zipf 1.0.3 → 1.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 4aac67c7168dbcf85aa211ec579cd47316ba3e3d
-  data.tar.gz: 5263572ce34158227f6615ead03755e36a3175f6
+  metadata.gz: fdbe716da58c6e94e0f7066226378cc113799088
+  data.tar.gz: 9c7a4d0c26ad22a5df8e51a3b794b3bcbf348883
 SHA512:
-  metadata.gz: 587d0e910acf176725667123b9cfffcd1692decaaddb11cb1656066d08dd862da120a86065c3e28b5f9e8fd79d7c86e9c3efd90654cb396133d023da2e9c7b63
-  data.tar.gz: 39688a72137f1c1741a442b168002defd1134394ec87e25c0101ff38cf1cef03ebbccab70053c2f75eae6084f5c573deac864cc026d18d4a37d4f55b4e57c680
+  metadata.gz: 170b92d1d017ca7ac701e77723fd79b4a4a9c74414b5c6445da0929e89ff919c59006be4493b641f0a96c63b073b9c7258c65b4497f9071fbd96c209f2820ce8
+  data.tar.gz: 8267eeeb569ce8b9f92a2d1471825a39f8d786802caccdd9c66164e459b2648a950eb828135d0011b29d1e4e6744815ccbee48c7c71151335d04bd3f09edff8b

data/lib/zipf.rb

@@ -1,16 +1,15 @@
 #!/usr/bin/env ruby
 
-require 'zipf/stringutil'
+require 'zipf/bleu'
+require 'zipf/dag'
 require 'zipf/fileutil'
+require 'zipf/hypergraph'
+require 'zipf/misc'
+require 'zipf/semirings'
 require 'zipf/SparseVector'
+require 'zipf/stringutil'
 require 'zipf/tfidf'
 require 'zipf/Translation'
-require 'zipf/dag'
-require 'zipf/semirings'
-require 'zipf/bleu'
-require 'zipf/misc'
-require 'zipf/hg'
-require 'zipf/grammar'
 
 STDIN.set_encoding  'utf-8'
 STDOUT.set_encoding 'utf-8'

data/lib/zipf/SparseVector.rb

@@ -1,6 +1,5 @@
 require 'json'
 
-
 class SparseVector < Hash
 
   def initialize arg=nil

data/lib/zipf/bleu.rb

@@ -1,6 +1,5 @@
 module BLEU
 
-
 class BLEU::NgramCounts
   attr_accessor :sum, :clipped, :ref_len, :hyp_len, :n
 
@@ -125,6 +124,5 @@ def BLEU::per_sentence_bleu hypothesis, reference, n=4, smooth=0.0
   return Math.exp logbleu
 end
 
-
 end #module
 

data/lib/zipf/dag.rb

@@ -2,7 +2,6 @@ module DAG
 
 require 'json'
 
-
 class DAG::Node
   attr_accessor :label, :outgoing, :incoming, :score, :mark
 
@@ -200,6 +199,5 @@ def DAG::read_graph_from_json fn, semiring=RealSemiring.new
   return graph, nodes_by_label
 end
 
-
 end #module
 

data/lib/zipf/fileutil.rb

@@ -1,6 +1,5 @@
 require 'zlib'
 
-
 class ReadFile
 
   def initialize fn, encoding='utf-8'

data/lib/zipf/{hg.rb → hypergraph.rb}

@@ -3,10 +3,8 @@
 require_relative 'semirings'
 require 'json'
 
-
 module HG
 
-
 class HG::Node
   attr_accessor :label, :cat, :outgoing, :incoming, :score
 
@@ -168,6 +166,5 @@ def HG::all_paths hypergraph, root, semiring=ViterbiSemiring.new
   return paths
 end
 
-
 end #module
 

data/lib/zipf/misc.rb

@@ -1,6 +1,5 @@
 require 'timeout'
 
-
 class Array
   def max_index
     self.index(self.max)

data/lib/zipf/tfidf.rb

@@ -1,6 +1,5 @@
 module TFIDF
 
-
 # returns key='raw frequency' for an
 # array-like object
 def TFIDF::tf array, stopwords=[]

