?(bounds[:max_x]) &&
+          range_y.include?(bounds[:min_y]) &&
+          range_y.include?(bounds[:max_y])
+      end
+
+      private
+
+      def bounds
+        @bounds ||= begin
+          matches = coord.match %r{\A(?<lat>[EW].+-+.+)\s*/\s*(?<lng>[NS].+-+.+)\Z}
+
+          if matches
+            min_x, max_x = matches['lat'].split(/-+/).map { |x| coord_to_decimal(x) }.minmax
+            min_y, max_y = matches['lng'].split(/-+/).map { |y| coord_to_decimal(y) }.minmax
+
+            { min_x: min_x, min_y: min_y, max_x: max_x, max_y: max_y }
+          else
+            {}
+          end
+        end
+      end
+
+      def coord
+        matches = value.match(/^\(?([^)]+)\)?\.?$/)
+
+        if matches
+          matches[1]
+        else
+          value
+        end
+      end
+
+      def coord_to_decimal(point)
+        regex = /(?<dir>[NESW])\s*(?<deg>\d+)[°⁰º](?:(?<min>\d+)[ʹ'])?(?:(?<sec>\d+)[ʺ"])?/
+        match = regex.match(point)
+
+        return Float::INFINITY unless match
+
+        dec = match['deg'].to_i
+        dec += match['min'].to_f / 60
+        dec += match['sec'].to_f / 60 / 60
+        dec = -1 * dec if match['dir'] == 'W' || match['dir'] == 'S'
+
+        dec
+      end
+    end
+  end
+end

data/lib/stanford-mods/geo_spatial.rb

@@ -10,50 +10,23 @@ module Stanford
         Array(@mods_ng_xml.subject.cartographics.coordinates).map(&:text)
       end
 
-      def point_bbox
+      def coordinates_as_envelope
         coordinates.map do |n|
-          matches = n.match(/^\(?([^)]+)\)?\.?$/)
+          c = Stanford::Mods::Coordinate.new(n)
 
-          if matches
-            coord_to_bbox(matches[1])
-          else
-            coord_to_bbox(n)
-          end
+          c.as_envelope if c.valid?
         end.compact
       end
 
-      private
-
-      def coord_to_bbox(coord)
-        matches = coord.match %r{\A(?<lat>[EW].+-+.+)\s*/\s*(?<lng>[NS].+-+.+)\Z}
-        return unless matches
-
-        min_x, max_x = matches['lat'].split(/-+/).map { |x| coord_to_decimal(x) }.minmax
-        min_y, max_y = matches['lng'].split(/-+/).map { |y| coord_to_decimal(y) }.minmax
-
-        "#{min_x} #{min_y} #{max_x} #{max_y}" if valid_bbox?(min_x, max_x, min_y, max_y)
-      end
-
-      def coord_to_decimal(point)
-        regex = /(?<dir>[NESW])\s*(?<deg>\d+)[°⁰º](?:(?<min>\d+)[ʹ'])?(?:(?<sec>\d+)[ʺ"])?/
-        match = regex.match(point)
-
-        return Float::INFINITY unless match
-
-        dec = match['deg'].to_i
-        dec += match['min'].to_f / 60
-        dec += match['sec'].to_f / 60 / 60
-        dec = -1 * dec if match['dir'] == 'W' || match['dir'] == 'S'
+      def coordinates_as_bbox
+        coordinates.map do |n|
+          c = Stanford::Mods::Coordinate.new(n)
 
-        dec
+          c.as_bbox if c.valid?
+        end.compact
       end
 
-      def valid_bbox?(min_x, max_x, min_y, max_y)
-        range_x = -180.0..180.0
-        range_y = -90.0..90.0
-
-        range_x.include?(min_x) && range_x.include?(max_x) && range_y.include?(min_y) && range_y.include?(max_y)
-      end
+      alias point_bbox coordinates_as_bbox
     end # class Record
   end # Module Mods
 end # Module Stanford

data/lib/stanford-mods/version.rb

@@ -1,6 +1,6 @@
 module Stanford
   module Mods
     # this is the Ruby Gem version
-    VERSION = "1.4.0"
+    VERSION = "1.5.0"
   end
 end

data/spec/geo_spatial_spec.rb

@@ -53,7 +53,7 @@ describe "Cartographic coordinates" do
     end
   end
 
-  context "point_bbox" do
+  describe "#coordinates_as_bbox" do
     it "returns empty array if no coordinates in the mods" do
       smods_rec.from_str(no_coord)
       expect(smods_rec.point_bbox).to eq([])
@@ -66,44 +66,20 @@ describe "Cartographic coordinates" do
       smods_rec.from_str(with_coords)
       expect(smods_rec.point_bbox).to eq(["-16.0 -15.0 28.0 13.0"])
     end
+  end
 
