From 840d1b84a977b46180350bdb965a118150f238d9 Mon Sep 17 00:00:00 2001
From: redpony <redpony@ec762483-ff6d-05da-a07a-a48fb63a330f>
Date: Thu, 24 Jun 2010 04:05:17 +0000
Subject: pipeline scripts

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@17 ec762483-ff6d-05da-a07a-a48fb63a330f
---
 gi/pipeline/filter-for-test-set.pl | 73 ++++++++++++++++++++++++++++++++++++++
 gi/pipeline/local-gi-pipeline.pl   |  2 +-
 gi/pipeline/scripts/sort-by-key.sh |  5 +++
 gi/pipeline/sort-by-key.sh         |  5 ---
 4 files changed, 79 insertions(+), 6 deletions(-)
 create mode 100755 gi/pipeline/filter-for-test-set.pl
 create mode 100755 gi/pipeline/scripts/sort-by-key.sh
 delete mode 100755 gi/pipeline/sort-by-key.sh

diff --git a/gi/pipeline/filter-for-test-set.pl b/gi/pipeline/filter-for-test-set.pl
new file mode 100755
index 00000000..61edaf67
--- /dev/null
+++ b/gi/pipeline/filter-for-test-set.pl
@@ -0,0 +1,73 @@
+#!/usr/bin/perl -w
+use strict;
+my $SCRIPT_DIR; BEGIN { use Cwd qw/ abs_path cwd /; use File::Basename; $SCRIPT_DIR = dirname(abs_path($0)); push @INC, $SCRIPT_DIR; }
+
+use Getopt::Long "GetOptions";
+use IPC::Run3;
+use File::Temp qw ( tempdir );
+my $TEMP_DIR = tempdir( CLEANUP => 1 );
+
+my $GZIP = 'gzip';
+my $ZCAT = 'gunzip -c';
+
+my $EXTOOLS = "$SCRIPT_DIR/../../extools";
+die "Can't find extools: $EXTOOLS" unless -e $EXTOOLS && -d $EXTOOLS;
+
+my $FILTER = "$EXTOOLS/filter_grammar";
+my $SCORE = "$EXTOOLS/score_grammar";
+
+assert_exec($FILTER, $SCORE);
+
+usage() unless scalar @ARGV == 3;
+my $corpus = $ARGV[0];
+my $grammar = $ARGV[1];
+my $testset = $ARGV[2];
+die "Can't find corpus: $corpus" unless -f $corpus;
+die "Can't find corpus: $grammar" unless -f $grammar;
+die "Can't find corpus: $testset" unless -f $testset;
+print STDERR "  CORPUS: $corpus\n";
+print STDERR " GRAMMAR: $corpus\n";
+print STDERR "TEST SET: $corpus\n";
+print STDERR "Extracting...\n";
+
+safesystem("$ZCAT $grammar | $FILTER $testset | $SCORE $corpus") or die "Failed";
+
+sub usage {
+  print <<EOT;
+
+Usage: $0 corpus.src_trg_al grammar.gz test-set.txt > filtered-grammar.scfg.txt
+
+Filter and score a grammar for a test set.
+
+EOT
+  exit 1;
+};
+
+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 $file;
+  }
+};
+
+sub safesystem {
+  print STDERR "Executing: @_\n";
+  system(@_);
+  if ($? == -1) {
+      print STDERR "ERROR: Failed to execute: @_\n  $!\n";
+      exit(1);
+  }
+  elsif ($? & 127) {
+      printf STDERR "ERROR: Execution of: @_\n  died with signal %d, %s coredump\n",
+          ($? & 127),  ($? & 128) ? 'with' : 'without';
+      exit(1);
+  }
+  else {
+    my $exitcode = $? >> 8;
+    print STDERR "Exit code: $exitcode\n" if $exitcode;
+    return ! $exitcode;
+  }
+}
+
+
diff --git a/gi/pipeline/local-gi-pipeline.pl b/gi/pipeline/local-gi-pipeline.pl
index e52ad4ec..4707d5a3 100755
--- a/gi/pipeline/local-gi-pipeline.pl
+++ b/gi/pipeline/local-gi-pipeline.pl
@@ -28,7 +28,7 @@ my $S2L = "$PYPSCRIPTS/spans2labels.py";
 
 my $PYP_TOPICS_TRAIN="$PYPTOOLS/pyp-topics-train";
 
-my $SORT_KEYS = "$SCRIPT_DIR/sort-by-key.sh";
+my $SORT_KEYS = "$SCRIPT_DIR/scripts/sort-by-key.sh";
 my $EXTRACTOR = "$EXTOOLS/extractor";
 my $FILTER = "$EXTOOLS/filter_grammar";
 my $SCORER = "$EXTOOLS/score_grammar";
diff --git a/gi/pipeline/scripts/sort-by-key.sh b/gi/pipeline/scripts/sort-by-key.sh
new file mode 100755
index 00000000..948dd4df
--- /dev/null
+++ b/gi/pipeline/scripts/sort-by-key.sh
@@ -0,0 +1,5 @@
+#!/bin/bash
+
+export LANG=C
+sort -t $'\t' -k 1
+
diff --git a/gi/pipeline/sort-by-key.sh b/gi/pipeline/sort-by-key.sh
deleted file mode 100755
index 948dd4df..00000000
--- a/gi/pipeline/sort-by-key.sh
+++ /dev/null
@@ -1,5 +0,0 @@
-#!/bin/bash
-
-export LANG=C
-sort -t $'\t' -k 1
-
-- 
cgit v1.2.3