rubyfish 0.0.4 → 0.0.5

data/README.md

@@ -1,21 +1,20 @@
-
 RubyFish
 =========
 
-RubyFish is a ruby port of python library jellyfish (http://github.com/sunlightlabs/jellyfish) for doing approximate and phonetic matching of strings.
+RubyFish is a ruby port of python library <a href = "http://github.com/sunlightlabs/jellyfish">jellyfish</a> for doing approximate and phonetic matching of strings.
+
+-------------
 
 RubyFish is a project of AnjLab (c) 2010.
 All code is released under a BSD-style license, see LICENSE for details.
 
-Originally written by
-Written by Michael Stephens <mstephens@sunlightfoundation.com> and James Turk
-<jturk@sunlightfoundation.com>.
+Originally written by <a href="mailto:mstephens@sunlightfoundation.com">Michael Stephens</a> and <a href="mailto:jturk@sunlightfoundation.com">James Turk</a>.
 
-Ported by Yury Korolev <yury.korolev@gmail.com>
-Source is available at http://github.com/anjlab/rubyfish
+Ported by <a href="mailto:yury.korolev@gmail.com">Yury Korolev</a>
+Source is available on <a href="http://github.com/anjlab/rubyfish">GitHub</a>
 
 Included Algorithms
-===================
+-------------------
 
 String comparison:
 
@@ -32,7 +31,7 @@ Phonetic encoding:
   * Double Metaphone
 
 Example Usage
-=============
+-------------
 
     ruby-1.9.2-p0 > require 'rubyfish'
     ruby-1.9.2-p0 > RubyFish::Levenshtein.distance("jellyfish", "rubyfish")

data/lib/rubyfish/damerau_levenshtein.rb

@@ -1,41 +1,59 @@
-require 'matrix'
-
 module RubyFish::DamerauLevenshtein
 
-  def distance a, b
+  # http://en.wikipedia.org/wiki/Levenshtein_distance
+  # http://en.wikipedia.org/wiki/Damerau%E2%80%93Levenshtein_distance
+
+  def _distance a, b, opts = {}
+    allow_swaps = opts[:allow_swaps]
+    ignore_case = opts[:ignore_case]
+    
     as = a.to_s
     bs = b.to_s
+    
+    if ignore_case
+    	as.downcase!
+    	bs.downcase!
+    end
 
     rows = as.size + 1
     cols = bs.size + 1
 
-    dist = ::RubyFish::MMatrix.new(rows, cols)
-
-    (0...rows).each {|i| dist[i, 0] = i}
-    (0...cols).each {|j| dist[0, j] = j}
-
+    dist = [
+        Array.new(cols) {|k| k},
+        Array.new(cols) {0},
+        Array.new(cols) {0}
+    ]
+	
     (1...rows).each do |i|
+      k = i % 3
+      dist[k][0] = i
+
       (1...cols).each do |j|
         cost = as[i - 1] == bs[j - 1] ? 0 : 1
 
         #minimum of deletion, insertion, substitution
-        d1 = dist[i - 1, j] + 1
-        d2 = dist[i, j - 1] + 1
-        d3 = dist[i - 1, j - 1] + cost
+        d1 = dist[k - 1][j] + 1
+        d2 = dist[k][j - 1] + 1
+        d3 = dist[k - 1][j - 1] + cost
 
         d_now = [d1, d2, d3].min
 
-        if i > 2 && j > 2 && as[i - 1] == bs[j - 2] && as[i - 2] == bs[j - 1]
-          d1 = dist[i - 2, j - 2] + cost
+        if allow_swaps && i > 2 && j > 2 && as[i - 1] == bs[j - 2] && as[i - 2] == bs[j - 1]
+          d1 = dist[k - 2][j - 2] + cost
           d_now = [d_now, d1].min;
         end
 
