geomash 0.2.1 → 0.3.1

data/lib/geomash/version.rb

@@ -1,3 +1,3 @@
 module Geomash
-  VERSION = "0.2.1"
+  VERSION = "0.3.1"
 end

data/lib/geomash.rb

@@ -5,6 +5,8 @@ module Geomash
   require "geomash/tgn"
   require "geomash/geonames"
   require "geomash/town_lookup"
+  require "geomash/autoexpire_cache_dalli"
+  require "geomash/autoexpire_cache_redis"
   require "geocoder"
   require "countries"
   require "unidecoder"
@@ -55,25 +57,13 @@ module Geomash
       #FIXME
       return_hash[:tgn] = Geomash::TGN.tgn_id_from_geo_hash(return_hash)
 
-      if return_hash[:tgn].blank?
+      if return_hash[:tgn].blank? || (return_hash[:tgn][:original_string_differs] && return_hash[:state_part].present?)
         geo_hash_temp =  Geomash::Standardizer.try_with_entered_names(return_hash)
-        return_hash[:tgn] = Geomash::TGN.tgn_id_from_geo_hash(geo_hash_temp) if geo_hash_temp.present?
-
-        if return_hash[:tgn].blank? && return_hash[:neighborhood_part].present?
-
-          geo_hash_temp = return_hash.clone
-          geo_hash_temp[:neighborhood_part] = nil
-          geo_hash_temp[:original_string_differs] = true
-          return_hash[:tgn] = Geomash::TGN.tgn_id_from_geo_hash(geo_hash_temp)
-          return_hash[:tgn][:original_string_differs] = true if return_hash[:tgn].present?
-        elsif return_hash[:city_part].present? && return_hash[:tgn].blank?
-
-          geo_hash_temp = return_hash.clone
-          geo_hash_temp[:city_part] = nil
-          geo_hash_temp[:original_string_differs] = true
-          return_hash[:tgn] = Geomash::TGN.tgn_id_from_geo_hash(geo_hash_temp)
-          return_hash[:tgn][:original_string_differs] = true if return_hash[:tgn].present?
-
+        geo_hash_temp[:tgn] = Geomash::TGN.tgn_id_from_geo_hash(geo_hash_temp)  if geo_hash_temp.present?
+        if geo_hash_temp.present? && return_hash[:tgn].blank?
+          return_hash[:tgn] = geo_hash_temp[:tgn]
+        elsif geo_hash_temp.present? && geo_hash_temp[:tgn][:parse_depth] > return_hash[:tgn][:parse_depth]
+          return_hash[:tgn] = geo_hash_temp[:tgn]
         end
 
       end

data/test/geomash_test.rb

@@ -16,9 +16,9 @@ class GeomashTest < ActiveSupport::TestCase
     assert_equal 'France', result[:country_part]
     assert_equal nil, result[:street_part]
     assert_equal '7010587', result[:tgn][:id] if Geomash::TGN.tgn_enabled == true
-    assert_equal true, result[:tgn][:original_string_differs] if Geomash::TGN.tgn_enabled == true
-    assert_equal '2987374', result[:geonames][:id] if Geomash::Geonames.geonames_username != '<username>'
-    assert_equal true, result[:geonames][:original_string_differs] if Geomash::Geonames.geonames_username != '<username>'
+    assert_equal false, result[:tgn][:original_string_differs] if Geomash::TGN.tgn_enabled == true
+    #assert_equal '2987374', result[:geonames][:id] if Geomash::Geonames.geonames_username != '<username>'
+    #assert_equal true, result[:geonames][:original_string_differs] if Geomash::Geonames.geonames_username != '<username>'
 
     #Slight variation problem with neighborhood: 11. Bezirk (Vienna, Austria)--Biography
     result = Geomash.parse('15. Bezirk (Rudolfsheim-Fünfhaus, Vienna, Austria)--Exhibitions', true)
@@ -43,6 +43,40 @@ class GeomashTest < ActiveSupport::TestCase
     assert_equal '2953481', result[:geonames][:id] if Geomash::Geonames.geonames_username != '<username>'
     assert_equal true, result[:geonames][:original_string_differs] if Geomash::Geonames.geonames_username != '<username>'
 
