From d53350dd95f22812d98ce4e6afc2f40a16aaac1c Mon Sep 17 00:00:00 2001
From: bhaddow <bhaddow@1f5c12ca-751b-0410-a591-d2e778427230>
Date: Mon, 12 Sep 2011 20:53:22 +0000
Subject: [PATCH] Regression test for lm oov feature

git-svn-id: https://mosesdecoder.svn.sourceforge.net/svnroot/mosesdecoder/trunk@4218 1f5c12ca-751b-0410-a591-d2e778427230
---
 .../phrase.basic-lm-oov/filter-stderr.pl      | 22 ++++++++
 .../phrase.basic-lm-oov/filter-stdout.pl      |  7 +++
 .../tests/phrase.basic-lm-oov/moses.ini       | 53 +++++++++++++++++++
 .../phrase.basic-lm-oov/to-translate.txt      |  5 ++
 .../phrase.basic-lm-oov/truth/results.txt     | 13 +++++
 5 files changed, 100 insertions(+)
 create mode 100755 regression-testing/tests/phrase.basic-lm-oov/filter-stderr.pl
 create mode 100755 regression-testing/tests/phrase.basic-lm-oov/filter-stdout.pl
 create mode 100644 regression-testing/tests/phrase.basic-lm-oov/moses.ini
 create mode 100644 regression-testing/tests/phrase.basic-lm-oov/to-translate.txt
 create mode 100644 regression-testing/tests/phrase.basic-lm-oov/truth/results.txt

diff --git a/regression-testing/tests/phrase.basic-lm-oov/filter-stderr.pl b/regression-testing/tests/phrase.basic-lm-oov/filter-stderr.pl
new file mode 100755
index 000000000..2f6e17607
--- /dev/null
+++ b/regression-testing/tests/phrase.basic-lm-oov/filter-stderr.pl
@@ -0,0 +1,22 @@
+#!/usr/bin/perl
+
+BEGIN { use Cwd qw/ abs_path /; use File::Basename; $script_dir = dirname(abs_path($0)); push @INC, "$script_dir/../perllib"; }
+use RegTestUtils;
+
+$x=0;
+while (<>) {
+  chomp;
+
+  if (/^Finished loading LanguageModels/) {
+    my $time = RegTestUtils::readTime($_);
+    print "LMLOAD_TIME ~ $time\n";
+  }
+  if (/^Finished loading phrase tables/) {
+    my $time = RegTestUtils::readTime($_);
+    print "PTLOAD_TIME ~ $time\n";
+  }
+  next unless /^BEST TRANSLATION:/;
+  my $pscore = RegTestUtils::readHypoScore($_);
+  $x++;
+  print "SCORE_$x = $pscore\n";
+}
diff --git a/regression-testing/tests/phrase.basic-lm-oov/filter-stdout.pl b/regression-testing/tests/phrase.basic-lm-oov/filter-stdout.pl
new file mode 100755
index 000000000..476ddf6e9
--- /dev/null
+++ b/regression-testing/tests/phrase.basic-lm-oov/filter-stdout.pl
@@ -0,0 +1,7 @@
+#!/usr/bin/perl
+$x=0;
+while (<>) {
+  chomp;
+  $x++;
+  print "TRANSLATION_$x=$_\n";
+}
diff --git a/regression-testing/tests/phrase.basic-lm-oov/moses.ini b/regression-testing/tests/phrase.basic-lm-oov/moses.ini
new file mode 100644
index 000000000..88bbc1ee6
--- /dev/null
+++ b/regression-testing/tests/phrase.basic-lm-oov/moses.ini
@@ -0,0 +1,53 @@
+# moses.ini for regression test
+
+[ttable-file]
+0 0 0 5 ${MODEL_PATH}/basic-surface-only/phrase-table.gz
+
+# language model
+[lmodel-file]
+0 0 3 ${LM_PATH}/europarl.en.srilm.gz
+# limit on how many phrase translations e for each phrase f are loaded
+[ttable-limit]
+#ttable element load limit 0 = all elements loaded
+20
+
+# distortion (reordering) weight
+[weight-d]
+0.141806519223522
+
+# language model weight
+[weight-l]
+0.142658800199951
+0.1
+
+# translation model weight (phrase translation, lexical weighting)
+[weight-t]
+0.00402447059454402
+0.0685647475075862
+0.294089113124688
+0.0328320356515851
+-0.0426081987467227
+
+# word penalty
+[weight-w]
+-0.273416114951401
+
+[distortion-limit]
+4
+
+[beam-threshold]
+0.03
+
+[input-factors]
+0
+
+[mapping]
+T 0
+
+[lmodel-oov-feature]
+1
+
+
+[verbose]
+2
+
diff --git a/regression-testing/tests/phrase.basic-lm-oov/to-translate.txt b/regression-testing/tests/phrase.basic-lm-oov/to-translate.txt
new file mode 100644
index 000000000..a505702d1
--- /dev/null
+++ b/regression-testing/tests/phrase.basic-lm-oov/to-translate.txt
@@ -0,0 +1,5 @@
+ich frage sie also , herr präsident : stellen die unterschiedlichen arbeitskosten somit nicht auch eine beschränkung des freien wettbewerbs in der europäischen union dar ?
+schaut man sich die fälligkeitspläne der ausführung des haushalts für die rubriken 2 , 3 , 4 und 7 an , stellt man fest , dass nur durchschnittlich 8 % aller verpflichtungen durch zahlungen gedeckt sind .
+vor drei jahren haben wir mit unserer beschäftigungsinitiative begonnen , indem wir kleinen und mittleren unternehmen halfen , chancenkapital zu bekommen .
+das parlament will das auf zweierlei weise tun .
+nur dann werden die europäischen institutionen auch ihrem auftrag gerecht .
diff --git a/regression-testing/tests/phrase.basic-lm-oov/truth/results.txt b/regression-testing/tests/phrase.basic-lm-oov/truth/results.txt
new file mode 100644
index 000000000..faca9f17b
--- /dev/null
+++ b/regression-testing/tests/phrase.basic-lm-oov/truth/results.txt
@@ -0,0 +1,13 @@
+TRANSLATION_1=i ask you , therefore , mr president , the different labour costs are therefore not a restriction of free competition in the european union ? 
+TRANSLATION_2=if we look at the fälligkeitspläne the implementation of the budget for the categories 2 , 3 , 4 and 7 to , we see that only an average of 8 % of commitments by payments are met . 
+TRANSLATION_3=three years ago our employment strategy , we started by small and medium-sized enterprises halfen , chancenkapital to obtain . 
+TRANSLATION_4=parliament wants the in two ways . 
+TRANSLATION_5=only then will the european institutions to its mandate . 
+LMLOAD_TIME ~ 8.00
+PTLOAD_TIME ~ 9.00
+SCORE_1 = -14.843
+SCORE_2 = -133.657
+SCORE_3 = -240.041
+SCORE_4 = -5.995
+SCORE_5 = -7.015
+TOTAL_WALLTIME ~ 28