iudex-simhash 1.0.0-java → 1.1.0-java

data/History.rdoc

@@ -1,2 +1,6 @@
+=== 1.1.0 (2011-11-13)
+* Update to iudex-html ~> 1.1.0
+* Update to minitest ~> 2.3
+
 === 1.0.0 (2011-04-04)
 * Initial release.

data/Manifest.txt

@@ -13,4 +13,4 @@ test/setup.rb
 test/test_fuzzy_set.rb
 test/test_simhash_generator.rb
 test/html/gentest.html
-lib/iudex-simhash/iudex-simhash-1.0.0.jar
+lib/iudex-simhash/iudex-simhash-1.1.0.jar

data/Rakefile

@@ -4,7 +4,7 @@ $LOAD_PATH << './lib'
 require 'iudex-simhash/base'
 
 require 'rubygems'
-gem     'rjack-tarpit', '~> 1.2'
+gem     'rjack-tarpit', '~> 1.4'
 require 'rjack-tarpit'
 
 t = RJack::TarPit.new( 'iudex-simhash',
@@ -13,10 +13,10 @@ t = RJack::TarPit.new( 'iudex-simhash',
 
 t.specify do |h|
   h.developer( "David Kellum", "dek-oss@gravitext.com" )
-  h.extra_deps += [ [ 'iudex-html',     '~> 1.0.0' ] ]
+  h.extra_deps += [ [ 'iudex-html',     '~> 1.1.0' ] ]
 
   h.testlib = :minitest
-  h.extra_dev_deps += [ [ 'minitest',      '>= 1.7.1', '< 2.1' ],
+  h.extra_dev_deps += [ [ 'minitest',      '~> 2.3' ],
                         [ 'rjack-logback', '~> 1.0' ] ]
 end
 

data/lib/iudex-simhash/base.rb

@@ -16,6 +16,6 @@
 
 module Iudex
   module SimHash
-    VERSION = '1.0.0'
+    VERSION = '1.1.0'
   end
 end

data/lib/iudex-simhash/factory_helper.rb

@@ -28,9 +28,10 @@ module Iudex
                                    'config', 'stopwords.en' )
 
         def simhash_stopwords( wfile = DEFAULT_WORDS )
-          words = File.open( wfile ) { |fin| fin.readlines }
-          words.map! { |w| w.strip }
-          words.reject! { |w| w =~ /^#/ }
+          words =
+            File.open( wfile ) { |fin| fin.readlines }.
+            map { |w| w.strip }.
+            reject { |w| w =~ /^#/ }
 
           Gen::StopWordSet.new( words )
         end
@@ -40,7 +41,9 @@ module Iudex
         def simhash_generator( input = :simhash_generator_inputs,
                                stopwords = simhash_stopwords )
 
-          inputs = send( input ).map { |r| r.to_a }.map do | key, ratio |
+          inputs = send( input ).
+            map { |r| Array( r ) }.
+            map do | key, ratio |
             key = key.to_k
             i = if( key.value_type == Element.java_class )
                   SimHashGenerator::Input.forTree( key )

data/pom.xml

@@ -3,13 +3,13 @@
   <groupId>iudex</groupId>
   <artifactId>iudex-simhash</artifactId>
   <packaging>jar</packaging>
-  <version>1.0.0</version>
+  <version>1.1.0</version>
   <name>Iudex simhash production and searching</name>
 
   <parent>
     <groupId>iudex</groupId>
     <artifactId>iudex-parent</artifactId>
-    <version>1.0</version>
+    <version>1.1</version>
     <relativePath>..</relativePath>
   </parent>
 
@@ -18,7 +18,7 @@
     <dependency>
       <groupId>iudex</groupId>
       <artifactId>iudex-html</artifactId>
-      <version>[1.0,1.1)</version>
+      <version>[1.1,1.2)</version>
     </dependency>
 
     <dependency>

data/test/test_fuzzy_set.rb

@@ -51,26 +51,26 @@ class TestFuzzySet < MiniTest::Unit::TestCase
 
   def test_match
     m = FuzzyList64.new( 100, 4 )
-    assert(   m.fuzzy_match( 0, 0 ) )
-    assert(   m.fuzzy_match( hex( '7FFF_FFFF_FFFF_FFFF' ),
-                             hex( '7FFF_FFFF_7777_FFFF' ) ) )
+    assert( m.fuzzy_match( 0, 0 ) )
+    assert( m.fuzzy_match( hex( '7FFF_FFFF_FFFF_FFFF' ),
+                           hex( '7FFF_FFFF_7777_FFFF' ) ) )
 
-    assert(   m.fuzzy_match( hex( 'FFFF_FFFF_FFFF_FFFF' ),
-                             hex( 'FFFF_FFFF_7777_FFFF' ) ) )
+    assert( m.fuzzy_match( hex( 'FFFF_FFFF_FFFF_FFFF' ),
+                           hex( 'FFFF_FFFF_7777_FFFF' ) ) )
 