+    result = Geomash.parse('Naroden Etnografski Muzeĭ (Sofia, Bulgaria)--Catalogs', true)
+    assert_equal 'Sofia', result[:city_part]
+    assert_equal 'Sofia', result[:state_part]
+    assert_equal 'Bulgaria', result[:country_part]
+    assert_equal nil, result[:neighborhood_part]
+    assert_equal nil, result[:street_part]
+    assert_equal '7009977', result[:tgn][:id] if Geomash::TGN.tgn_enabled == true
+    assert_equal true, result[:tgn][:original_string_differs] if Geomash::TGN.tgn_enabled == true
+    #assert_equal '727011', result[:geonames][:id] if Geomash::Geonames.geonames_username != '<username>'
+    #assert_equal true, result[:geonames][:original_string_differs] if Geomash::Geonames.geonames_username != '<username>'
+
+    result = Geomash.parse('Lettering--United States--History--19th century', true)
+    assert_equal nil, result[:city_part]
+    assert_equal nil, result[:state_part]
+    assert_equal 'United States', result[:country_part]
+    assert_equal nil, result[:neighborhood_part]
+    assert_equal nil, result[:street_part]
+    assert_equal '7012149', result[:tgn][:id] if Geomash::TGN.tgn_enabled == true
+    assert_equal false, result[:tgn][:original_string_differs] if Geomash::TGN.tgn_enabled == true
+    assert_equal '6252001', result[:geonames][:id] if Geomash::Geonames.geonames_username != '<username>'
+    assert_equal false, result[:geonames][:original_string_differs] if Geomash::Geonames.geonames_username != '<username>'
+
+    result = Geomash.parse('Lettering--United States--History--19th century', true)
+    assert_equal nil, result[:city_part]
+    assert_equal nil, result[:state_part]
+    assert_equal 'United States', result[:country_part]
+    assert_equal nil, result[:neighborhood_part]
+    assert_equal nil, result[:street_part]
+    assert_equal '7012149', result[:tgn][:id] if Geomash::TGN.tgn_enabled == true
+    assert_equal false, result[:tgn][:original_string_differs] if Geomash::TGN.tgn_enabled == true
+    assert_equal '6252001', result[:geonames][:id] if Geomash::Geonames.geonames_username != '<username>'
+    assert_equal false, result[:geonames][:original_string_differs] if Geomash::Geonames.geonames_username != '<username>'
+
+
   end
 
   def test_parse_with_no_flag
@@ -126,6 +160,17 @@ class GeomashTest < ActiveSupport::TestCase
     assert_equal nil, result[:street_part]
     assert_equal true, result[:tgn][:original_string_differs] if Geomash::TGN.tgn_enabled == true
 
+    #Case of a country with no states
+    result = Geomash.parse('Tokyo, Japan')
+    assert_equal 'Tokyo', result[:city_part]
+    assert_equal nil, result[:state_part]
+    assert_equal 'Japan', result[:country_part]
+    assert_equal nil, result[:neighborhood_part]
+    assert_equal '7004472', result[:tgn][:id] if Geomash::TGN.tgn_enabled == true
+    assert_equal false, result[:tgn][:original_string_differs] if Geomash::TGN.tgn_enabled == true
+    assert_equal '1850147', result[:geonames][:id] if Geomash::Geonames.geonames_username != '<username>'
+    assert_equal false, result[:geonames][:original_string_differs] if Geomash::Geonames.geonames_username != '<username>'
+
     #Should find the Michigan Atlanta over the Georgia Atlanta
     #State part from an API giving me Atlanta????
     result = Geomash.parse('Atlanta, MI')
@@ -138,7 +183,16 @@ class GeomashTest < ActiveSupport::TestCase
     assert_equal '4984500', result[:geonames][:id] if Geomash::Geonames.geonames_username != '<username>'
     assert_equal false, result[:geonames][:original_string_differs] if Geomash::Geonames.geonames_username != '<username>'
 
-
+    #TODO: This should also likely parse as North Korea as well...
+    result = Geomash.parse('Korea')
+    assert_equal nil, result[:city_part]
+    assert_equal nil, result[:state_part]
+    assert_equal 'South Korea', result[:country_part]
+    assert_equal nil, result[:neighborhood_part]
+    assert_equal '7000299', result[:tgn][:id] if Geomash::TGN.tgn_enabled == true
+    assert_equal false, result[:tgn][:original_string_differs] if Geomash::TGN.tgn_enabled == true
+    assert_equal '1835841', result[:geonames][:id] if Geomash::Geonames.geonames_username != '<username>'
+    assert_equal false, result[:geonames][:original_string_differs] if Geomash::Geonames.geonames_username != '<username>'
 
