Fix some Layer, Term & Format tests

PrinsINT · PrinsINT · commit 0afee27712f4 · 2025-12-19T10:57:55.000+01:00
diff --git a/server/src/main/kotlin/org/ivdnt/galahad/documents/DocumentFormat.kt b/server/src/main/kotlin/org/ivdnt/galahad/documents/DocumentFormat.kt
@@ -11,7 +11,6 @@ import javax.xml.stream.XMLStreamReader
 /** Format of a document. No 1-to-1 correspondence with extensions due to xml format overlap. */
 enum class DocumentFormat(val identifier: String, val extension: String) {
     TeiP4Legacy("tei-p4-legacy", "tei.xml"),
-    TeiP5Legacy("tei-p5-legacy", "tei.xml"),
     TeiP5("tei-p5", "tei.xml"),
     Naf("naf", "naf.xml"),
     Tsv("tsv", "tsv"),
diff --git a/server/src/main/kotlin/org/ivdnt/galahad/export/CorpusExport.kt b/server/src/main/kotlin/org/ivdnt/galahad/export/CorpusExport.kt
@@ -24,14 +24,7 @@ class CorpusExport private constructor(
     val shouldMerge: Boolean,
     val posHeadOnly: Boolean,
 ) : Logging {
-    private fun mergeFormatMatches(it: Document, format: DocumentFormat): Boolean {
-        var otherFormat = it.metadata.format
-        // Overwrite the format for legacy formats that can in fact be merged.
-        if (otherFormat == DocumentFormat.TeiP5Legacy) {
-            otherFormat = DocumentFormat.TeiP5
-        }
-        return otherFormat == format
-    }
+    private fun mergeFormatMatches(it: Document, format: DocumentFormat): Boolean = it.metadata.format == format
 
     private fun formatMapper(doc: Document, out: OutputStream) {
         try {
diff --git a/server/src/main/kotlin/org/ivdnt/galahad/export/LayerMerger.kt b/server/src/main/kotlin/org/ivdnt/galahad/export/LayerMerger.kt
@@ -20,7 +20,7 @@ abstract class LayerMerger protected constructor(protected val export: DocumentE
             DocumentFormat.Tsv -> TsvMerger(export)
             DocumentFormat.Folia -> FoliaMerger(export)
             DocumentFormat.Conllu -> ConlluMerger(export)
-            DocumentFormat.TeiP5Legacy, DocumentFormat.TeiP5 -> TeiMerger(export)
+            DocumentFormat.TeiP5 -> TeiMerger(export)
             else -> throw InvalidDocumentFormatException("Unsupported export conversion format: ${export.format}")
         }
     }
diff --git a/server/src/main/kotlin/org/ivdnt/galahad/formats/InternalFile.kt b/server/src/main/kotlin/org/ivdnt/galahad/formats/InternalFile.kt
@@ -32,7 +32,6 @@ abstract class InternalFile protected constructor() {
                 DocumentFormat.Pdf -> PdfFile(file)
                 // Multiple TEI formats
                 DocumentFormat.TeiP4Legacy,
-                DocumentFormat.TeiP5Legacy,
                 DocumentFormat.TeiP5,
                     -> TeiFile(file, format)
             }
diff --git a/server/src/test/kotlin/org/ivdnt/galahad/documents/DocumentMetadataTest.kt b/server/src/test/kotlin/org/ivdnt/galahad/documents/DocumentMetadataTest.kt
@@ -18,28 +18,25 @@ class DocumentMetadataTest {
 
     @Test
     fun `Properties for an unannotated file`() {
-        val path = "formats/shared/converter/input.txt"
-        val file = TestUtil.get(path)
-        val plaintext = file.readText()
+        val path = "formats/shared/converter/karel_en_martijn.txt"
         val doc = TestUtil.getDoc(path)
         val meta = doc.metadata
-        assertEquals("input.txt", meta.name)
+        assertEquals("karel_en_martijn.txt", meta.name)
         assertEquals(DocumentFormat.Txt, meta.format)
-        assertEquals(plaintext, meta.text) // This works because the preview is < MAX_PREVIEW_LENGTH
+        assert(meta.text.contains("Fraaie historie ende alwaer"))
         val total = meta.annotations.annotations[Annotation.TOKEN]
-        assertEquals(0, total)
+        assertEquals(39, total)
     }
 
