Always use the first fs for a token
Change-Id: Ibfa101cf7efd783f64d444e5f7ee3bbb112972e4
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 56f9e7e..1852c7c 100644
--- a/app/src/main/kotlin/de/ids_mannheim/korapxmltools/KorapXml2Conllu.kt
+++ b/app/src/main/kotlin/de/ids_mannheim/korapxmltools/KorapXml2Conllu.kt
@@ -434,13 +434,13 @@
val value = feature.textContent.trim()
if (value.isEmpty()) return@forEach
when (attr) {
- "lemma" -> if(fs.lemma == "_") fs.lemma = value
+ "lemma" -> if(fs.lemma == "_") fs.lemma = value.replace("UNKNOWN", "--")
"upos" -> fs.upos = value
- "xpos", "ctag", "pos" -> if(fs.xpos == "_") fs.xpos = value
+ "xpos", "ctag", "pos" -> if(fs.xpos == "_") fs.xpos = value.replace("UNKNOWN", "--")
"feats", "msd" -> if(fs.feats == "_" ) fs.feats = value
"type" -> if(fs.feats == "_") fs.feats = feature.getElementsByTagName("symbol").item(0).attributes.getNamedItem("value").textContent.trim()
// "subtype" -> if(fs.feats == "_") fs.feats += ":" + feature.getElementsByTagName("symbol").item(0).attributes.getNamedItem("value").textContent
- "certainty" -> fs.misc = value
+ "certainty" -> if(fs.misc == "_") fs.misc = value
}
}
res[fromTo] = fs