de-tok.rb: (de)tokenization daemon wrapper

1 files changed, 36 insertions, 0 deletions

diff --git a/de-tok.rb b/de-tok.rb
new file mode 100755
index 0000000..92c563f
--- /dev/null
+++ b/de-tok.rb
@@ -0,0 +1,36 @@
+#!/usr/bin/env ruby
+
+require 'nanomsg'
+require 'open3'
+require 'trollop'
+
+conf = Trollop::options do
+  opt :action, "tokenize (T) or detokenize (D)", :type => :string, :requred => true
+  opt :addr, "socket address", :short => "-S", :type => :string, :required => true
+  opt :scripts, "path to scripts directory", :short => "-p", :type => :string, :required => true
+  opt :lang, "language", :short => "-l", :type => :string, :required => true
+end
+
+sock = NanoMsg::PairSocket.new
+sock.bind conf[:addr]
+sock.send "hello"
+
+if conf[:action] == "D"
+  cmd = "#{conf[:scripts]}/detokenizer.perl -q -b -u -l #{conf[:lang]}"
+elsif conf[:action] == "T"
+  cmd = "#{conf[:scripts]}/tokenizer-no-escape.perl -q -b -a -l #{conf[:lang]}"
+else
+  # ERROR
+end
+while true
+  inp = sock.recv
+  break if !inp||inp=="shutdown"
+  Open3.popen3(cmd) do |pin, pout, perr|
+    pin.write inp
+    pin.close
+    sock.send pout.gets.strip
+  end
+end
+
+sock.send "off"
+