     @Test
     fun `Properties for an annotated file`() {
-        val path = "all-formats/input/input.tei.xml"
+        val path = "formats/shared/converter/karel_en_martijn.tei.xml"
         val doc = TestUtil.getDoc(path)
-        val plaintext = TestUtil.getLayer(doc).toString()
         val meta = doc.metadata
-        assertEquals("input.tei.xml", meta.name)
+        assertEquals("karel_en_martijn.tei.xml", meta.name)
         assertEquals(DocumentFormat.TeiP5, meta.format)
-        assertEquals(plaintext, meta.text) // This works because the preview is < MAX_PREVIEW_LENGTH
+        assert(meta.text.contains("Fraaie historie ende alwaer"))
         val total = meta.annotations.annotations[Annotation.TOKEN]
-        assertEquals(21, total)
+        assertEquals(52, total)
     }
 }
diff --git a/server/src/test/kotlin/org/ivdnt/galahad/documents/DocumentsTest.kt b/server/src/test/kotlin/org/ivdnt/galahad/documents/DocumentsTest.kt
@@ -19,22 +19,22 @@ class DocumentsTest {
     @Test
     fun `Create and delete files`() {
         // Add two files
-        addFile("all-formats/input/input.folia.xml")
-        addFile("all-formats/input/input.conllu")
+        addFile("formats/shared/converter/karel_en_martijn.folia.xml")
+        addFile("formats/shared/converter/karel_en_martijn.conllu")
         // 2 files exist
-        assertDocsinDocuments(setOf("input.folia.xml", "input.conllu"))
+        assertDocsinDocuments(setOf("karel_en_martijn.folia.xml", "karel_en_martijn.conllu"))
         // Delete the first
-        deleteFile("input.folia.xml")
+        deleteFile("karel_en_martijn.folia.xml")
         // 1 is left
-        assertDocsinDocuments(setOf("input.conllu"))
+        assertDocsinDocuments(setOf("karel_en_martijn.conllu"))
         // Try to access deleted file
-        assertFileDeleted("input.folia.xml")
+        assertFileDeleted("karel_en_martijn.folia.xml")
         // Delete the last file
-        deleteFile("input.conllu")
+        deleteFile("karel_en_martijn.conllu")
         // 0 left
         assertDocsinDocuments(setOf())
         // Try to access deleted file
-        assertFileDeleted("input.conllu")
+        assertFileDeleted("karel_en_martijn.conllu")
     }
 
     private fun addFile(path: String) {
diff --git a/server/src/test/kotlin/org/ivdnt/galahad/documents/FormatInducerTest.kt b/server/src/test/kotlin/org/ivdnt/galahad/documents/FormatInducerTest.kt
@@ -8,9 +8,9 @@ import org.junit.jupiter.api.Test
 class FormatInducerTest {
     @Test
     fun `Parse simple formats based on extension`() {
-        assertEquals(DocumentFormat.Tsv, DocumentFormat.fromFile(TestUtil.get("all-formats/input/input.tsv")))
-        assertEquals(DocumentFormat.Conllu, DocumentFormat.fromFile(TestUtil.get("all-formats/input/input.conllu")))
-        assertEquals(DocumentFormat.Txt, DocumentFormat.fromFile(TestUtil.get("all-formats/input/input.txt")))
+        assertEquals(DocumentFormat.Tsv, DocumentFormat.fromFile(TestUtil.get("formats/shared/converter/karel_en_martijn.tsv")))
+        assertEquals(DocumentFormat.Conllu, DocumentFormat.fromFile(TestUtil.get("formats/shared/converter/karel_en_martijn.conllu")))
+        assertEquals(DocumentFormat.Txt, DocumentFormat.fromFile(TestUtil.get("formats/shared/converter/karel_en_martijn.txt")))
     }
 
     @Test
@@ -20,26 +20,20 @@ class FormatInducerTest {
 
     @Test
     fun `Parse non-legacy XML formats`() {
-        assertEquals(DocumentFormat.Folia, DocumentFormat.fromFile(TestUtil.get("all-formats/input/input.folia.xml")))
-        assertEquals(DocumentFormat.TeiP5, DocumentFormat.fromFile(TestUtil.get("all-formats/input/input.tei.xml")))
-        assertEquals(DocumentFormat.Naf, DocumentFormat.fromFile(TestUtil.get("all-formats/input/input.naf.xml")))
+        assertEquals(DocumentFormat.Folia, DocumentFormat.fromFile(TestUtil.get("formats/shared/converter/karel_en_martijn.folia.xml")))
+        assertEquals(DocumentFormat.TeiP5, DocumentFormat.fromFile(TestUtil.get("formats/shared/converter/karel_en_martijn.tei.xml")))
+        assertEquals(DocumentFormat.TeiP5, DocumentFormat.fromFile(TestUtil.get("formats/tei/reader/teicorpus/input.tei.xml")))
+        assertEquals(DocumentFormat.Naf, DocumentFormat.fromFile(TestUtil.get("formats/shared/converter/karel_en_martijn.naf.xml")))
     }
 
