summaryrefslogtreecommitdiff
path: root/prototype/weaver.rb
diff options
context:
space:
mode:
authorPatrick Simianer <p@simianer.de>2014-09-20 16:24:39 +0100
committerPatrick Simianer <p@simianer.de>2014-09-20 16:24:39 +0100
commit8949948243cb709e65ec172b3014c9e6a3181ed3 (patch)
tree6342152d91054601b107a1f75c1271f06a6c732e /prototype/weaver.rb
parent129a22cfcc7651daa4b11ed52e7870249f6373a5 (diff)
prototype: fixed grammar handling
Diffstat (limited to 'prototype/weaver.rb')
-rwxr-xr-xprototype/weaver.rb30
1 files changed, 21 insertions, 9 deletions
diff --git a/prototype/weaver.rb b/prototype/weaver.rb
index 966d4c8..5afdb93 100755
--- a/prototype/weaver.rb
+++ b/prototype/weaver.rb
@@ -20,11 +20,11 @@ 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 => '-l'
- opt :add_pass_through, "", :type => :bool, :default => false, :short => '-p'
+ opt :input, "", :type => :string, :default => '-', :short => '-i'
+ opt :grammar, "", :type => :string, :required => true, :short => '-g'
+ opt :weights, "", :type => :string, :required => true, :short => '-w'
+ opt :add_glue, "", :type => :bool, :default => false, :short => '-l'
+ opt :add_pass_through, "", :type => :bool, :default => false, :short => '-p'
end
grammar = nil
@@ -32,17 +32,29 @@ def main
grammar = read_grammar cfg[:grammar], cfg[:add_glue], cfg[:add_pass_through]
end
+ sgm_input = false
+ if ['sgm', 'xml'].include? cfg[:input].split('.')[-1]
+ sgm_input = true
+ 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
+ if sgm_input
+ x = XmlSimple.xml_in(input)
+ input = x['content'].split
+ else
+ input = input.split
+ end
n = input.size
- if x['grammar']
+ if sgm_input && x['grammar']
grammar = read_grammar x['grammar'], cfg[:add_glue], cfg[:add_pass_through], input
+ elsif cfg[:add_pass_through]
+ grammar.add_pass_through input
end
+
STDERR.write "> initializing charts\n"
passive_chart = Parse::Chart.new n
active_chart = Parse::Chart.new n
@@ -62,7 +74,7 @@ def main
semiring = ViterbiSemiring.new
path, score = HG::viterbi_path hypergraph, hypergraph.nodes_by_id[-1], semiring
s = HG::derive path, path.last.head, []
- STDERR.write " #{s.map { |i| i.word }.join ' '} ||| #{Math.log score}\n"
+ STDOUT.write "#{s.map { |i| i.word }.join ' '} ||| #{Math.log score}\n"
}
end