Added langCode, withoutFormatExt, TeiMetadata, distinguish PID and title

Author: PrinsINT

server/src/main/kotlin/org/ivdnt/galahad/annotations/Layer.kt

@@ -10,7 +10,7 @@ const val SOURCE_LAYER_NAME: String = "sourceLayer"
 class Layer(
     /** Documents in this layer. (Formats like conllu support multiple documents.) */
     val documents: Array<DocumentLayer>,
-    /** ID of this layer. Ideally the file PID as documents may have different ids. */
+    /** ID of this layer. Ideally this is the file PID, so that documents may have different ids. */
     val id: String = UUID.randomUUID().toString(),
 ) {
     /** Terms in this layer. Documents, paragraphs, sentences flattened. */

server/src/main/kotlin/org/ivdnt/galahad/corpora/MutableCorpusMetadata.kt

@@ -33,6 +33,11 @@ open class MutableCorpusMetadata(
     @JsonIgnore
     var user: User? = null
 
+    @get:JsonIgnore
+    val langCode: String
+        // iso 639 code. "und" for undefined.
+        get() = Locale.getAvailableLocales().find { it.getDisplayLanguage(Locale.ENGLISH).equals(language, true) }?.isO3Language ?: "und"
+
     /**
      * Whether the user is in the list of collaborators of this corpus.
      * Note that this is not the same as having write access: use [hasWriteAccess].

server/src/main/kotlin/org/ivdnt/galahad/documents/DocumentFormat.kt

@@ -26,6 +26,9 @@ enum class DocumentFormat(val identifier: String, val extension: String) {
     override fun toString(): String = identifier
 
     companion object {
+        val extensions: Set<String>
+            get() = entries.map { "." + it.extension }.toSet()
+
         // Used by Spring.
         @JsonCreator
         fun fromString(s: String): DocumentFormat =

server/src/main/kotlin/org/ivdnt/galahad/export/CmdiMetadata.kt

@@ -7,6 +7,7 @@ import org.ivdnt.galahad.util.XmlUtil
 import org.ivdnt.galahad.util.child
 import org.ivdnt.galahad.util.childElements
 import org.ivdnt.galahad.util.ifNullOrBlank
+import org.ivdnt.galahad.util.withoutFormatExt
 import org.w3c.dom.Element
 import org.w3c.dom.Node
 import java.io.OutputStream
@@ -22,7 +23,7 @@ import javax.xml.transform.stream.StreamResult
  * the metadata values in the DOM with each instance.
  */
 class CmdiMetadata(val export: DocumentExport) {
-    private val docTitle = export.document.uploadedFile.nameWithoutExtension
+    private val docTitle = export.document.uploadedFile.withoutFormatExt
     private val corpus: MutableCorpusMetadata = export.corpus.mutableMetadata
     private val format = export.format.identifier
     private val now = Date()

server/src/main/kotlin/org/ivdnt/galahad/formats/naf/NafWriter.kt

@@ -4,6 +4,7 @@ import org.ivdnt.galahad.annotations.Annotation
 import org.ivdnt.galahad.export.DocumentExport
 import org.ivdnt.galahad.export.LayerWriter
 import org.ivdnt.galahad.util.XmlUtil
+import org.ivdnt.galahad.util.withoutFormatExt
 import org.w3c.dom.Document
 import org.w3c.dom.Element
 import java.io.OutputStream
@@ -17,7 +18,7 @@ class NafWriter(export: DocumentExport) : LayerWriter(export) {
         val xml = XmlUtil.builder.newDocument()
         val root = xml.createElement("NAF").apply {
             setAttribute("version", "v3.3")
-            setAttribute("xml:lang", "dum")
+            setAttribute("xml:lang", export.corpus.mutableMetadata.langCode)
         }
         xml.appendChild(root)
 
@@ -40,7 +41,7 @@ class NafWriter(export: DocumentExport) : LayerWriter(export) {
         root.appendChild(nafHeader)
 
         val fileDesc = xml.createElement("fileDesc").apply {
-            setAttribute("title", export.document.name)
+            setAttribute("title", export.document.uploadedFile.withoutFormatExt)
             setAttribute("author", export.user.id)
             setAttribute("creationtime", now.toString())
             setAttribute("filename", export.document.name)

server/src/main/kotlin/org/ivdnt/galahad/formats/tei/TeiMetadata.kt

@@ -6,6 +6,7 @@ import org.ivdnt.galahad.export.DocumentExport
 import org.ivdnt.galahad.util.XmlUtil
 import org.ivdnt.galahad.util.childOrNull
 import org.ivdnt.galahad.util.ifNullOrBlank
+import org.ivdnt.galahad.util.withoutFormatExt
 import org.w3c.dom.Document
 import org.w3c.dom.Element
 import org.w3c.dom.Node
@@ -32,7 +33,7 @@ class TeiMetadata(
                 ?.childOrNull("titleStmt")
                 ?.childOrNull("title")?.textContent
                 ?: // if null, use filename without extension
-                export.document.uploadedFile.nameWithoutExtension
+                export.document.uploadedFile.withoutFormatExt
         }
 
     private val corpusMetadata: MutableCorpusMetadata = export.corpus.mutableMetadata