     @Test
     fun `Parse legacy XML formats`() {
-        assertEquals(DocumentFormat.TeiP4Legacy, DocumentFormat.fromFile(TestUtil.get("all-formats/tei/teip4legacy.xml")))
-        // if 1 or more pos are present, it's TeiP5
-        assertEquals(DocumentFormat.TeiP5, DocumentFormat.fromFile(TestUtil.get("all-formats/tei/1-pos-0-type.tei.xml")))
-        // even if there are types, it's still TeiP5
-        assertEquals(DocumentFormat.TeiP5, DocumentFormat.fromFile(TestUtil.get("all-formats/tei/1-pos-1-type.tei.xml")))
-        // if no pos or type are present, it's unannotated and we default to TeiP5
-        assertEquals(DocumentFormat.TeiP5, DocumentFormat.fromFile(TestUtil.get("all-formats/tei/0-pos-0-type.tei.xml")))
-        // if no pos are present, but at least one type is present, it's TeiP5Legacy
-        assertEquals(DocumentFormat.TeiP5Legacy, DocumentFormat.fromFile(TestUtil.get("all-formats/tei/0-pos-1-type.tei.xml")))
+        assertEquals(DocumentFormat.TeiP4Legacy, DocumentFormat.fromFile(TestUtil.get("formats/tei/reader/teip4/input.tei.xml")))
 
     }
 
     @Test
     fun `Parse unknown XML root node`() {
+        assertThrows(Exception::class.java) { DocumentFormat.fromFile(TestUtil.get("documents/invalid-root.xml")) }
     }
 }
diff --git a/server/src/test/kotlin/org/ivdnt/galahad/evaluation/EvaluationEntryTest.kt b/server/src/test/kotlin/org/ivdnt/galahad/evaluation/EvaluationEntryTest.kt
@@ -14,7 +14,6 @@ class EvaluationEntryTest {
         val b = EvaluationEntry(1, mutableListOf())
         val result = EvaluationEntry.add(a, b)
         assertEquals(2, result.count)
-        assertEquals(0, result.jsonSamples.size)
         assertEquals(0, result.samples.size)
     }
 
@@ -24,7 +23,6 @@ class EvaluationEntryTest {
         val b = EvaluationEntry(4, mutableListOf(TermComparison(Term.EMPTY, Term.EMPTY)))
         val result = EvaluationEntry.add(a, b)
         assertEquals(7, result.count)
-        assertEquals(2, result.jsonSamples.size)
         assertEquals(2, result.samples.size)
     }
 
@@ -34,12 +32,10 @@ class EvaluationEntryTest {
         val b = EvaluationEntry(9, MutableList(9) { TermComparison(Term.EMPTY, Term.EMPTY) })
         var result = EvaluationEntry.add(a, b)
         assertEquals(16, result.count)
-        assertEquals(10, result.jsonSamples.size)
         assertEquals(16, result.samples.size)
         val c = EvaluationEntry(3, MutableList(3) { TermComparison(Term.EMPTY, Term.EMPTY) })
         result = EvaluationEntry.add(result, c)
         assertEquals(19, result.count)
-        assertEquals(10, result.jsonSamples.size)
         assertEquals(16, result.samples.size) // should not grow once the truncation limit is reached
     }
 
diff --git a/server/src/test/kotlin/org/ivdnt/galahad/evaluation/LayerComparisonTest.kt b/server/src/test/kotlin/org/ivdnt/galahad/evaluation/LayerComparisonTest.kt
@@ -1,6 +1,7 @@
 package org.ivdnt.galahad.evaluation
 
 import org.ivdnt.galahad.annotations.Annotation
+import org.ivdnt.galahad.annotations.Term
 import org.ivdnt.galahad.evaluation.comparison.*
 import org.ivdnt.galahad.util.LayerBuilder
 import org.junit.jupiter.api.Test
@@ -11,7 +12,7 @@ import org.junit.jupiter.api.Nested
 class LayerComparisonTest {
 
