vector_embed 0.5.1 → 0.5.2

data/CHANGELOG

@@ -1,3 +1,9 @@
+0.5.2 / 2013-07-10
+
+* Bug fixes
+
+  * Treat nil as 0 in number feature labels (still disallowed in boolean labels)
+
 0.5.1 / 2013-07-09
 
 * Bug fixes

data/lib/vector_embed/maker/number.rb

@@ -36,7 +36,7 @@ class VectorEmbed
             FORMAT % num
           end
         when NilClass, NULL, SLASH_N
-          nil
+          keep_zero? ? 0 : nil
         else
           raise ArgumentError, "Can't embed #{v.inspect} in number feature #{k.inspect}"
         end

data/lib/vector_embed/version.rb

@@ -1,3 +1,3 @@
 class VectorEmbed
-  VERSION = '0.5.1'
+  VERSION = '0.5.2'
 end

data/spec/vector_embed_spec.rb

@@ -30,6 +30,25 @@ describe VectorEmbed do
       v.line(true)
       lambda { v.line(5.4) }.should raise_error(/Can't embed.*boolean/)
     end
+
+    it "stores nil as 0 in number" do
+      v = VectorEmbed.new
+      v.line(1)
+      v.line(nil).should == '0'
+    end
+
+    it "doesn't let you store nil in boolean" do
+      v = VectorEmbed.new
+      v.line(true)
+      lambda { v.line(nil) }.should raise_error(/Can't embed.*boolean/)
+    end
+
+    it "treats nil as evidence of number" do
+      v = VectorEmbed.new
+      v.line(nil).should == '0'
+      v.line(1).should == '1'
+    end
+
   end
 
   describe 'using a dictionary' do
@@ -170,6 +189,23 @@ describe VectorEmbed do
       v.line(3, 1 => '0').should == "3"
     end
 
+    it "DOES output blank strings in phrase attributes" do
+      v = VectorEmbed.new
+      v.line(3, 5 => 'hello').should == "3 #{l_h("5\x00hello")}:1"
+      v.line(3, 5 => nil).should == "3 #{l_h("5\x00")}:1"
+      v.line(3, 5 => '').should == "3 #{l_h("5\x00")}:1"
+      v.line(3, 5 => ' ').should == "3 #{l_h("5\x00")}:1"
+    end
+
+    it "strips OUTER whitespace" do
+      l_h("5\x00").should_not == l_h("5\x00 ")
+      l_h("5\x00 ").should_not == l_h("5\x00  ")
+      v = VectorEmbed.new
+      v.line(3, 5 => 'hello').should == "3 #{l_h("5\x00hello")}:1"
+      v.line(3, 5 => ' hello ').should == "3 #{l_h("5\x00hello")}:1"
+      v.line(3, 5 => ' hel lo ').should == "3 #{l_h("5\x00hel lo")}:1"
+    end
+
     it "treats nil like zero in number attributes" do
       v = VectorEmbed.new
       v.line(1, 1 => 1) # establish it's a number

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: vector_embed
 version: !ruby/object:Gem::Version
-  version: 0.5.1
+  version: 0.5.2
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-07-09 00:00:00.000000000 Z
+date: 2013-07-11 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: murmurhash3