 
 

data/test/geonames_test.rb

@@ -11,7 +11,7 @@ class GeonamesTest < ActiveSupport::TestCase
       assert_equal '-84.18404', result[:coords][:box][:west]
       assert_equal '45.01697', result[:coords][:box][:north]
       assert_equal '-84.11884', result[:coords][:box][:east]
-      assert_equal '44.98859', result[:coords][:box][:south]
+      assert_equal '44.9886', result[:coords][:box][:south]
       assert_equal 'Atlanta', result[:hier_geo][:ppla2]
       assert_equal 'Montmorency County', result[:hier_geo][:adm2]
       assert_equal 'Michigan', result[:hier_geo][:adm1]

data/test/standardizer_test.rb

@@ -22,5 +22,42 @@ class ParserTest < ActiveSupport::TestCase
 
   end
 
+  def test_geographic_parser
+    #Nil results... problem cases
+    result = Geomash::Standardizer.parse_for_geographic_term('Yuma Indians')
+    assert_equal '', result
+
+    result = Geomash::Standardizer.parse_for_geographic_term('Yuma Indians--Woodworking')
+    assert_equal '', result
+
+    result = Geomash::Standardizer.parse_for_geographic_term('Norway maple')
+    assert_equal '', result
+
+    result = Geomash::Standardizer.parse_for_geographic_term('Some Value--German Engineering--Cars')
+    assert_equal '', result
+
+    result = Geomash::Standardizer.parse_for_geographic_term('Art, Japanese')
+    assert_equal '', result
+
+    #Normal cases
+    result = Geomash::Standardizer.parse_for_geographic_term('Palmer (Mass) - history or Stores (retail trade) - Palmer, Mass')
+    assert_equal 'Palmer, Mass', result
+
+    result = Geomash::Standardizer.parse_for_geographic_term('Naroden Etnografski Muzeĭ (Sofia, Bulgaria)--Catalogs')
+    assert_equal 'Naroden Etnografski Muzeĭ, Sofia, Bulgaria', result
+
+    result = Geomash::Standardizer.parse_for_geographic_term('Germany')
+    assert_equal 'Germany', result
+
+    result = Geomash::Standardizer.parse_for_geographic_term('United States')
+    assert_equal 'United States', result
+
+    result = Geomash::Standardizer.parse_for_geographic_term('South Korea')
+    assert_equal 'South Korea', result
+
+    result = Geomash::Standardizer.parse_for_geographic_term('Blah (North Korea)')
+    assert_equal 'Blah, North Korea', result
+  end
+
 
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: geomash
 version: !ruby/object:Gem::Version
-  version: 0.2.1
+  version: 0.3.1
 platform: ruby
 authors:
 - Boston Public Library
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-01-13 00:00:00.000000000 Z
+date: 2015-05-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -160,21 +160,23 @@ extra_rdoc_files: []
 files:
 - config/geomash.yml.sample
 - lib/geomash.rb
-- lib/geomash/geonames.rb
-- lib/geomash/parser.rb
-- lib/geomash/standardizer.rb
-- lib/geomash/version.rb
 - lib/geomash/tgn.rb
+- lib/geomash/autoexpire_cache_dalli.rb
 - lib/geomash/town_lookup.rb
+- lib/geomash/geonames.rb
 - lib/geomash/constants.rb
+- lib/geomash/version.rb
+- lib/geomash/standardizer.rb
+- lib/geomash/autoexpire_cache_redis.rb
+- lib/geomash/parser.rb
 - Rakefile
 - test/standardizer_test.rb
-- test/parser_test.rb
 - test/geomash_test.rb
-- test/geonames_test.rb
-- test/town_lookup_test.rb
 - test/test_helper.rb
 - test/tgn_test.rb
+- test/town_lookup_test.rb
+- test/parser_test.rb
+- test/geonames_test.rb
 homepage: http://www.bpl.org
 licenses: []
 metadata: {}
@@ -194,17 +196,16 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.1.10
+rubygems_version: 2.0.3
 signing_key: 
 specification_version: 4
 summary: Parse string for potential geographic matches and return that data along
   with the TGN ID and Geonames ID.
 test_files:
 - test/standardizer_test.rb
-- test/parser_test.rb
 - test/geomash_test.rb
-- test/geonames_test.rb
-- test/town_lookup_test.rb
 - test/test_helper.rb
 - test/tgn_test.rb
-has_rdoc: 
+- test/town_lookup_test.rb
+- test/parser_test.rb
+- test/geonames_test.rb