     @Nested
-    inner class PosFilter {
+    inner class TermFilter {
         @Test
         fun `Matching layers with pos filter`() {
             val builder =
@@ -26,10 +27,24 @@ class LayerComparisonTest {
             assertEquals(100, comparison.matches.size)
         }
 
+        @Test
+        fun `Layers with partial match`() {
+            val hypo = LayerBuilder().loadDummies(100, pos = "NOU").build()
+            val ref = LayerBuilder().loadDummies(50, pos = "NOU").loadDummies(20, pos = "VRB").loadDummies(30, pos = "ADJ").build()
+            val comparison = LayerComparison(
+                hypothesis = hypo,
+                reference = ref,
+                filter = NouVrbFilter()
+            )
+            assertEquals(20, comparison.matches.size)
+        }
+
         private fun NouNouFilter(): LayerFilter {
             val termFilter = HeadGroupTermFilter(Annotation.POS, "NOU")
             return ConfusionLayerFilter(termFilter, termFilter)
         }
+        private fun NouVrbFilter(): LayerFilter =
+            ConfusionLayerFilter(HeadGroupTermFilter(Annotation.POS, "NOU"), HeadGroupTermFilter(Annotation.POS, "VRB"))
 
         @Test
         fun `Matching layers with pos filter, but different pos`() {
@@ -49,10 +64,10 @@ class LayerComparisonTest {
     inner class Punctuation {
         @Test
         fun `Matching layers despite punctuation difference`() {
-            assertPunctuationDifference("dummy, dummy", 2, 1, 0)
-            assertPunctuationDifference("dummy dummy, dummy", 3, 1, 0)
-            assertPunctuationDifference("dummy dummy, dummy dummy", 4, 1, 0)
-            assertPunctuationDifference("dummy, dummy, dummy,", 3, 3, 0)
+            assertPunctuationDifference("dummy, dummy", 3, 0, 1)
+            assertPunctuationDifference("dummy dummy, dummy", 4, 0, 1)
+            assertPunctuationDifference("dummy dummy, dummy dummy", 5, 0, 1)
+            assertPunctuationDifference("dummy, dummy, dummy,", 6, 0, 3)
         }
 
         // Because we only fix for one punctuation mark at the end of a word. No more.
@@ -63,7 +78,9 @@ class LayerComparisonTest {
             val ref = LayerBuilder() // dummy":
                 .loadDummies(1, "dummy\":").build()
             val comparison = LayerComparison(hypothesis = hypo, reference = ref)
-            assertEquals(0, comparison.matches.size)
+            assertEquals(3, comparison.matches.size)
+            assertEquals(2, comparison.matches.filter { it.ref == Term.EMPTY }.size)
+            assertEquals(0, comparison.matches.filter { it.hyp == Term.EMPTY }.size)
         }
 
         /**
@@ -96,6 +113,8 @@ class LayerComparisonTest {
             // Compare
             val comparison = LayerComparison(hypothesis = hypo, reference = ref)
             assertEquals(numMatches, comparison.matches.size)
+            assertEquals(numHypoMissing, comparison.matches.filter { it.hyp == Term.EMPTY }.size)
+            assertEquals(numRefMissing, comparison.matches.filter { it.ref == Term.EMPTY }.size)
         }
     }
 
@@ -113,6 +132,10 @@ class LayerComparisonTest {
             val ref = LayerBuilder().loadDummies(numRefTerms).build()
             val comparison = LayerComparison(hypothesis = hypo, reference = ref)
             assertEquals(numMatches, comparison.matches.size)
+            val hypoNoMatch = comparison.matches.filter { it.hyp == Term.EMPTY }
+            val refNoMatch = comparison.matches.filter { it.ref == Term.EMPTY }
+            assertEquals(numHypoMissing, hypoNoMatch.size)
+            assertEquals(numRefMissing, refNoMatch.size)
         }
 
