vector_embed 0.4.0 → 0.5.0

data/CHANGELOG

@@ -1,3 +1,13 @@
+0.5.0 / 2013-07-08
+
+* Breaking changes
+
+  * Treat .1 as 0.1
+
+* Bug fixes
+
+  * Fix error when mixing ngrams and phrases
+
 0.4.0 / 2013-07-01
 
 * Breaking changes

data/lib/vector_embed.rb

@@ -10,6 +10,7 @@ require 'vector_embed/stop_word'
 class VectorEmbed
   # http://stackoverflow.com/questions/638565/parsing-scientific-notation-sensibly
   JUST_A_NUMBER = /\A\s*[+\-]?(?:0|[1-9]\d*)(?:\.\d*)?(?:[eE][+\-]?\d+)?\s*\z/
+  UGLY_FLOAT = /\A\.\d+\z/
   BLANK = /\A\s*\z/
   NULL = /\Anull\z/i
   SLASH_N = '\N'

data/lib/vector_embed/maker/ngram.rb

@@ -9,6 +9,9 @@ class VectorEmbed
         end
       end
 
+      # TODO make sure you can't collide with these
+      IM_AN_NGRAM = 'ngram'
+
       attr_reader :len
       attr_reader :delim
 
@@ -20,7 +23,7 @@ class VectorEmbed
       end
 
       def pairs(v)
-        raise "Ngram can't handle #{v.inspect}, only a single string for now" unless v.is_a?(String)
+        raise "Ngram can't handle #{v.inspect}, only a single string for now" unless v.is_a?(::String)
         v = parent.preprocess v.to_s
         if len == 1
           # word mode
@@ -31,7 +34,7 @@ class VectorEmbed
         else
           raise "Word n-gram not supported yet"
         end.map do |ngram|
-          [ [ parent.index([k, 'ngram', ngram]), 1 ] ]
+          [ parent.index([k, IM_AN_NGRAM, ngram]), 1 ]
         end
       end
     end

data/lib/vector_embed/maker/number.rb

@@ -8,8 +8,8 @@ class VectorEmbed
           case v
           when Numeric, NilClass, NULL, SLASH_N
             true
-          else
-            v =~ JUST_A_NUMBER
+          when String
+            v =~ JUST_A_NUMBER or v =~ UGLY_FLOAT
           end
         end
 
@@ -31,7 +31,7 @@ class VectorEmbed
 
       def value(v)
         case v
-        when Numeric, JUST_A_NUMBER
+        when Numeric, JUST_A_NUMBER, UGLY_FLOAT
           Number.numify v
         when NilClass, NULL, SLASH_N
           nil

data/lib/vector_embed/version.rb

@@ -1,3 +1,3 @@
 class VectorEmbed
-  VERSION = '0.4.0'
+  VERSION = '0.5.0'
 end

data/spec/vector_embed_spec.rb

@@ -136,6 +136,11 @@ describe VectorEmbed do
       v.line(1, 1 => '9e9').should == "1 #{l_h('1')}:9000000000"
     end
 
+    it "is ok with floats that don't start with 0" do
+      v = VectorEmbed.new
+      v.line(3, 5 => '.9').should == "3 #{l_h('5')}:0.9"
+    end
+
     it "stores dates as days since 2000-01-01" do
       v = VectorEmbed.new
       v.line(1, 3 => Date.new(1999,12,31)).should == "1 #{l_h('3')}:-1"
@@ -295,6 +300,12 @@ describe VectorEmbed do
       v.line(1, 1 => 'foo', 2 => 'bar', 'baz' => 'zoo').should == sortme("1 #{l_h("1\x00foo")}:1 #{l_h("2\x00bar")}:1 #{l_h("baz\x00zoo")}:1")
     end
 
+    it "simultaneously produces phrases and n-grams" do
+      v = VectorEmbed.new ngram_len: 2, ngram_delim: '', features: { 2 => :Phrase }
+      v.line(3, 1 => 'foo', 2 => 'bar').should == sortme("3 #{l_h("1\x00ngram\x00fo")}:1 #{l_h("1\x00ngram\x00oo")}:1 #{l_h("2\x00bar")}:1")
+
+    end
+
   end
 
   describe 'hinting' do

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: vector_embed
 version: !ruby/object:Gem::Version
-  version: 0.4.0
+  version: 0.5.0
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-07-01 00:00:00.000000000 Z
+date: 2013-07-08 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: murmurhash3