-    {
-      %((W 123°23ʹ16ʺ--W 122°31ʹ22ʺ/N 39°23ʹ57ʺ--N 38°17ʹ53ʺ)) =>
-        ['-123.38777777777779 38.29805555555556 -122.52277777777778 39.399166666666666'],
-      %(E 10°03'00"--E 12°58'00"/N 45°00'00"--N 41°46'00") =>
-        ['10.05 41.766666666666666 12.966666666666667 45.0'],
-      %(E 8°41'-E 12°21'/N 46°04'-N 44°23') =>
-        ['8.683333333333334 44.38333333333333 12.35 46.06666666666667'],
-      %((E17°--E11°/N14°--N18°).) =>
-        ['11.0 14.0 17.0 18.0'], # coordinates need to be reordered
-      %((W 170⁰--E 55⁰/N 40⁰--S 36⁰).) =>
-        ['-170.0 -36.0 55.0 40.0'], # superscript 0 is almost a degree character..
-      %(W80°--E100°/N487°--S42°) =>
-        [], # N487 is out of bounds for the bounding box
-      %((W 0°-W 0°/S 90°---S 90°)) =>
-        ['-0.0 -90.0 -0.0 -90.0'], # one dash, two dashes, three dashes.. what's the difference?
-      %(W 650--W 100/N 700--N 550) =>
-        [] # missing degree character, and all coordinates are out of bounds.
-    }.each do |value, expected|
-      describe 'data mappings' do
-        let(:mods) do
-          <<-EOF
-            <mods xmlns="#{Mods::MODS_NS}">
-              <subject>
-                <cartographics>
-                  <coordinates>#{value}</coordinates>
-                </cartographics>
-              </subject>
-            </mods>
-          EOF
-        end
-
-        let(:smods_rec) { Stanford::Mods::Record.new.from_str(mods) }
-
-        it 'maps to the right bounding box' do
-          expect(smods_rec.point_bbox).to eq expected
-        end
-      end
+  describe "#coordinates_as_envelope" do
+    it "returns empty array if no coordinates in the mods" do
+      smods_rec.from_str(no_coord)
+      expect(smods_rec.coordinates_as_envelope).to eq([])
+    end
+    it "returns empty array if bad data is in the mods" do
+      smods_rec.from_str(with_bad_data)
+      expect(smods_rec.coordinates_as_envelope).to eq([])
+    end
+    it "returns decimal representation of latitude and longitude" do
+      smods_rec.from_str(with_coords)
+      expect(smods_rec.coordinates_as_envelope).to eq(["ENVELOPE(-16.0, 28.0, -15.0, 13.0)"])
     end
   end
 end # describe Cartographic coordinates

