tei-import: Map blank lemmas and pos to null

INL · Oct 16, 2024 · 70a5618 · 70a5618
1 parent aed57d4
commit 70a5618
Showing 1 changed file with 6 additions and 2 deletions.
diff --git a/server/src/main/kotlin/org/ivdnt/galahad/port/xml/BLFXMLParser.kt b/server/src/main/kotlin/org/ivdnt/galahad/port/xml/BLFXMLParser.kt
@@ -287,10 +287,14 @@ class BLFXMLParser (
 
         // Extraction
         val literal = literalExtractor(node).trim() // wordPathExpression.evaluate( node )
-        val lem = lemmaExtractor(node) // lemPathExpression.evaluate( node )
-        val pos = posExtractor(node) // posPathExpression.evaluate( node )
+        var lem: String? = lemmaExtractor(node) // lemPathExpression.evaluate( node )
+        var pos: String? = posExtractor(node) // posPathExpression.evaluate( node )
         val id = idExtractor(node)
 
+        // Map empty strings to null
+        if(lem.isNullOrBlank()) lem = null
+        if(pos.isNullOrBlank()) pos = null
+
         // Add the word to the source layer
         val wordForm = WordForm(literal, trueWordOffset, literal.length, id ?: "no-id" )
         val term = Term(lem, pos, mutableListOf(wordForm))