add test for sparse feature PhraseLengthFeature

6 files changed, 84 insertions, 0 deletions

diff --git a/tests/phrase.phrase-length-feature/filter-stderr.pl b/tests/phrase.phrase-length-feature/filter-stderr.pl
new file mode 100755
index 0000000..2f6e176
--- /dev/null
+++ b/tests/phrase.phrase-length-feature/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 = $pscore\n";
+}
diff --git a/tests/phrase.phrase-length-feature/filter-stdout.pl b/tests/phrase.phrase-length-feature/filter-stdout.pl
new file mode 100755
index 0000000..476ddf6
--- /dev/null
+++ b/tests/phrase.phrase-length-feature/filter-stdout.pl
@@ -0,0 +1,7 @@
+#!/usr/bin/perl
+$x=0;
+while (<>) {
+  chomp;
+  $x++;
+  print "TRANSLATION_$x=$_\n";
+}
diff --git a/tests/phrase.phrase-length-feature/moses.ini b/tests/phrase.phrase-length-feature/moses.ini
new file mode 100644
index 0000000..b36abb7
--- /dev/null
+++ b/tests/phrase.phrase-length-feature/moses.ini
@@ -0,0 +1,41 @@
+# moses.ini for regression test
+
+# limit on how many phrase translations e for each phrase f are loaded
+[ttable-limit]
+#ttable element load limit 0 = all elements loaded
+20
+
+[distortion-limit]
+4
+
+[beam-threshold]
+0.03
+
+[input-factors]
+0
+
+[mapping]
+T 0
+
+
+[verbose]
+2
+
+[feature]
+WordPenalty
+Distortion
+KENLM lazyken=0 order=3 factor=0 path=${LM_PATH}/europarl.en.srilm.gz
+PhraseLengthFeature
+PhraseDictionaryMemory num-features=5 input-factor=0 output-factor=0 path=${MODEL_PATH}/ptable-with-alignment/phrase-table.gz
+UnknownWordPenalty
+
+[weight]
+WordPenalty0= -0.273416114951401
+KENLM0= 0.142658800199951
+Distortion0= 0.141806519223522
+PhraseDictionaryMemory0= 0.00402447059454402 0.0685647475075862 0.294089113124688 0.0328320356515851 -0.0426081987467227
+
+
+[weight-file]
+${TEST_PATH}/weights
+
diff --git a/tests/phrase.phrase-length-feature/to-translate.txt b/tests/phrase.phrase-length-feature/to-translate.txt
new file mode 100644
index 0000000..78d7ca1
--- /dev/null
+++ b/tests/phrase.phrase-length-feature/to-translate.txt
@@ -0,0 +1,2 @@
+vor allem aber die Anwalt
+Hand hat Mehrheit
diff --git a/tests/phrase.phrase-length-feature/truth/results.txt b/tests/phrase.phrase-length-feature/truth/results.txt
new file mode 100644
index 0000000..c551880
--- /dev/null
+++ b/tests/phrase.phrase-length-feature/truth/results.txt
@@ -0,0 +1,5 @@
+TRANSLATION_1=especially the lawyers 
+TRANSLATION_2=country majority 
+SCORE_1 = 7.298
+SCORE_2 = -4.648
+TOTAL_WALLTIME ~ 2
diff --git a/tests/phrase.phrase-length-feature/weights b/tests/phrase.phrase-length-feature/weights
new file mode 100644
index 0000000..a02a9df
--- /dev/null
+++ b/tests/phrase.phrase-length-feature/weights
@@ -0,0 +1,7 @@
+PhraseLengthFeature0_4,2 4.5
+PhraseLengthFeature0_t2 2.3
+PhraseLengthFeature0_s4 5.343
+PhraseLengthFeature0_s1 -2.55
+PhraseLengthFeature0_t1 -0.232
+PhraseLengthFeature0_1,1 2.756
+