diff options
author | Patrick Simianer <p@simianer.de> | 2014-09-27 15:17:43 +0100 |
---|---|---|
committer | Patrick Simianer <p@simianer.de> | 2014-09-27 15:17:43 +0100 |
commit | f65e428784cfd2264f3fdfddd574c37acd38f54f (patch) | |
tree | 6a98f82964cbce52957a2d62f8b2c5b7ff8be095 /lib | |
parent | f28694f9bdd5059050e420cca2ca2694cbb0dda4 (diff) |
cosmetic changes, bump to 1.1
Diffstat (limited to 'lib')
-rwxr-xr-x | lib/zipf.rb | 3 | ||||
-rw-r--r-- | lib/zipf/SparseVector.rb | 1 | ||||
-rw-r--r-- | lib/zipf/bleu.rb | 2 | ||||
-rw-r--r-- | lib/zipf/dag.rb | 2 | ||||
-rw-r--r-- | lib/zipf/fileutil.rb | 1 | ||||
-rw-r--r-- | lib/zipf/grammar.rb | 123 | ||||
-rw-r--r-- | lib/zipf/hypergraph.rb (renamed from lib/zipf/hg.rb) | 3 | ||||
-rw-r--r-- | lib/zipf/misc.rb | 1 | ||||
-rw-r--r-- | lib/zipf/tfidf.rb | 1 |
9 files changed, 1 insertions, 136 deletions
diff --git a/lib/zipf.rb b/lib/zipf.rb index 2f59ccc..681e2cd 100755 --- a/lib/zipf.rb +++ b/lib/zipf.rb @@ -9,8 +9,7 @@ require 'zipf/dag' require 'zipf/semirings' require 'zipf/bleu' require 'zipf/misc' -require 'zipf/hg' -require 'zipf/grammar' +require 'zipf/hypergraph' STDIN.set_encoding 'utf-8' STDOUT.set_encoding 'utf-8' diff --git a/lib/zipf/SparseVector.rb b/lib/zipf/SparseVector.rb index 3f950c4..e9af799 100644 --- a/lib/zipf/SparseVector.rb +++ b/lib/zipf/SparseVector.rb @@ -1,6 +1,5 @@ require 'json' - class SparseVector < Hash def initialize arg=nil diff --git a/lib/zipf/bleu.rb b/lib/zipf/bleu.rb index 69de00b..c07315e 100644 --- a/lib/zipf/bleu.rb +++ b/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 diff --git a/lib/zipf/dag.rb b/lib/zipf/dag.rb index 45ede20..a0edde7 100644 --- a/lib/zipf/dag.rb +++ b/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 diff --git a/lib/zipf/fileutil.rb b/lib/zipf/fileutil.rb index eb69136..ac701da 100644 --- a/lib/zipf/fileutil.rb +++ b/lib/zipf/fileutil.rb @@ -1,6 +1,5 @@ require 'zlib' - class ReadFile def initialize fn, encoding='utf-8' diff --git a/lib/zipf/grammar.rb b/lib/zipf/grammar.rb deleted file mode 100644 index 568b9fc..0000000 --- a/lib/zipf/grammar.rb +++ /dev/null @@ -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 - diff --git a/lib/zipf/hg.rb b/lib/zipf/hypergraph.rb index f86bf60..6666062 100644 --- a/lib/zipf/hg.rb +++ b/lib/zipf/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 diff --git a/lib/zipf/misc.rb b/lib/zipf/misc.rb index 0319a5f..4d29a06 100644 --- a/lib/zipf/misc.rb +++ b/lib/zipf/misc.rb @@ -1,6 +1,5 @@ require 'timeout' - class Array def max_index self.index(self.max) diff --git a/lib/zipf/tfidf.rb b/lib/zipf/tfidf.rb index 13a40a3..7fb92c9 100644 --- a/lib/zipf/tfidf.rb +++ b/lib/zipf/tfidf.rb @@ -1,6 +1,5 @@ module TFIDF - # returns key='raw frequency' for an # array-like object def TFIDF::tf array, stopwords=[] |