summaryrefslogtreecommitdiff
path: root/word-aligner/ortho-norm/cz.pl
diff options
context:
space:
mode:
authorredpony <redpony@ec762483-ff6d-05da-a07a-a48fb63a330f>2010-10-20 17:19:17 +0000
committerredpony <redpony@ec762483-ff6d-05da-a07a-a48fb63a330f>2010-10-20 17:19:17 +0000
commiteb43ccf1e1a7d59a0f672956230644ebb7a5cbf1 (patch)
treec46852a34a5ded277b2330d4fe5338ad65defd15 /word-aligner/ortho-norm/cz.pl
parente6c49facea8bc83f348e112872a837048b5ca6ad (diff)
cz norm
git-svn-id: https://ws10smt.googlecode.com/svn/trunk@685 ec762483-ff6d-05da-a07a-a48fb63a330f
Diffstat (limited to 'word-aligner/ortho-norm/cz.pl')
-rwxr-xr-xword-aligner/ortho-norm/cz.pl17
1 files changed, 17 insertions, 0 deletions
diff --git a/word-aligner/ortho-norm/cz.pl b/word-aligner/ortho-norm/cz.pl
new file mode 100755
index 00000000..9d8bb07b
--- /dev/null
+++ b/word-aligner/ortho-norm/cz.pl
@@ -0,0 +1,17 @@
+#!/usr/bin/perl -w
+use strict;
+use utf8;
+
+while(<STDIN>) {
+ $_ = lc $_;
+ s/ al-/ al/g;
+ s/^al-/al/;
+ s/j/i/g;
+ s/dž/j/g;
+ s/š/sh/g;
+ s/á/a/g;
+ s/č/ch/g;
+ s/í/i/g;
+ print;
+}
+