c2k: throw away alternative pos interpretations for now
(separated by '|')
Change-Id: Iee2e4b24e3e421b9ff8a547b85554ebfd92e79c9
diff --git a/script/conllu2korapxml b/script/conllu2korapxml
index 1fdad0d..a9bb030 100755
--- a/script/conllu2korapxml
+++ b/script/conllu2korapxml
@@ -135,12 +135,14 @@
</rel>
</span>
@;
- }
- $morpho .= qq( <span id="s${s}_n$t" from="$spansFrom[$t]" to="$spansTo[$t]">
+ }
+ my $pos = $parsed[3];
+ $pos =~ s/\|.*//;
+ $morpho .= qq( <span id="s${s}_n$t" from="$spansFrom[$t]" to="$spansTo[$t]">
<fs type="lex" xmlns="http://www.tei-c.org/ns/1.0">
<f name="lex">
<fs>
- <f name="pos">$parsed[3]</f>
+ <f name="pos">$pos</f>
);
$morpho .= qq( <f name="lemma">$parsed[2]</f>\n) if($parsed[2] ne "_" || $parsed[1] eq '_');
$morpho .= qq( <f name="msd">$parsed[5]</f>\n) if($parsed[5] ne "_");