data/spec/lib/stanford-mods/coordinate_spec.rb

@@ -0,0 +1,84 @@
+# encoding: utf-8
+require 'spec_helper'
+
+describe Stanford::Mods::Coordinate do
+  describe '#valid' do
+    it 'is valid for well-formed coordinates' do
+      expect(described_class.new('W 123°23ʹ16ʺ--W 122°31ʹ22ʺ/N 39°23ʹ57ʺ--N 38°17ʹ53ʺ')).to be_valid
+    end
+
+    it 'rejects out-of-bounds coordinates' do
+      expect(described_class.new('W80°--E100°/N487°--S42°')).not_to be_valid
+    end
+
+    it 'rejects coordinates without degree symbols' do
+      expect(described_class.new('W 650--W 100/N 700--N 550')).not_to be_valid
+    end
+
+    it 'rejects malformed coordinates' do
+      expect(described_class.new('(E29°--E35/°S12°--S16°).')).not_to be_valid
+    end
+  end
+
+  describe '#as_bbox' do
+    it 'is nil for invalid data' do
+      expect(described_class.new('x').as_bbox).to eq nil
+    end
+  end
+
+  describe '#as_envelope' do
+    it 'is nil for invalid data' do
+      expect(described_class.new('x').as_envelope).to eq nil
+    end
+  end
+
+  context '#as_bbox' do
+    {
+      %((W 123°23ʹ16ʺ--W 122°31ʹ22ʺ/N 39°23ʹ57ʺ--N 38°17ʹ53ʺ)) =>
+        '-123.38777777777779 38.29805555555556 -122.52277777777778 39.399166666666666',
+      %(E 10°03'00"--E 12°58'00"/N 45°00'00"--N 41°46'00") =>
+        '10.05 41.766666666666666 12.966666666666667 45.0',
+      %(E 8°41'-E 12°21'/N 46°04'-N 44°23') =>
+        '8.683333333333334 44.38333333333333 12.35 46.06666666666667',
+      %((E17°--E11°/N14°--N18°).) =>
+        '11.0 14.0 17.0 18.0', # coordinates need to be reordered
+      %((W 170⁰--E 55⁰/N 40⁰--S 36⁰).) =>
+        '-170.0 -36.0 55.0 40.0', # superscript 0 is almost a degree character..
+      %((W 0°-W 0°/S 90°---S 90°)) =>
+        '-0.0 -90.0 -0.0 -90.0' # one dash, two dashes, three dashes.. what's the difference?
+    }.each do |value, expected|
+      describe 'parsing' do
+        let(:subject) { described_class.new(value) }
+
+        it 'transforms into the right bbox' do
+          expect(subject.as_bbox).to eq expected
+        end
+      end
+    end
+  end
+
+  context '#as_envelope' do
+    {
+      %((W 123°23ʹ16ʺ--W 122°31ʹ22ʺ/N 39°23ʹ57ʺ--N 38°17ʹ53ʺ)) =>
+        'ENVELOPE(-123.38777777777779, -122.52277777777778, 38.29805555555556, 39.399166666666666)',
+      %(E 10°03'00"--E 12°58'00"/N 45°00'00"--N 41°46'00") =>
+        'ENVELOPE(10.05, 12.966666666666667, 41.766666666666666, 45.0)',
+      %(E 8°41'-E 12°21'/N 46°04'-N 44°23') =>
+        'ENVELOPE(8.683333333333334, 12.35, 44.38333333333333, 46.06666666666667)',
+      %((E17°--E11°/N14°--N18°).) =>
+        'ENVELOPE(11.0, 17.0, 14.0, 18.0)', # coordinates need to be reordered
+      %((W 170⁰--E 55⁰/N 40⁰--S 36⁰).) =>
+        'ENVELOPE(-170.0, 55.0, -36.0, 40.0)', # superscript 0 is almost a degree character..
+      %((W 0°-W 0°/S 90°---S 90°)) =>
+        'ENVELOPE(-0.0, -0.0, -90.0, -90.0)' # one dash, two dashes, three dashes.. what's the difference?
+    }.each do |value, expected|
+      describe 'parsing' do
+        let(:subject) { described_class.new(value) }
+
+        it 'transforms into the right envelope' do
+          expect(subject.as_envelope).to eq expected
+        end
+      end
+    end
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: stanford-mods
 version: !ruby/object:Gem::Version
-  version: 1.4.0
+  version: 1.5.0
 platform: ruby
 authors:
 - Naomi Dushay
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-01-13 00:00:00.000000000 Z
+date: 2016-01-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: mods
@@ -120,6 +120,7 @@ files:
 - Rakefile
 - config/mappings_hash.rb
 - lib/stanford-mods.rb
+- lib/stanford-mods/coordinate.rb
 - lib/stanford-mods/date_parsing.rb
 - lib/stanford-mods/geo_spatial.rb
 - lib/stanford-mods/name.rb
@@ -132,6 +133,7 @@ files:
 - spec/date_parsing_spec.rb
 - spec/fixtures/spotlight_pub_date_data.rb
 - spec/geo_spatial_spec.rb
+- spec/lib/stanford-mods/coordinate_spec.rb
 - spec/name_spec.rb
 - spec/origin_info_spec.rb
 - spec/physical_location_spec.rb
@@ -171,6 +173,7 @@ test_files:
 - spec/date_parsing_spec.rb
 - spec/fixtures/spotlight_pub_date_data.rb
 - spec/geo_spatial_spec.rb
+- spec/lib/stanford-mods/coordinate_spec.rb
 - spec/name_spec.rb
 - spec/origin_info_spec.rb
 - spec/physical_location_spec.rb