-    assert( ! m.fuzzy_match( hex( '7FFF_FFFF_FFFF_FFFF' ),
-                             hex( '7FFF_FFFF_EFFF_7777' ) ) )
+    refute( m.fuzzy_match( hex( '7FFF_FFFF_FFFF_FFFF' ),
+                           hex( '7FFF_FFFF_EFFF_7777' ) ) )
 
-    assert( ! m.fuzzy_match( hex( 'FFFF_FFFF_FFFF_FFFF' ),
-                             hex( 'FFFF_FFFF_EFFF_7777' ) ) )
+    refute( m.fuzzy_match( hex( 'FFFF_FFFF_FFFF_FFFF' ),
+                           hex( 'FFFF_FFFF_EFFF_7777' ) ) )
   end
 
   def test_add
     m = FuzzyList64.new( 100, 4 )
-    assert(   m.addIfNotFound(  0x0 ) )
-    assert(   m.addIfNotFound( 0xFF ) )
-    assert( ! m.addIfNotFound( 0xFE ) )
-    assert( ! m.addIfNotFound(  0x1 ) )
+    assert( m.addIfNotFound(  0x0 ) )
+    assert( m.addIfNotFound( 0xFF ) )
+    refute( m.addIfNotFound( 0xFE ) )
+    refute( m.addIfNotFound(  0x1 ) )
   end
 
   def test_series_list
@@ -92,7 +92,7 @@ class TestFuzzySet < MiniTest::Unit::TestCase
     s = s.dup
     last = s.pop # Remove last for now
     assert_series_all( fset, s )
-    assert( ! fset.addIfNotFound( hex( last ) ), last )
+    refute( fset.addIfNotFound( hex( last ) ), last )
   end
 
   def assert_series_all( fset, s )
@@ -119,21 +119,21 @@ class TestFuzzySet < MiniTest::Unit::TestCase
     last = s.pop # Remove last for now
     assert_find_series_all( fset, s )
     l = Java::java.util.ArrayList.new;
-    assert( ! fset.addFindAll( hex( last ), l ) )
+    refute( fset.addFindAll( hex( last ), l ) )
     assert( l.size(), 1 );
 
     # Remove the match and try again.
     assert( fset.remove( l.get( 0 ) ), "remove match" )
     assert( fset.remove( hex( last ) ), "remove last" )
     l.clear
-    assert( ! fset.addFindAll( hex( last ), l ) )
+    refute( fset.addFindAll( hex( last ), l ) )
     assert( l.empty? )
   end
 
   def assert_find_series_all( fset, s )
     s.each do |k|
       l = Java::java.util.ArrayList.new;
-      assert( ! fset.addFindAll( hex( k ), l ) )
+      refute( fset.addFindAll( hex( k ), l ) )
       assert( l.empty? )
     end
   end

metadata

@@ -2,7 +2,7 @@
 name: iudex-simhash
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: 1.0.0
+  version: 1.1.0
 platform: java
 authors: 
   - David Kellum
@@ -10,8 +10,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-04-04 00:00:00 -07:00
-default_executable: 
+date: 2011-11-13 00:00:00 Z
 dependencies: 
   - !ruby/object:Gem::Dependency 
     name: iudex-html
@@ -21,7 +20,7 @@ dependencies:
       requirements: 
         - - ~>
           - !ruby/object:Gem::Version 
-            version: 1.0.0
+            version: 1.1.0
     type: :runtime
     version_requirements: *id001
   - !ruby/object:Gem::Dependency 
@@ -30,12 +29,9 @@ dependencies:
     requirement: &id002 !ruby/object:Gem::Requirement 
       none: false
       requirements: 
-        - - ">="
-          - !ruby/object:Gem::Version 
-            version: 1.7.1
-        - - <
+        - - ~>
           - !ruby/object:Gem::Version 
-            version: "2.1"
+            version: "2.3"
     type: :development
     version_requirements: *id002
   - !ruby/object:Gem::Dependency 
@@ -57,7 +53,7 @@ dependencies:
       requirements: 
         - - ~>
           - !ruby/object:Gem::Version 
-            version: 1.3.0
+            version: 1.4.0
     type: :development
     version_requirements: *id004
 description: |-
@@ -90,8 +86,8 @@ files:
   - test/test_fuzzy_set.rb
   - test/test_simhash_generator.rb
   - test/html/gentest.html
-  - lib/iudex-simhash/iudex-simhash-1.0.0.jar
-has_rdoc: true
+  - lib/iudex-simhash/iudex-simhash-1.1.0.jar
+  - .gemtest
 homepage: http://github.com/dekellum/iudex
 licenses: []
 
@@ -116,7 +112,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
 requirements: []
 
 rubyforge_project: iudex-simhash
-rubygems_version: 1.5.1
+rubygems_version: 1.8.9
 signing_key: 
 specification_version: 3
 summary: Iudex is a general purpose web crawler and feed processor in ruby/java