Merge branch 'master' of github.com:pks/cdec-dtrain

6 files changed, 106 insertions, 7 deletions

diff --git a/corpus/filter-length.pl b/corpus/filter-length.pl
index 38837f14..2e257cda 100755
--- a/corpus/filter-length.pl
+++ b/corpus/filter-length.pl
@@ -44,11 +44,11 @@ while(<F>) {
   $lines++;
   if ($lines % 100000 == 0) { print STDERR " [$lines]\n"; }
   elsif ($lines % 2500 == 0) { print STDERR "."; }
-  my ($sf, $se, @d) = split / \|\|\| /;
+  my ($sf, $se, @d) = split /\s*\|\|\|\s*/;
   if (scalar @d != 0 or !defined $se) {
     $bad_format++;
     if ($bad_format > 100 && ($bad_format / $lines) > 0.02) {
-      die "Corpus appears to be incorretly formatted, example: $_";
+      die "$bad_format / $lines : Corpus appears to be incorretly formatted, example: $_";
     }
     next;
   }
diff --git a/corpus/paste-files.pl b/corpus/paste-files.pl
index 0b788386..4cb424ad 100755
--- a/corpus/paste-files.pl
+++ b/corpus/paste-files.pl
@@ -41,9 +41,9 @@ while(1) {
     }
     warn "$ARGV[$anum]:$lc contains a ||| symbol - please remove.\n" if $r =~ /\|\|\|/;
     $r =~ s/\|\|\|/ /g;
-    $r =~ s/  +//g;
-    $r =~ s/^ //;
-    $r =~ s/ $//;
+    $r =~ s/\s+/ /g;
+    $r =~ s/^ +//;
+    $r =~ s/ +$//;
     $anum++;
     push @line, $r;
   }
@@ -56,5 +56,5 @@ for (my $i = 1; $i < scalar @fhs; $i++) {
   my $r = <$fh>;
   die "Mismatched number of lines.\n" if defined $r;
 }
-print STDERR "Bad lines containing ||| were $bad\n";
+print STDERR "Number of lines containing ||| was: $bad\n" if $bad > 0;
 
diff --git a/corpus/support/quote-norm.pl b/corpus/support/quote-norm.pl
index 72b0064d..e4e5055e 100755
--- a/corpus/support/quote-norm.pl
+++ b/corpus/support/quote-norm.pl
@@ -27,6 +27,20 @@ while(<STDIN>) {
   s/¨/"/g;
   s/¡/ ¡ /g;
   s/¿/ ¿ /g;
+
+  s/ п. п. / п.п. /g;
+  s/ ст. л. / ст.л. /g;
+  s/ т. е. / т.е. /g;
+  s/ т. к. / т.к. /g;
+  s/ т. ч. / т.ч. /g;
+  s/ т. д. / т.д. /g;
+  s/ т. п. / т.п. /g;
+  s/ и. о. / и.о. /g;
+  s/ с. г. / с.г. /g;
+  s/ г. р. / г.р. /g;
+  s/ т. н. / т.н. /g;
+  s/ т. ч. / т.ч. /g;
+  s/ н. э. / н.э. /g;
   # â<U+0080><U+0099>
   s/â(\x{80}\x{99}|\x{80}\x{98})/'/g;
   s/â(\x{80}\x{9c}|\x{80}\x{9d})/"/g;
diff --git a/corpus/support/token_list b/corpus/support/token_list
index 28eb4396..d470cb22 100644
--- a/corpus/support/token_list
+++ b/corpus/support/token_list
@@ -210,3 +210,85 @@ W.
 X.
 Y.
 Z.
+А.
+Б.
+В.
+Г.
+Д.
+Е.
+Ё.
+Ж.
+З.
+И.
+Й.
+К.
+Л.
+М.
+Н.
+О.
+П.
+Р.
+С.
+Т.
+У.
+Ф.
+Х.
+Ц.
+Ч.
+Ш.
+Щ.
+Ъ.
+Ы.
+Ь.
+Э.
+Ю.
+Я.
+л.
+г.
+обл.
+гг.
+в.
+вв.
+мин.
+ч.
+тыс.
+млн.
+млрд.
+трлн.
+кв.
+куб.
+руб.
+коп.
+долл.
+Прим.
+прим.
+чел.
+грн.
+мин.
+им.
+проф.
+акад.
+ред.
+авт.
+корр.
+соб.
+спец.
+см.
+тж.
+др.
+пр.
+букв.
+# Two-letter abbreviations - can be written with space
+п.п.
+ст.л.
+т.е.
+т.к.
+т.ч.
+т.д.
+т.п.
+и.о.
+с.г.
+г.р.
+т.н.
+т.ч.
+н.э.
diff --git a/corpus/support/token_patterns b/corpus/support/token_patterns
index 8e69432b..de64fb2a 100644
--- a/corpus/support/token_patterns
+++ b/corpus/support/token_patterns
@@ -1,3 +1,5 @@
 /^(al|el|ul|e)\-[a-z]+$/
+/^((а|А)(ль|ш)|уль)-\p{Cyrillic}+$/
+/^\p{Cyrillic}\.\p{Cyrillic}\.$/
 /^(\d|\d\d|\d\d\d)\.$/
 
diff --git a/corpus/support/tokenizer.pl b/corpus/support/tokenizer.pl
index e9c3a37d..0350a894 100755
--- a/corpus/support/tokenizer.pl
+++ b/corpus/support/tokenizer.pl
@@ -73,6 +73,7 @@ my $dict_file = "$workdir/token_list";
 my $word_patt_file = "$workdir/token_patterns";
 
 open(my $dict_fp, "$dict_file") or die;
+binmode($dict_fp, ":utf8");
 
 # read in the list of words that should not be segmented, 
 ##  e.g.,"I.B.M.", co-operation.
@@ -89,6 +90,7 @@ while(<$dict_fp>){
 }
 
 open(my $patt_fp, "$word_patt_file") or die;
+binmode($patt_fp, ":utf8");
 my @word_patts = ();
 my $word_patt_num = 0;
 while(<$patt_fp>){
@@ -147,7 +149,6 @@ while(<STDIN>){
 
     print STDOUT " $new_line\n";
 }
-print STDERR "\n";
 
 ########################################################################