-        dist[i, j] = d_now;
+        dist[k][j] = d_now;
       end
     end
 
-    dist[as.size, bs.size]
+    dist[(rows - 1) % 3][-1]
+  end
+
+  def distance a, b, opts = {}
+    _distance(a, b, :allow_swaps => true, :ignore_case => opts[:ignore_case])
   end
 
   module_function :distance
+  module_function :_distance
 end

data/lib/rubyfish/hamming.rb

@@ -1,7 +1,13 @@
 module RubyFish::Hamming
-  def distance a, b
+  def distance a, b, opts={}
+    ignore_case = opts[:ignore_case]
     distance = 0
     as, bs = a.to_s, b.to_s
+    
+    if ignore_case
+    	as.downcase!
+    	bs.downcase!
+    end
 
     short, long = [as, bs].sort
 

data/lib/rubyfish/jaro.rb

@@ -2,8 +2,8 @@ module RubyFish::Jaro
   include ::RubyFish::JaroWinkler
   extend ::RubyFish::JaroWinkler
 
-  def distance a, b
-    _distance(a, b, :winklerize => false)
+  def distance a, b, opts={}
+    _distance(a, b, :winklerize => false, :ignore_case => opts[:ignore_case])
   end
 
   module_function :distance

data/lib/rubyfish/jaro_winkler.rb

@@ -3,9 +3,15 @@ module RubyFish::JaroWinkler
   def _distance a, b, opts = {}
     long_tolerance = opts[:long_tolerance]
     winklerize = opts[:winklerize]
+    ignore_case = opts[:ignore_case]
 
     as = a.to_s
     bs = b.to_s
+    
+    if ignore_case
+    	as.downcase!
+    	bs.downcase!
+    end
 
     as_length = as.size
     bs_length = bs.size
@@ -98,7 +104,7 @@ module RubyFish::JaroWinkler
   end
 
   def distance a, b, opts = {}
-    _distance(a, b, :winklerize => true)
+    _distance(a, b, :winklerize => true, :ignore_case => opts[:ignore_case])
   end
 
   module_function :distance

data/lib/rubyfish/levenshtein.rb

@@ -1,36 +1,13 @@
 module RubyFish::Levenshtein
 
-  # http://en.wikipedia.org/wiki/Levenshtein_distance
-  def distance a, b
-    as, bs = a.to_s, b.to_s
+  include ::RubyFish::DamerauLevenshtein
+  extend ::RubyFish::DamerauLevenshtein
 
-    as_length = as.size
-    bs_length = bs.size
-
-    rows = as_length + 1
-    cols = bs_length + 1
-
-    dist = ::RubyFish::MMatrix.new rows, cols
-    (0...rows).each {|i| dist[i, 0] = i}
-    (0...cols).each {|j| dist[0, j] = j}
-
-    (1...cols).each do |j|
-      (1...rows).each do |i|
-        if as[i - 1] == bs[j - 1]
-         dist[i, j] = dist[i - 1, j - 1]
-        else
-          d1 = dist[i - 1, j] + 1
-          d2 = dist[i, j - 1] + 1
-          d3 = dist[i - 1, j - 1] + 1
-
-          dist[i, j] = [d1, d2, d3].min;
-        end
-      end
-    end
-
-    dist[as_length, bs_length];
+  def distance a, b, opts={}
+    _distance(a, b, :allowswaps => false, :ignore_case => opts[:ignore_case])
   end
 
   module_function :distance
 
+
 end

data/lib/rubyfish/longest_subsequence.rb

@@ -1,10 +1,17 @@
 module RubyFish::LongestSubsequence
 
   # http://en.wikibooks.org/wiki/Algorithm_Implementation/Strings/Longest_common_subsequence#Ruby
-  def distance a, b
+  def distance a, b, opts={}
+    ignore_case = opts[:ignore_case]
+    
     as = a.to_s
     bs = b.to_s
 
