summaryrefslogtreecommitdiff
path: root/main.rb
diff options
context:
space:
mode:
Diffstat (limited to 'main.rb')
-rwxr-xr-xmain.rb75
1 files changed, 75 insertions, 0 deletions
diff --git a/main.rb b/main.rb
new file mode 100755
index 0000000..229454a
--- /dev/null
+++ b/main.rb
@@ -0,0 +1,75 @@
+#!/usr/bin/env ruby
+
+require 'trollop'
+require 'xmlsimple'
+require_relative 'parse'
+
+
+def read_grammar fn, add_glue, add_pass_through
+ STDERR.write "> reading grammar '#{fn}'\n"
+ grammar = Grammar::Grammar.new fn
+ if add_glue
+ STDERR.write ">> adding glue grammar\n"
+ grammar.add_glue_rules
+ end
+ if add_pass_through
+ STDERR.write ">> adding pass-through grammar\n"
+ grammar.add_pass_through_rules input
+ end
+ return grammar
+end
+
+def main
+ cfg = Trollop::options do
+ opt :input, "", :type => :string, :default => '-', :short => '-i'
+ opt :grammar, "", :type => :string, :default => nil, :short => '-g'
+ opt :weights, "", :type => :string, :default => nil, :short => '-w'
+ opt :add_glue, "", :type => :bool, :default => false, :short => '-h'
+ opt :add_pass_through, "", :type => :bool, :default => false, :short => '-p'
+ end
+
+ grammar = nil
+ if cfg[:grammar]
+ grammar = read_grammar cfg[:grammar], cfg[:add_glue], cfg[:add_pass_through]
+ end
+
+ STDERR.write "> reading input from '#{cfg[:input]}'\n"
+ ReadFile.readlines_strip(cfg[:input]).each { |input|
+
+ x = XmlSimple.xml_in(input)
+ input = x['content'].split
+ n = input.size
+
+ if x['grammar']
+ grammar = read_grammar x['grammar'], cfg[:add_glue], cfg[:add_pass_through]
+ end
+
+ STDERR.write "> initializing charts\n"
+ passive_chart = Parse::Chart.new n
+ active_chart = Parse::Chart.new n
+ Parse::init input, n, active_chart, passive_chart, grammar
+
+ STDERR.write "> parsing\n"
+ Parse::parse input, n, active_chart, passive_chart, grammar
+
+ weights = SparseVector.from_kv(ReadFile.read(cfg[:weights]), ' ', "\n")
+ if !weights
+ weights = SparseVector.new
+ end
+
+ hypergraph, nodes_by_id = passive_chart.to_hg weights
+
+ STDERR.write "> viterbi\n"
+ semiring = ViterbiSemiring.new
+ path, score = HG::viterbi_path hypergraph, nodes_by_id[-1], semiring
+ s = HG::derive path, path.last.head, []
+ puts "#{s.map { |i| i.word }.join ' '}"
+ puts Math.log score
+ puts
+
+ }
+end
+
+
+main
+