From 1a6e8eb7b5f848079f162fb2fca49e81cdabb698 Mon Sep 17 00:00:00 2001
From: redpony <redpony@ec762483-ff6d-05da-a07a-a48fb63a330f>
Date: Sun, 25 Jul 2010 21:03:19 +0000
Subject: configure number of translations to keep

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@410 ec762483-ff6d-05da-a07a-a48fb63a330f
---
 gi/pipeline/scripts/filter-by-f.pl | 7 +++++--
 gi/pipeline/scripts/refilter.pl    | 5 ++++-
 2 files changed, 9 insertions(+), 3 deletions(-)

(limited to 'gi/pipeline/scripts')

diff --git a/gi/pipeline/scripts/filter-by-f.pl b/gi/pipeline/scripts/filter-by-f.pl
index 3dd03bdd..0cef0606 100755
--- a/gi/pipeline/scripts/filter-by-f.pl
+++ b/gi/pipeline/scripts/filter-by-f.pl
@@ -8,13 +8,16 @@ my $REFILTER="$SCRIPT_DIR/refilter.pl";
 my $SORT="$SCRIPT_DIR/sort-by-key.sh";
 assert_exec($REKEY, $REFILTER, $SORT);
 
-die "Usage: $0 ingrammar.gz outgrammar.gz\n" unless scalar @ARGV == 2;
+
+die "Usage: $0 NUM-TRANSLATIONS ingrammar.gz outgrammar.gz\n" unless scalar @ARGV == 3;
+my $translations = shift @ARGV;
+die "Need number: $translations" unless $translations > 0;
 die unless $ARGV[0] =~ /\.gz$/;
 die unless $ARGV[1] =~ /\.gz$/;
 die if $ARGV[0] eq $ARGV[1];
 die "Can't find $ARGV[0]" unless -f $ARGV[0];
 
-my $cmd = "gunzip -c $ARGV[0] | $REKEY | $SORT | $REFILTER | gzip > $ARGV[1]";
+my $cmd = "gunzip -c $ARGV[0] | $REKEY | $SORT | $REFILTER $translations | gzip > $ARGV[1]";
 safesystem($ARGV[1], $cmd) or die "Filtering failed";
 exit 0;
 
diff --git a/gi/pipeline/scripts/refilter.pl b/gi/pipeline/scripts/refilter.pl
index 11a36ebe..a783eb4e 100755
--- a/gi/pipeline/scripts/refilter.pl
+++ b/gi/pipeline/scripts/refilter.pl
@@ -1,7 +1,10 @@
 #!/usr/bin/perl -w
 use strict;
 
-my $NUM_TRANSLATIONS = 30;
+my $NUM_TRANSLATIONS = shift @ARGV;
+unless ($NUM_TRANSLATIONS) { $NUM_TRANSLATIONS=30; }
+print STDERR "KEEPING $NUM_TRANSLATIONS TRANSLATIONS FOR SOURCE\n";
+
 my $pk = '';
 my %dict;
 while(<>) {
-- 
cgit v1.2.3