+    if ignore_case
+    	as.downcase!
+    	bs.downcase!
+    end
+    
     rows = as.size
     cols = bs.size
 

data/lib/rubyfish/longest_substring.rb

@@ -1,9 +1,16 @@
 module RubyFish::LongestSubstring
 
   # http://en.wikibooks.org/wiki/Algorithm_implementation/Strings/Longest_common_substring#Ruby
-  def distance a, b
+  def distance a, b, opts={}
+  	ignore_case = opts[:ignore_case]
+  	
     as = a.to_s
     bs = b.to_s
+    
+    if ignore_case
+    	as.downcase!
+    	bs.downcase!
+    end
 
     rows = as.size
     cols = bs.size
@@ -29,9 +36,16 @@ module RubyFish::LongestSubstring
     ans
   end
 
-  def longest_substring a, b
+  def longest_substring a, b, opts={}
+  	ignore_case = opts[:ignore_case]
+  	
     as = a.to_s
     bs = b.to_s
+    
+    if ignore_case
+    	as.downcase!
+    	bs.downcase!
+    end
 
     rows = as.size
     cols = bs.size
@@ -70,8 +84,8 @@ module RubyFish::LongestSubstring
     res
   end
 
-  def longest_substring_index(a, b)
-    a.index(longest_substring(a, b))
+  def longest_substring_index(a, b, opts={})
+    a.index(longest_substring(a, b, :ignore_case => opts[:ignore_case]))
   end
 
   module_function :distance

data/lib/rubyfish/version.rb

@@ -1,3 +1,3 @@
 module RubyFish
-  VERSION = "0.0.4"
+  VERSION = "0.0.5"
 end

metadata

@@ -1,33 +1,23 @@
---- !ruby/object:Gem::Specification 
+--- !ruby/object:Gem::Specification
 name: rubyfish
-version: !ruby/object:Gem::Version 
-  prerelease: false
-  segments: 
-  - 0
-  - 0
-  - 4
-  version: 0.0.4
+version: !ruby/object:Gem::Version
+  version: 0.0.5
+  prerelease: 
 platform: ruby
-authors: 
+authors:
 - Yury Korolev
 autorequire: 
 bindir: bin
 cert_chain: []
-
-date: 2010-09-20 00:00:00 +04:00
-default_executable: 
+date: 2012-08-16 00:00:00.000000000 Z
 dependencies: []
-
 description: Port of http://github.com/sunlightlabs/jellyfish
-email: 
+email:
 - yury.korolev@gmail.com
 executables: []
-
 extensions: []
-
 extra_rdoc_files: []
-
-files: 
+files:
 - lib/rubyfish/damerau_levenshtein.rb
 - lib/rubyfish/double_metaphone.rb
 - lib/rubyfish/hamming.rb
@@ -43,40 +33,31 @@ files:
 - CHANGELOG.md
 - README.md
 - ROADMAP.md
-has_rdoc: true
 homepage: http://github.com/anjlab/rubyfish
 licenses: []
-
 post_install_message: 
 rdoc_options: []
-
-require_paths: 
+require_paths:
 - lib
-required_ruby_version: !ruby/object:Gem::Requirement 
+required_ruby_version: !ruby/object:Gem::Requirement
   none: false
-  requirements: 
-  - - ">="
-    - !ruby/object:Gem::Version 
-      hash: 3404657759252333384
-      segments: 
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+      segments:
       - 0
-      version: "0"
-required_rubygems_version: !ruby/object:Gem::Requirement 
+      hash: -749425856707131905
+required_rubygems_version: !ruby/object:Gem::Requirement
   none: false
-  requirements: 
-  - - ">="
-    - !ruby/object:Gem::Version 
-      segments: 
-      - 1
-      - 3
-      - 6
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
       version: 1.3.6
 requirements: []
-
 rubyforge_project: rubyfish
-rubygems_version: 1.3.7
+rubygems_version: 1.8.24
 signing_key: 
 specification_version: 3
 summary: Library for doing approximate and phonetic matching of string
 test_files: []
-