string_metric 0.1.1 → 0.1.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 229ed953c5aadc5c929cb2b47ad7898cdc19d742
-  data.tar.gz: b41a66496dd7c0cb2627af22b9764d2bf8d06c37
+  metadata.gz: 08ed461c4edc7dfd5e32fb34e93b13b4442c5167
+  data.tar.gz: 69d6cd7273238442f184d6a63fec949ffcc6f6db
 SHA512:
-  metadata.gz: c0e932fbe3e03ec7b8268a2cc85f276b7b0041cdaf8fea5fb98d7e8a966ded52280fbe51b01a89da4e26430020a4995cc3b043110f2a95731e70c07ad3a30748
-  data.tar.gz: 19f270701a5e4383b84ab089ea534e7930fb980ff7dba494c51201436c69408cf54b0cca7d64098aefdd24214cc6e250168a1dc975abbf49e3672ba262b7eead
+  metadata.gz: f5d7e8664514c5f31075755c3b09b2a7c4a0cd963601aefd26f1b928426dea2051a12b8abd039dddb0d8986fe25c6ce860d8c4a867757a4bf5b5a1883c6ace1e
+  data.tar.gz: d9b631aad62da262d7a9afec9c7c1295e8b238cbfe4ef432354e0a2c42d9bf3b777758cf00abe2b14b702069eeb335df505ba9db37faa165335d748f4d748f2e

data/lib/string_metric/levenshtein/experiment.rb

@@ -4,10 +4,6 @@ module StringMetric
   module Levenshtein
     class Experiment
       def self.distance(from, to, options = {})
-        return 0 if from == to
-        return to.size if from.size.zero?
-        return from.size if to.size.zero?
-
         m = from.length
         n = to.length
 
@@ -25,4 +21,4 @@ module StringMetric
       end
     end
   end
-end
+end

data/lib/string_metric/levenshtein/iterative_with_full_matrix.rb

@@ -4,24 +4,32 @@ module StringMetric
   module Levenshtein
     class IterativeWithFullMatrix
       def self.distance(from, to, options = {})
-        return 0 if from == to
-        return to.size if from.size.zero?
-        return from.size if to.size.zero?
-
         max_distance      = options[:max_distance]
         insertion_cost    = options.fetch(:insertion_cost, 1)
         deletion_cost     = options.fetch(:deletion_cost, 1)
         substitution_cost = options.fetch(:substitution_cost, 1)
 
-        d = (0..to.size).map do |i|
-          [0] * (from.size + 1)
+        m = from.length
+        n = to.length
+
+        if max_distance && (m - n).abs >= max_distance
+          return max_distance
+        end
+
+        return 0 if from == to
+        return n if m.zero?
+        return m if n.zero?
+
+        d = (0..n).map do |i|
+          [0] * (m + 1)
         end
 
-        (1..from.size).each { |j| d[0][j] = j }
-        (1..to.size).each { |i| d[i][0] = i }
+        (1..m).each { |j| d[0][j] = j }
+        (1..n).each { |i| d[i][0] = i }
 
-        (1..from.size).each do |j|
-          (1..to.size).each do |i|
+        to_column = 0
+        (1..m).each do |j|
+          (1..n).each do |i|
             if from[j-1] == to[i-1]
               d[i][j] = d[i -1][j-1]
             else
@@ -30,12 +38,13 @@ module StringMetric
                          d[i-1][j-1] + substitution_cost  # substitution
                         ].min
             end
+            to_column = i
           end
 
-          break if max_distance and d[j][j] > max_distance
+          break if max_distance and d[to_column].min > max_distance
         end
 
-        x = d[to.size][from.size]
+        x = d[n][m]
         if max_distance && x > max_distance
           max_distance
         else

data/lib/string_metric/levenshtein/iterative_with_two_matrix_rows.rb

@@ -4,10 +4,6 @@ module StringMetric
   module Levenshtein
     class IterativeWithTwoMatrixRows
       def self.distance(from, to, options = {})
-        return 0 if from == to
-        return to.size if from.size.zero?
-        return from.size if to.size.zero?
-
         max_distance      = options[:max_distance]
         insertion_cost    = options.fetch(:insertion_cost, 1)
         deletion_cost     = options.fetch(:deletion_cost, 1)
@@ -16,13 +12,19 @@ module StringMetric
         m = from.length
         n = to.length
 