         @Test
@@ -122,15 +145,15 @@ class LayerComparisonTest {
 
         @Test
         fun `A layer is 1 term larger than the other`() {
-            assertLayers(numHypoTerms = 2, numRefTerms = 3, numMatches = 2, numHypoMissing = 0, numRefMissing = 1)
-            assertLayers(numHypoTerms = 100, numRefTerms = 101, numMatches = 100, numHypoMissing = 0, numRefMissing = 1)
-            assertLayers(numHypoTerms = 101, numRefTerms = 100, numMatches = 100, numHypoMissing = 1, numRefMissing = 0)
+            assertLayers(numHypoTerms = 2, numRefTerms = 3, numMatches = 3, numHypoMissing = 1, numRefMissing = 0)
+            assertLayers(numHypoTerms = 100, numRefTerms = 101, numMatches = 101, numHypoMissing = 1, numRefMissing = 0)
+            assertLayers(numHypoTerms = 101, numRefTerms = 100, numMatches = 101, numHypoMissing = 0, numRefMissing = 1)
         }
 
         @Test
         fun `A layer is empty`() {
-            assertLayers(numHypoTerms = 100, numRefTerms = 0, numMatches = 0, numHypoMissing = 100, numRefMissing = 0)
-            assertLayers(numHypoTerms = 0, numRefTerms = 100, numMatches = 0, numHypoMissing = 0, numRefMissing = 100)
+            assertLayers(numHypoTerms = 100, numRefTerms = 0, numMatches = 100, numHypoMissing = 0, numRefMissing = 100)
+            assertLayers(numHypoTerms = 0, numRefTerms = 100, numMatches = 100, numHypoMissing = 100, numRefMissing = 0)
         }
     }
 }
diff --git a/server/src/test/kotlin/org/ivdnt/galahad/evaluation/TermComparisonTest.kt b/server/src/test/kotlin/org/ivdnt/galahad/evaluation/TermComparisonTest.kt
@@ -16,8 +16,8 @@ class TermComparisonTest {
             hypoLemma: String? = "school", refLemma: String? = "school",
             hypoPos: String? = "NOU", refPos: String? = "NOU",
         ) {
-            val hypoTerm = Term("", 0, mapOf(Annotation.LEMMA to hypoLemma, Annotation.POS to hypoPos))
-            val refTerm = Term("", 0, mapOf(Annotation.LEMMA to refLemma, Annotation.POS to refPos))
+            val hypoTerm = Term("", 0, mapOf(Annotation.TOKEN to "dummy", Annotation.LEMMA to hypoLemma, Annotation.POS to hypoPos))
+            val refTerm = Term("", 0, mapOf(Annotation.TOKEN to "dummy", Annotation.LEMMA to refLemma, Annotation.POS to refPos))
             TermComparison(hypoTerm, refTerm).apply {
                 assertEquals(lemmaEqual, equal(Annotation.LEMMA))
                 assertEquals(posEqual, equal(Annotation.POS))
diff --git a/server/src/test/kotlin/org/ivdnt/galahad/taggers/TaggerStoreControllerTest.kt b/server/src/test/kotlin/org/ivdnt/galahad/taggers/TaggerStoreControllerTest.kt
@@ -25,6 +25,6 @@ class TaggerStoreControllerTest(
 
     @Test
     fun `Get health of invalid tagger`() {
-        assertEquals(TaggerHealthStatus.ERROR, ctrl.getTaggerHealth("invalid").status)
+        assertThrows<TaggerNotFoundException> { ctrl.getTaggerHealth("invalid").status }
     }
 }

Original file line number	Diff line number	Diff line change
`@@ -20,7 +20,7 @@ abstract class LayerMerger protected constructor(protected val export: DocumentE`
`20`	`20`	`DocumentFormat.Tsv -> TsvMerger(export)`
`21`	`21`	`DocumentFormat.Folia -> FoliaMerger(export)`
`22`	`22`	`DocumentFormat.Conllu -> ConlluMerger(export)`
`23`		`- DocumentFormat.TeiP5Legacy, DocumentFormat.TeiP5 -> TeiMerger(export)`
	`23`	`+ DocumentFormat.TeiP5 -> TeiMerger(export)`
`24`	`24`	`else -> throw InvalidDocumentFormatException("Unsupported export conversion format: ${export.format}")`
`25`	`25`	`}`
`26`	`26`	`}`
Original file line number	Diff line number	Diff line change
`@@ -32,7 +32,6 @@ abstract class InternalFile protected constructor() {`
`32`	`32`	`DocumentFormat.Pdf -> PdfFile(file)`
`33`	`33`	`// Multiple TEI formats`
`34`	`34`	`DocumentFormat.TeiP4Legacy,`
`35`		`- DocumentFormat.TeiP5Legacy,`
`36`	`35`	`DocumentFormat.TeiP5,`
`37`	`36`	`-> TeiFile(file, format)`
`38`	`37`	`}`