commit 4af5edea32dc2cdbf505a3ced0f93f4c8801200a
author Marc Kupietz <kupietz@ids-mannheim.de> Thu Mar 07 12:38:18 2024 +0100
committer Marc Kupietz <kupietz@ids-mannheim.de> Thu Mar 07 15:58:54 2024 +0100
tree 3be113213d70e130ba10ab5f38ca45a5a43f1bfd
parent 4d4b0c87f6a02292dcf9b43d59b8c86646388487

Print no offsets for empty sentences

Change-Id: Ia392ac704da90aea88912436c39e82eaebe09de1

app/src/main/kotlin/de/ids_mannheim/korapxmltools/KorapXml2Conllu.kt

diff --git a/app/src/main/kotlin/de/ids_mannheim/korapxmltools/KorapXml2Conllu.kt b/app/src/main/kotlin/de/ids_mannheim/korapxmltools/KorapXml2Conllu.kt
index a76694c..f10600d 100644
--- a/app/src/main/kotlin/de/ids_mannheim/korapxmltools/KorapXml2Conllu.kt
+++ b/app/src/main/kotlin/de/ids_mannheim/korapxmltools/KorapXml2Conllu.kt
@@ -415,12 +415,10 @@
         token_index: Int,
         tokens: ConcurrentHashMap<String, Array<Span>>
     ): String {
-        val sentenceEndOffset: Int
-        sentenceEndOffset = if (sentences[docId] == null) {
-            -1
-        } else {
-            sentences[docId]!![sentence_index].to
+        if (sentences[docId] == null || sentences[docId]!!.size <= sentence_index) {
+            return ""
         }
+        val sentenceEndOffset = sentences[docId]!![sentence_index].to
         var i = token_index
         val start_offsets_string = StringBuilder()
         val end_offsets_string = StringBuilder()