1 files changed, 36 insertions, 0 deletions
diff --git a/add-seg b/add-seg
new file mode 100755
index 0000000..e5db580
--- /dev/null
+++ b/add-seg
@@ -0,0 +1,36 @@
+#!/usr/bin/env ruby
+
+require 'trollop'
+require 'zipf'
+
+o = Trollop::options do
+  opt :grammar, "(Absolute) path of folder containing grammars.", :type => :string, :short => '-g', :default => nil
+  opt :loo, "leave one out", :type => :bool, :default => false
+  opt :start_id, "start with this id", :type => :int, :default => 0, :short => '-i'
+  opt :nogz, "grammar files not gzipped", :type => :bool, :default => false
+  opt :index, "number according to index", :type => :string, :default => nil
+end
+
+index = []
+if o[:index]
+  index = ReadFile.readlines_strip(o[:index]).map{ |i| i.to_i }
+end
+
+i = o[:start_id]
+j = 0
+while line = STDIN.gets
+  ext = '.gz'
+  ext = '' if o[:nogz]
+  s = "<seg"
+  if o[:loo] then s += " exclude=\"#{i}\"" end
+  if index.size > 0
+    if o[:grammar] then s += " grammar=\"#{o[:grammar]}/grammar.#{index[j]}#{ext}\"" end
+    puts s + " id=\"#{index[j]}\"> #{line.strip} </seg>"
+  else
+    if o[:grammar] then s += " grammar=\"#{o[:grammar]}/grammar.#{i}#{ext}\"" end
+    puts s + " id=\"#{i}\"> #{line.strip} </seg>"
+  end
+  i += 1
+  j += 1
+end
+