From ea83bf53c1fa99231d68f53ff7e2b3d0645c8dd9 Mon Sep 17 00:00:00 2001
From: Chris Dyer <cdyer@allegro.clab.cs.cmu.edu>
Date: Fri, 24 Oct 2014 15:31:16 -0400
Subject: conll2cdec conversion

---
 corpus/conll2cdec.pl | 39 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)
 create mode 100755 corpus/conll2cdec.pl

(limited to 'corpus')

diff --git a/corpus/conll2cdec.pl b/corpus/conll2cdec.pl
new file mode 100755
index 00000000..f65b86f8
--- /dev/null
+++ b/corpus/conll2cdec.pl
@@ -0,0 +1,39 @@
+#!/usr/bin/perl -w
+use strict;
+
+my @xx;
+my @yy;
+my @os;
+my $sec = undef;
+my $i = 0;
+while(<>) {
+  chomp;
+  if (/^\s*$/) {
+    print "<seg id=\"$i\"";
+    $i++;
+    for (my $j = 0; $j < $sec; $j++) {
+      my @oo = ();
+      for (my $k = 0; $k < scalar @xx; $k++) {
+        my $sym = $os[$k]->[$j];
+        $sym =~ s/"/'/g;
+        push @oo, $sym;
+      }
+      my $zz = $j + 1;
+      print " feat$zz=\"@oo\"";
+    }
+
+    print "> @xx ||| @yy </seg>\n";
+    @xx = ();
+    @yy = ();
+    @os = ();
+  } else {
+    my ($x, @fs) = split /\s+/;
+    my $y = pop @fs;
+    if (!defined $sec) { $sec = scalar @fs; }
+    die unless $sec == scalar @fs;
+    push @xx, $x;
+    push @yy, $y;
+    push @os, \@fs;
+  }
+}
+
-- 
cgit v1.2.3