From 5374ff923659a030c9ffed3f30241661e8a8c331 Mon Sep 17 00:00:00 2001
From: redpony <redpony@ec762483-ff6d-05da-a07a-a48fb63a330f>
Date: Thu, 22 Jul 2010 21:31:54 +0000
Subject: forgotten

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@371 ec762483-ff6d-05da-a07a-a48fb63a330f
---
 gi/pipeline/scripts/filter-by-f.pl | 53 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 53 insertions(+)
 create mode 100755 gi/pipeline/scripts/filter-by-f.pl

(limited to 'gi')

diff --git a/gi/pipeline/scripts/filter-by-f.pl b/gi/pipeline/scripts/filter-by-f.pl
new file mode 100755
index 00000000..3dd03bdd
--- /dev/null
+++ b/gi/pipeline/scripts/filter-by-f.pl
@@ -0,0 +1,53 @@
+#!/usr/bin/perl -w
+use strict;
+
+my $SCRIPT_DIR; BEGIN { use Cwd qw/ abs_path /; use File::Basename; $SCRIPT_DIR = dirname(abs_path($0)); push @INC, $SCRIPT_DIR; }
+
+my $REKEY="$SCRIPT_DIR/rekey.pl";
+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 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 > $ARGV[1]";
+safesystem($ARGV[1], $cmd) or die "Filtering failed";
+exit 0;
+
+sub assert_exec {
+  my @files = @_;
+  for my $file (@files) {
+    die "Can't find $file - did you run make?\n" unless -e $file;
+    die "Can't execute $file" unless -e $file;
+  }
+};
+
+sub safesystem {
+  my $output = shift @_;
+  print STDERR "Executing: @_\n";
+  system(@_);
+  if ($? == -1) {
+      print STDERR "ERROR: Failed to execute: @_\n  $!\n";
+      if (defined $output && -e $output) { printf STDERR "Removing $output\n"; `rm -rf $output`; }
+      exit(1);
+  }
+  elsif ($? & 127) {
+      printf STDERR "ERROR: Execution of: @_\n  died with signal %d, %s coredump\n",
+          ($? & 127),  ($? & 128) ? 'with' : 'without';
+      if (defined $output && -e $output) { printf STDERR "Removing $output\n"; `rm -rf $output`; }
+      exit(1);
+  }
+  else {
+    my $exitcode = $? >> 8;
+    if ($exitcode) {
+      print STDERR "Exit code: $exitcode\n";
+      if (defined $output && -e $output) { printf STDERR "Removing $output\n"; `rm -rf $output`; }
+    }
+    return ! $exitcode;
+  }
+}
+
-- 
cgit v1.2.3