mort666_simhash 0.2.7 → 0.2.8

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 3b3b62e04b6e50547de7e7edf3441da79ce41869
-  data.tar.gz: dde5d235ee1d2bbeaa9304fe235768883c6525cf
+  metadata.gz: e7b5060a56cd39db1a85ff2bca91022d8279742f
+  data.tar.gz: 1501e136b7d5c32e11580f42d13ce5cae8a38682
 SHA512:
-  metadata.gz: c333facb22f608bf1caec2de3023f30ccc552f116a323063819dfcfe2bc49d1bc397a3f930d3ac8caa4e19d1f6308542096b4732e5ae308a9c1dc241083674ee
-  data.tar.gz: 5c1ef0e033e82009858a7136e23c01936ce7083c551783c8987f36a31a294e0a1babdc82e2bd3a943c60a444bb724fe09affc2b1998495b2172dd8d58b1e06cf
+  metadata.gz: 9f851149712bb5384d1657b3c87cfa90fc81ae0775118a2651ccd710016f980d1377685f55d8a9bc5210204bbbf860e206c80896996376da8ff669833e924d39
+  data.tar.gz: 8c3eee0868acdbf881cf61727a82388f17509bb91c57de6926d0fe409ca4734e3f51aa309a976438a15c9bd095e4205b73e4a5dd2d7863ac6587679f1d7e1875

data/Rakefile

@@ -23,7 +23,7 @@ end
 desc 'Start an IRB session with all necessary files required.'
 task :shell do |t|
   chdir File.dirname(__FILE__)
-  exec 'irb -I lib/ -I lib/simhash -I lib/string -I lib/integer -r rubygems'
+  exec 'irb -I lib/ -I lib/simhash -I lib/simhash/string -I lib/simhash/integer -r rubygems'
 end
 
 desc 'Build the gemspec.'

data/lib/{string.rb → simhash/string.rb}

@@ -2,7 +2,7 @@ class String
   def simhash(options={})
     split_by = options.delete(:split_by) || " "
 
-    # Do the punctuation clean before the split.. 
+    # Do the punctuation clean before the split..
     # You could argue this is not preserving the meaning doing so actually preserves the edge case where a hyphen is removed
     # resulting hash does not match the same string with a space in there before the split
     if options[:preserve_punctuation]
@@ -11,7 +11,7 @@ class String
       Simhash.hash(self.gsub(Simhash::PUNCTUATION_REGEXP, ' ') .split(split_by), options)
     end
   end
-  
+
   def hash_vl_rb(length)
     return 0 if self == ""
 
@@ -25,4 +25,4 @@ class String
     x
   end
 
-end
+end

data/lib/simhash.rb

@@ -3,23 +3,23 @@
 require 'active_support/core_ext/string/multibyte'
 require 'unicode'
 
-require 'string'
-require 'integer'
+require 'simhash/string'
+require 'simhash/integer'
 require 'simhash/stopwords'
 
-begin
+#begin
   require 'string_hashing'
-rescue LoadError
-end
+#rescue LoadError
+#end
 
-module Simhash  
+module Simhash
   DEFAULT_STRING_HASH_METHOD = String.public_instance_methods.include?("hash_vl") ? :hash_vl : :hash_vl_rb
   PUNCTUATION_REGEXP = if RUBY_VERSION >= "1.9"
     /(\s|\d|[^\p{L}]|\302\240| *— *|[«»…\-–—]| )+/u
   else
     /(\s|\d|\W|\302\240| *— *|[«»…\-–—]| )+/u
   end
-  
+
   # Compare calculates the Hamming distance between two 64-bit integers
   #
   # Currently, this is calculated using the Kernighan method [1]. Other methods
@@ -40,25 +40,25 @@ module Simhash
   def self.hash(tokens, options={})
     hashbits = options[:hashbits] || 64
     hashing_method = options[:hashing_method] || DEFAULT_STRING_HASH_METHOD
-        
+
     v = [0] * hashbits
     masks = v.dup
     masks.each_with_index {|e, i| masks[i] = (1 << i)}
-    
+
     self.each_filtered_token(tokens, options) do |token|
       hashed_token = token.send(hashing_method, hashbits).to_i
       hashbits.times do |i|
         v[i] += (hashed_token & masks[i]).zero? ? -1 : +1
       end
     end
-   
+
     fingerprint = 0
 
-    hashbits.times { |i| fingerprint += 1 << i if v[i] >= 0 }  
-      
-    fingerprint    
+    hashbits.times { |i| fingerprint += 1 << i if v[i] >= 0 }
+
+    fingerprint
   end
-  
+
   def self.each_filtered_token(tokens, options={})
     token_min_size = options[:token_min_size].to_i
     stop_sentenses = options[:stop_sentenses]
@@ -66,28 +66,28 @@ module Simhash
       # cutting punctuation (\302\240 is unbreakable space)
       # Moved up to string class
       # token = token.gsub(PUNCTUATION_REGEXP, ' ') if !options[:preserve_punctuation]
-      
+
       token = Unicode::downcase(token.strip)
-      
+
       # cutting stop-words
       token = token.split(" ").reject{ |w| Stopwords::ALL.index(" #{w} ") != nil }.join(" ") if options[:stop_words]
-    
+
       # cutting stop-sentenses
       next if stop_sentenses && stop_sentenses.include?(" #{token} ")
-            
+
       next if token.size.zero? || token.mb_chars.size < token_min_size
-      
-      yield token      
+
+      yield token
     end
   end
-  
+
   def self.filtered_tokens(tokens, options={})
     filtered_tokens = []
     self.each_filtered_token(tokens, options) { |token| filtered_tokens << token }
-    filtered_tokens  
+    filtered_tokens
   end
-  
+
   def self.hm
     @@string_hash_method
   end
-end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: mort666_simhash
 version: !ruby/object:Gem::Version
-  version: 0.2.7
+  version: 0.2.8
 platform: ruby
 authors:
 - Alex Gusev
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-09-26 00:00:00.000000000 Z
+date: 2017-08-30 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -109,12 +109,12 @@ files:
 - Rakefile
 - ext/string_hashing/extconf.rb
 - ext/string_hashing/string_hashing.c
-- lib/integer.rb
 - lib/simhash.rb
+- lib/simhash/integer.rb
 - lib/simhash/stopwords.rb
 - lib/simhash/stopwords/en.rb
 - lib/simhash/stopwords/ru.rb
-- lib/string.rb
+- lib/simhash/string.rb
 homepage: http://github.com/mort666/simhash
 licenses: []
 metadata: {}
@@ -134,7 +134,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.4.3
+rubygems_version: 2.6.12
 signing_key: 
 specification_version: 4
 summary: 'Gives you possbility to convert string into simhashes to futher use: finding