+        if max_distance && (n - m).abs >= max_distance
+          return max_distance
+        end
+
+        return 0 if from == to
+        return n if m.zero?
+        return m if n.zero?
+
         v0 = (0..m).to_a
-        v1 = []
         x = 0
 
         n.times do |i|
-          x = v1[0] = i + 1
-
+          current = x = i + 1
           sub_cell = v0[0]
 
           m.times do |j|
@@ -30,20 +32,18 @@ module StringMetric
 
             ins_cell = v0[j+1]
 
-            x = [x + deletion_cost,         # deletion
+            x = [current + deletion_cost,   # deletion
                  ins_cell + insertion_cost, # insertion
                  sub_cell + cost            # substitution
                 ].min
 
-
-            v1[j + 1] = x
-
+            v0[j] = current
+            current = x
             sub_cell = ins_cell
           end
 
-          break if max_distance && v0[i] > max_distance
-
-          v0 = v1.dup
+          v0[m] = x
+          break if max_distance && v0.min > max_distance
         end
 
         if max_distance && x > max_distance
@@ -54,4 +54,4 @@ module StringMetric
       end
     end
   end
-end
+end

data/lib/string_metric/levenshtein/iterative_with_two_matrix_rows_optimized.rb

@@ -4,10 +4,6 @@ module StringMetric
   module Levenshtein
     class IterativeWithTwoMatrixRowsOptimized
       def self.distance(from, to, options = {})
-        return 0 if from == to
-        return to.size if from.size.zero?
-        return from.size if to.size.zero?
-
         max_distance      = options[:max_distance]
         insertion_cost    = options[:insertion_cost]    || 1
         deletion_cost     = options[:deletion_cost]     || 1
@@ -16,36 +12,40 @@ module StringMetric
         m = from.length
         n = to.length
 
+        if max_distance && (n - m).abs >= max_distance
+          return max_distance
+        end
+
+        return 0 if from == to
+        return n if m.zero?
+        return m if n.zero?
+
         from = from.codepoints.to_a
         to = to.codepoints.to_a
 
         v0 = (0..m).to_a
-        v1 = []
         x = 0
 
         n.times do |i|
-          x = v1[0] = i + 1
-
+          current = x = i + 1
           sub_cell = v0[0]
 
           m.times do |j|
             cost = (from[j] == to[i]) ? 0 : substitution_cost
-
             ins_cell = v0[j + 1]
 
-            x = [x + deletion_cost,         # deletion
+            x = [current + deletion_cost,   # deletion
                  ins_cell + insertion_cost, # insertion
                  sub_cell + cost            # substitution
                 ].sort![0]
 
-            v1[j + 1] = x
-
+            v0[j] = current
+            current = x
             sub_cell = ins_cell
           end
 
-          break if max_distance && v0[i] > max_distance
-
-          v0 = v1.dup
+          v0[m] = x
+          break if max_distance && v0.sort[0] > max_distance
         end
 
         if max_distance && x > max_distance

data/lib/string_metric/levenshtein/recursive.rb

@@ -4,15 +4,19 @@ module StringMetric
   module Levenshtein
     class Recursive
       def self.distance(from, to, options = {})
-        return 0 if from == to
-        return to.size if from.size.zero?
-        return from.size if to.size.zero?
-
         max_distance      = options[:max_distance]
         insertion_cost    = options.fetch(:insertion_cost, 1)
         deletion_cost     = options.fetch(:deletion_cost, 1)
         substitution_cost = options.fetch(:substitution_cost, 1)
 
+        if max_distance && (from.size - to.size).abs >= max_distance
+          return max_distance
+        end
+
+        return 0 if from == to
+        return to.size if from.size.zero?
+        return from.size if to.size.zero?
+
         if from.chars.to_a.last == to.chars.to_a.last
           cost = 0
         else
@@ -35,4 +39,4 @@ module StringMetric
       end
     end
   end
-end
+end

data/lib/string_metric/version.rb

@@ -1,3 +1,3 @@
 module StringMetric
-  VERSION = "0.1.1"
+  VERSION = "0.1.2"
 end

data/spec/fixtures/levenshtein.csv

@@ -6,6 +6,9 @@ hello, hellol, 1
 hello, heloll, 2
 hello, cheese, 4
 hello, saint, 5
+hellol, hello, 1
 hello,, 5
+, hello, 5
+, , 0
 sitting, kitten, 3
-αλφα, βητα, 3
+αλφα, βητα, 3

