commit 093b21cfc1af7c868e46a810afdfb736ce32d835
author Marc Kupietz <kupietz@ids-mannheim.de> Sat Jul 31 23:39:51 2021 +0200
committer Marc Kupietz <kupietz@ids-mannheim.de> Sat Jul 31 23:53:07 2021 +0200
tree c57968b37efdc7fc511d9b7f6170fd02712df1ce
parent 1db65e5492f819beab80044e0e7d918c3282210d

Fix structure not being read for last text

Change-Id: Id4b7c86312d4a70eccaffd131180c163a1133bb0

script/korapxml2conllu

diff --git a/script/korapxml2conllu b/script/korapxml2conllu
index b4f673d..dd01ba6 100755
--- a/script/korapxml2conllu
+++ b/script/korapxml2conllu
@@ -301,7 +301,7 @@
       s/&/&/go;
       tr/…•⋅»«ˮ“”„›‹ʼ‘’‚′‐‑‒–—―⁓⁻₋−﹣－/...""""""'''''''-/;
       $plain_texts{$docid} = $_;
-      last if($text_count++ > 1 && $plain_texts{$target_id});
+      last if(!$extract_attributes_regex && ($text_count++ > 1 && $plain_texts{$target_id}));
     } elsif (m@<text>(.*)@) {
       $_= decode("utf-8", $1, Encode::FB_DEFAULT);
       s/&lt;/</go;
@@ -318,7 +318,7 @@
       tr/…•⋅»«ˮ“”„›‹ʼ‘’‚′‐‑‒–—―⁓⁻₋−﹣－/...""""""'''''''-/;
       $plain_texts{$docid} .= $_;
       $text_started=0;
-      last if($text_count++ > 1 && $plain_texts{$target_id});
+      last if(!$extract_attributes_regex && ($text_count++ > 1 && $plain_texts{$target_id}));
     } elsif ($text_started) {
       chomp;
       $_ = decode("utf-8", $_, Encode::FB_DEFAULT) . ' ';