From cfab8c127b55fdf0016308cfee1cab9ce990dcce Mon Sep 17 00:00:00 2001
From: Jonas Natten <jonas.natten@knowit.no>
Date: Wed, 5 Feb 2025 08:28:51 +0100
Subject: [PATCH] concept-api: Migrate `:`-separated tags to separate tags

---
 .../db/migration/V22__SplitTags.scala         | 37 +++++++++++++++
 .../db/migration/V22__SplitTagsTest.scala     | 45 +++++++++++++++++++
 2 files changed, 82 insertions(+)
 create mode 100644 concept-api/src/main/scala/no/ndla/conceptapi/db/migration/V22__SplitTags.scala
 create mode 100644 concept-api/src/test/scala/no/ndla/conceptapi/db/migration/V22__SplitTagsTest.scala

diff --git a/concept-api/src/main/scala/no/ndla/conceptapi/db/migration/V22__SplitTags.scala b/concept-api/src/main/scala/no/ndla/conceptapi/db/migration/V22__SplitTags.scala
new file mode 100644
index 0000000000..0ff82057f7
--- /dev/null
+++ b/concept-api/src/main/scala/no/ndla/conceptapi/db/migration/V22__SplitTags.scala
@@ -0,0 +1,37 @@
+/*
+ * Part of NDLA concept-api
+ * Copyright (C) 2025 NDLA
+ *
+ * See LICENSE
+ *
+ */
+
+package no.ndla.conceptapi.db.migration
+
+import io.circe.parser
+import io.circe.syntax.EncoderOps
+import io.circe.generic.auto.*
+import no.ndla.database.DocumentMigration
+
+case class TagsObject(tags: List[String], language: String)
+
+class V22__SplitTags extends DocumentMigration {
+  override val columnName: String = "document"
+  override val tableName: String  = "conceptdata"
+
+  private def convertTags(tags: List[TagsObject]): List[TagsObject] = tags.map { to =>
+    val splitTags = to.tags.flatMap(_.split(":")).filterNot(_.isEmpty)
+    to.copy(tags = splitTags)
+  }
+
+  override def convertColumn(document: String): String = {
+    val oldDocument = parser.parse(document).toTry.get
+    oldDocument.hcursor.downField("tags").as[Option[List[TagsObject]]].toTry.get match {
+      case None => document
+      case Some(tags) =>
+        val convertedTags = convertTags(tags).asJson
+        val newDocument   = oldDocument.mapObject(_.remove("tags").add("tags", convertedTags))
+        newDocument.noSpaces
+    }
+  }
+}
diff --git a/concept-api/src/test/scala/no/ndla/conceptapi/db/migration/V22__SplitTagsTest.scala b/concept-api/src/test/scala/no/ndla/conceptapi/db/migration/V22__SplitTagsTest.scala
new file mode 100644
index 0000000000..bd46b146d5
--- /dev/null
+++ b/concept-api/src/test/scala/no/ndla/conceptapi/db/migration/V22__SplitTagsTest.scala
@@ -0,0 +1,45 @@
+/*
+ * Part of NDLA concept-api
+ * Copyright (C) 2025 NDLA
+ *
+ * See LICENSE
+ *
+ */
+
+package no.ndla.conceptapi.db.migration
+
+import no.ndla.conceptapi.UnitSuite
+
+class V22__SplitTagsTest extends UnitSuite {
+
+  val migration = new V22__SplitTags
+
+  test("That converting colon separated tags works") {
+    val oldDocument =
+      """{"tags":[{"tags":["tag1:tag2:tag3"],"language":"nb"},{"tags":["tag4:tag5:tag6"],"language":"en"}]}"""
+    val converted = migration.convertColumn(oldDocument)
+    val expectedDocument =
+      """{"tags":[{"tags":["tag1","tag2","tag3"],"language":"nb"},{"tags":["tag4","tag5","tag6"],"language":"en"}]}"""
+    converted should be(expectedDocument)
+  }
+
+  test("That converting colon separated tags works with empty values") {
+    val oldDocument      = """{"tags":[{"tags":["tag1::","apekatt",":snabeldyr:"],"language":"nb"}]}"""
+    val converted        = migration.convertColumn(oldDocument)
+    val expectedDocument = """{"tags":[{"tags":["tag1","apekatt","snabeldyr"],"language":"nb"}]}"""
+    converted should be(expectedDocument)
+  }
+
+  test("That non colon separated tags are not changed") {
+    val oldDocument      = """{"tags":[{"tags":["tag1","apekatt","snabeldyr"],"language":"nb"}]}"""
+    val converted        = migration.convertColumn(oldDocument)
+    val expectedDocument = """{"tags":[{"tags":["tag1","apekatt","snabeldyr"],"language":"nb"}]}"""
+    converted should be(expectedDocument)
+  }
+
+  test("That no tags doesn't crash") {
+    val oldDocument = """{"someotherfield":"yabadabado"}"""
+    val converted   = migration.convertColumn(oldDocument)
+    converted should be(oldDocument)
+  }
+}