data/spec/support/levenshtein.rb

@@ -24,11 +24,48 @@ shared_examples "Levenshtein Distance" do |options|
 
     context "when max_distance is passed as option" do
       context "and normal distance is greater than max_distance" do
-        let(:max_distance) { 2 }
-
         it "is trimmed to max_distance" do
-          expect(described_class.distance("kitten", "sitting",
-            max_distance: max_distance)).to eq max_distance
+          expect(described_class.distance("gumbo", "gambol", max_distance: 1)).to eq 1
+          expect(described_class.distance("test", "tasf", max_distance: 1)).to eq 1
+          expect(described_class.distance("kitten", "sitting", max_distance: 2)).to eq 2
+          expect(described_class.distance("kitten", "kittenss", max_distance: 1)).to eq 1
+          expect(described_class.distance("kittenss", "kitten", max_distance: 1)).to eq 1
+          expect(described_class.distance("sitting", "kitten", max_distance: 2)).to eq 2
+          expect(described_class.distance("gambol", "gumbo", max_distance: 1)).to eq 1
+          expect(described_class.distance("kitten", "", max_distance: 2)).to eq 2
+          expect(described_class.distance("", "kitten", max_distance: 3)).to eq 3
+        end
+      end
+      context "and normal distance is less than max_distance" do
+        it "is calculated distance" do
+          expect(described_class.distance("", "t", max_distance: 2)).to eq 1
+          expect(described_class.distance("t", "", max_distance: 3)).to eq 1
+          expect(described_class.distance("test", "test", max_distance: 1)).to eq 0
+          expect(described_class.distance("test", "tent", max_distance: 2)).to eq 1
+          expect(described_class.distance("gumbo", "gambol", max_distance: 3)).to eq 2
+          expect(described_class.distance("kitten", "sitting", max_distance: 4)).to eq 3
+          expect(described_class.distance("kitten", "kittenss", max_distance: 4)).to eq 2
+          expect(described_class.distance("kittenss", "kitten", max_distance: 4)).to eq 2
+          expect(described_class.distance("sitting", "kitten", max_distance: 4)).to eq 3
+          expect(described_class.distance("gambol", "gumbo", max_distance: 3)).to eq 2
+          expect(described_class.distance("", "cat", max_distance: 4)).to eq 3
+          expect(described_class.distance("cat", "", max_distance: 5)).to eq 3
+          expect(described_class.distance("", "", max_distance: 2)).to eq 0
+        end
+      end
+      context "and normal distance is same as max_distance" do
+        it "is calculated distance" do
+          expect(described_class.distance("test", "test", max_distance: 0)).to eq 0
+          expect(described_class.distance("test", "tent", max_distance: 1)).to eq 1
+          expect(described_class.distance("gumbo", "gambol", max_distance: 2)).to eq 2
+          expect(described_class.distance("kitten", "sitting", max_distance: 3)).to eq 3
+          expect(described_class.distance("kitten", "kittenss", max_distance: 2)).to eq 2
+          expect(described_class.distance("kittenss", "kitten", max_distance: 2)).to eq 2
+          expect(described_class.distance("sitting", "kitten", max_distance: 3)).to eq 3
+          expect(described_class.distance("gambol", "gumbo", max_distance: 2)).to eq 2
+          expect(described_class.distance("", "cat", max_distance: 3)).to eq 3
+          expect(described_class.distance("cat", "", max_distance: 3)).to eq 3
+          expect(described_class.distance("", "", max_distance: 0)).to eq 0
         end
       end
     end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: string_metric
 version: !ruby/object:Gem::Version
-  version: 0.1.1
+  version: 0.1.2
 platform: ruby
 authors:
 - Giorgos Tsiftsis
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-02-03 00:00:00.000000000 Z
+date: 2015-04-13 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -133,7 +133,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.2.0
+rubygems_version: 2.4.6
 signing_key: 
 specification_version: 4
 summary: A simple library with String Metric algorithms
@@ -147,3 +147,4 @@ test_files:
 - spec/lib/levenshtein_spec.rb
 - spec/spec_helper.rb
 - spec/support/levenshtein.rb
+has_rdoc: