From 298eed4f842bcc01b392a40c4a316110a6347f6c Mon Sep 17 00:00:00 2001
From: "linh.kitty" <linh.kitty@ec762483-ff6d-05da-a07a-a48fb63a330f>
Date: Tue, 13 Jul 2010 20:20:55 +0000
Subject: linh added

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@241 ec762483-ff6d-05da-a07a-a48fb63a330f
---
 gi/scfg/abc/Release/process_grammar.pl | 36 ++++++++++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)
 create mode 100644 gi/scfg/abc/Release/process_grammar.pl

(limited to 'gi/scfg/abc/Release/process_grammar.pl')

diff --git a/gi/scfg/abc/Release/process_grammar.pl b/gi/scfg/abc/Release/process_grammar.pl
new file mode 100644
index 00000000..f82a8e5a
--- /dev/null
+++ b/gi/scfg/abc/Release/process_grammar.pl
@@ -0,0 +1,36 @@
+#!perl
+
+use warnings;
+use strict;
+
+my $grammar_file = $ARGV[0];
+
+my %nt_count; #maps nt--> count rules whose lhs is nt 
+
+open(G, "<$grammar_file") or die "Can't open file $grammar_file";
+
+while (<G>){
+
+    chomp();
+
+    s/\|\|\|.*//g;
+    s/\s//g;
+
+    $nt_count{$_}++;
+}
+
+
+close (G);
+
+open(G, "<$grammar_file") or die "Can't open file $grammar_file";
+
+while (<G>){
+
+    chomp();
+
+    (my $nt = $_) =~ s/\|\|\|.*//g;
+    $nt =~ s/\s//g;
+
+    s/(.+\|\|\|.+\|\|\|.+\|\|\|).+/$1/g;
+    print $_ . " MinusLogP=" .(log($nt_count{$nt})) ."\n";
+}
-- 
cgit v1.2.3