feat(lapis): omit sequences with null values in unaligned fasta dow…

…nloads

closes #912

lapis/src/main/kotlin/org/genspectrum/lapis/controller/ControllerDescriptions.kt

@@ -36,7 +36,8 @@ const val ALIGNED_MULTI_SEGMENTED_NUCLEOTIDE_SEQUENCE_ENDPOINT_DESCRIPTION =
     Only sequences matching the specified sequence filters are considered."""
 const val UNALIGNED_MULTI_SEGMENTED_NUCLEOTIDE_SEQUENCE_ENDPOINT_DESCRIPTION =
     """Returns a string of fasta formatted unaligned nucleotide sequences of the requested segment.
-    Only sequences matching the specified sequence filters are considered."""
+    Only sequences matching the specified sequence filters are considered.
+    Sequences that don't have an unaligned sequence for this segment are omitted."""
 
 const val AGGREGATED_GROUP_BY_FIELDS_DESCRIPTION =
     """The fields to stratify by. If empty, only the overall count is returned"""

lapis/src/main/kotlin/org/genspectrum/lapis/response/SiloResponse.kt

@@ -8,6 +8,7 @@ import com.fasterxml.jackson.databind.JsonDeserializer
 import com.fasterxml.jackson.databind.JsonNode
 import com.fasterxml.jackson.databind.JsonSerializer
 import com.fasterxml.jackson.databind.SerializerProvider
+import com.fasterxml.jackson.databind.node.JsonNodeType
 import com.fasterxml.jackson.databind.node.NullNode
 import io.swagger.v3.oas.annotations.media.Schema
 import jakarta.servlet.http.HttpServletResponse
@@ -103,7 +104,7 @@ data class InsertionData(
 
 data class SequenceData(
     val sequenceKey: String,
-    val sequence: String,
+    val sequence: String?,
 ) {
     fun writeAsString() = ">$sequenceKey\n$sequence"
 }
@@ -116,10 +117,9 @@ fun Stream<SequenceData>.writeFastaTo(
     if (response.contentType == null) {
         response.contentType = MediaType(TEXT_X_FASTA, Charset.defaultCharset()).toString()
     }
-    response.outputStream.writer().use {
-        for (sequenceData in this) {
-            it.appendLine(sequenceData.writeAsString())
-        }
+    response.outputStream.writer().use { stream ->
+        this.filter { it.sequence != null }
+            .forEach { stream.appendLine(it.writeAsString()) }
     }
 }
 
@@ -135,8 +135,19 @@ class SequenceDataDeserializer(val databaseConfig: DatabaseConfig) : JsonDeseria
     ): SequenceData {
         val node = p.readValueAsTree<JsonNode>()
         val sequenceKey = node.get(databaseConfig.schema.primaryKey).asText()
-        val sequence =
-            node.fields().asSequence().first { it.key != databaseConfig.schema.primaryKey }.value.asText()
-        return SequenceData(sequenceKey, sequence)
+
+        val value = node.fields().asSequence().first { it.key != databaseConfig.schema.primaryKey }.value
+        val sequence = when (value.nodeType) {
+            JsonNodeType.NULL -> null
+            JsonNodeType.STRING -> value.asText()
+            else -> throw RuntimeException(
+                "Error deserializing sequence data: got unexpected value $value of type ${value.nodeType}",
+            )
+        }
+
+        return SequenceData(
+            sequenceKey = sequenceKey,
+            sequence = sequence,
+        )
     }
 }

lapis/src/test/kotlin/org/genspectrum/lapis/controller/LapisControllerTest.kt

@@ -256,7 +256,11 @@ class LapisControllerTest(
                 SequenceType.ALIGNED,
                 "geneName",
             )
-        } returns Stream.of(SequenceData("key1", "the sequence"), SequenceData("key2", "the other sequence"))
+        } returns Stream.of(
+            SequenceData("key1", "the sequence"),
+            SequenceData("key2", "the other sequence"),
+            SequenceData("key3", null),
+        )
 
         mockMvc.perform(request("$ALIGNED_AMINO_ACID_SEQUENCES_ROUTE/geneName"))
             .andExpect(status().isOk)

lapis/src/test/kotlin/org/genspectrum/lapis/controller/MultiSegmentedSequenceControllerTest.kt

@@ -33,7 +33,11 @@ import java.util.stream.Stream
 class MultiSegmentedSequenceControllerTest(
     @Autowired val mockMvc: MockMvc,
 ) {
-    val returnedValue: Stream<SequenceData> = Stream.of(SequenceData("sequenceKey", "theSequence"))
+    val returnedValue: Stream<SequenceData> = Stream.of(
+        SequenceData("sequenceKey", "theSequence"),
+        SequenceData("sequenceKeyWithNullValue", null),
+    )
+
     val expectedFasta = """
         >sequenceKey
         theSequence

lapis/src/test/kotlin/org/genspectrum/lapis/controller/SingleSegmentedSequenceControllerTest.kt

@@ -31,7 +31,11 @@ import java.util.stream.Stream
 class SingleSegmentedSequenceControllerTest(
     @Autowired val mockMvc: MockMvc,
 ) {
-    val returnedValue: Stream<SequenceData> = Stream.of(SequenceData("sequenceKey", "theSequence"))
+    val returnedValue: Stream<SequenceData> = Stream.of(
+        SequenceData("sequenceKey", "theSequence"),
+        SequenceData("sequenceKeyWithNullValue", null),
+    )
+
     val expectedFasta = """
         >sequenceKey
         theSequence

lapis/src/test/kotlin/org/genspectrum/lapis/silo/SiloClientTest.kt

@@ -149,6 +149,7 @@ class SiloClientTest(
                     """
                         {"primaryKey": "key1","someSequenceName": "ABCD"}
                         {"primaryKey": "key2","someSequenceName": "DEFG"}
+                        {"primaryKey": "key3","someSequenceName": null}
                     """,
                 ),
         )
@@ -159,12 +160,13 @@ class SiloClientTest(
         )
         val result = underTest.sendQuery(query).toList()
 
-        assertThat(result, hasSize(2))
+        assertThat(result, hasSize(3))
         assertThat(
             result,
             containsInAnyOrder(
-                SequenceData("key1", "ABCD"),
-                SequenceData("key2", "DEFG"),
+                SequenceData(sequenceKey = "key1", sequence = "ABCD"),
+                SequenceData(sequenceKey = "key2", sequence = "DEFG"),
+                SequenceData(sequenceKey = "key3", sequence = null),
             ),
         )
     }