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: zipf
 version: !ruby/object:Gem::Version
-  version: 1.0.3
+  version: '1.1'
 platform: ruby
 authors:
 - Patrick Simianer
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-06-17 00:00:00.000000000 Z
+date: 2014-09-27 00:00:00.000000000 Z
 dependencies: []
 description: NLP related tools and classes
 email: p@simianer.de
@@ -19,12 +19,11 @@ files:
 - lib/zipf.rb
 - lib/zipf/stringutil.rb
 - lib/zipf/misc.rb
-- lib/zipf/grammar.rb
-- lib/zipf/hg.rb
 - lib/zipf/fileutil.rb
 - lib/zipf/semirings.rb
 - lib/zipf/dag.rb
 - lib/zipf/SparseVector.rb
+- lib/zipf/hypergraph.rb
 - lib/zipf/tfidf.rb
 - lib/zipf/bleu.rb
 - lib/zipf/Translation.rb
@@ -48,7 +47,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.0.3
+rubygems_version: 2.0.14
 signing_key: 
 specification_version: 4
 summary: zipf

data/lib/zipf/grammar.rb

@@ -1,123 +0,0 @@
-module Grammar
-
-
-class T
-  attr_accessor :word
-
-  def initialize word
-    @word = word
-  end
-
-  def to_s
-    "T<#{@word}>"
-  end
-end
-
-class NT
-  attr_accessor :symbol, :index, :span
-
-  def initialize symbol, index=0
-    @symbol = symbol
-    @index = index
-    @span = Span.new
-  end
-
-  def to_s
-    "NT(#{@span.left},#{@span.right})<#{@symbol},#{@index}>"
-  end
-end
-
-class Rule
-  attr_accessor :lhs, :rhs, :e
-
-  def initialize lhs=nil, rhs=[], e=''
-    @lhs = lhs
-    @rhs = rhs
-    @e = e
-  end
-
-  def to_s
-    "#{lhs} -> #{rhs.map{ |i| i.to_s }.join ' '} [arity=#{arity}] ||| #{@e}"
-  end
-
-  def arity
-    rhs.select { |i| i.class == NT }.size
-  end
-
-  def from_s s
-    _ = splitpipe s, 3
-    @lhs = NT.new _[0].strip.gsub!(/(\[|\])/, "")
-    _[1].split.each { |x|
-      x.strip!
-      if x[0]=='[' && x[x.size-1] == ']'
-        @rhs << NT.new(x.gsub!(/(\[|\])/, "").split(',')[0])
-      else
-        @rhs << T.new(x)
-      end
-    }
-    @e = _[2]
-  end
-
-  def self.from_s s
-    r = self.new
-    r.from_s s
-    return r
-  end
-end
-
-class Span
-  attr_accessor :left, :right
-
-  def initialize left=nil, right=nil
-    @left = left
-    @right = right
-  end
-end
-
-class Grammar
-  attr_accessor :rules, :startn, :startt, :flat
-
-  def initialize fn
-    @rules = []; @startn = []; @startt = [] ;@flat = []
-    ReadFile.readlines_strip(fn).each_with_index { |s,i|
-      STDERR.write '.'; STDERR.write " #{i+1}\n" if (i+1)%80==0
-      @rules << Rule.from_s(s)
-      if @rules.last.rhs.first.class == NT
-        @startn << @rules.last
-      else
-        if rules.last.arity == 0
-          @flat << @rules.last
-        else
-          @startt << @rules.last
-        end
-      end
-    }
-    STDERR.write "\n"
-  end
-
-  def to_s
-    s = ''
-    @rules.each { |r| s += r.to_s+"\n" }
-    return s
-  end
-
-  def add_glue_rules
-    @rules.map { |r| r.lhs.symbol }.select { |s| s != 'S' }.uniq.each { |symbol|
-      @rules << Rule.new(NT.new('S'), [NT.new(symbol)])
-      @startn << @rules.last
-      @rules << Rule.new(NT.new('S'), [NT.new('S'), NT.new('X')])
-      @startn << @rules.last
-    }
-  end
-
-  def add_pass_through_rules s
-    s.each { |word|
-      @rules << Rule.new(NT.new('X'), [T.new(word)])
-      @flat << @rules.last
-    }
-  end
-end
-
-
-end #module
-