|
| 1 | +/* SPDX-License-Identifier: Apache-2.0 |
| 2 | + Copyright 2023 Atlan Pte. Ltd. */ |
| 3 | +package com.atlan.pkg.rab |
| 4 | + |
| 5 | +import AssetImportCfg |
| 6 | +import RelationalAssetsBuilderCfg |
| 7 | +import com.atlan.model.assets.Asset |
| 8 | +import com.atlan.model.assets.Column |
| 9 | +import com.atlan.model.assets.Connection |
| 10 | +import com.atlan.model.assets.Table |
| 11 | +import com.atlan.model.enums.AtlanConnectorType |
| 12 | +import com.atlan.model.fields.AtlanField |
| 13 | +import com.atlan.pkg.PackageTest |
| 14 | +import com.atlan.pkg.Utils |
| 15 | +import com.atlan.pkg.rab.Importer.PREVIOUS_FILES_PREFIX |
| 16 | +import org.testng.Assert.assertFalse |
| 17 | +import org.testng.Assert.assertTrue |
| 18 | +import java.io.File |
| 19 | +import java.nio.file.Paths |
| 20 | +import kotlin.test.Test |
| 21 | +import kotlin.test.assertEquals |
| 22 | +import kotlin.test.assertNotNull |
| 23 | + |
| 24 | +/** |
| 25 | + * Test that RAB correctly expands STRUCT/ARRAY/MAP complex-type columns into nested child columns. |
| 26 | + */ |
| 27 | +class ComplexTypeColumnsRABTest : PackageTest("ctcol") { |
| 28 | + override val logger = Utils.getLogger(this.javaClass.name) |
| 29 | + |
| 30 | + private val conn1 = makeUnique("c1") |
| 31 | + private val conn1Type = AtlanConnectorType.ICEBERG |
| 32 | + |
| 33 | + private val testFile = "input.csv" |
| 34 | + private val files = listOf(testFile, "debug.log") |
| 35 | + |
| 36 | + private fun prepFile() { |
| 37 | + val input = Paths.get("src", "test", "resources", "assets-complex.csv").toFile() |
| 38 | + val output = Paths.get(testDirectory, testFile).toFile() |
| 39 | + input.useLines { lines -> |
| 40 | + lines.forEach { line -> |
| 41 | + val revised = |
| 42 | + line |
| 43 | + .replace("{{CONNECTION1}}", conn1) |
| 44 | + .replace("{{API_TOKEN_USER}}", client.users.currentUser.username) |
| 45 | + output.appendText("$revised\n") |
| 46 | + } |
| 47 | + } |
| 48 | + } |
| 49 | + |
| 50 | + private val connectionAttrs: List<AtlanField> = |
| 51 | + listOf( |
| 52 | + Connection.NAME, |
| 53 | + Connection.CONNECTOR_TYPE, |
| 54 | + ) |
| 55 | + |
| 56 | + private val tableAttrs: List<AtlanField> = |
| 57 | + listOf( |
| 58 | + Table.NAME, |
| 59 | + Table.CONNECTION_QUALIFIED_NAME, |
| 60 | + Table.COLUMN_COUNT, |
| 61 | + Table.COLUMNS, |
| 62 | + ) |
| 63 | + |
| 64 | + private val columnAttrs: List<AtlanField> = |
| 65 | + listOf( |
| 66 | + Column.NAME, |
| 67 | + Column.STATUS, |
| 68 | + Column.CONNECTION_QUALIFIED_NAME, |
| 69 | + Column.TABLE_NAME, |
| 70 | + Column.TABLE_QUALIFIED_NAME, |
| 71 | + Column.DATA_TYPE, |
| 72 | + Column.RAW_DATA_TYPE_DEFINITION, |
| 73 | + Column.ORDER, |
| 74 | + Column.PARENT_COLUMN_QUALIFIED_NAME, |
| 75 | + Column.PARENT_COLUMN_NAME, |
| 76 | + Column.COLUMN_DEPTH_LEVEL, |
| 77 | + Column.COLUMN_HIERARCHY, |
| 78 | + Column.NESTED_COLUMN_ORDER, |
| 79 | + ) |
| 80 | + |
| 81 | + override fun setup() { |
| 82 | + prepFile() |
| 83 | + runCustomPackage( |
| 84 | + RelationalAssetsBuilderCfg( |
| 85 | + assetsFile = Paths.get(testDirectory, testFile).toString(), |
| 86 | + assetsUpsertSemantic = "upsert", |
| 87 | + assetsFailOnErrors = true, |
| 88 | + deltaSemantic = "full", |
| 89 | + ), |
| 90 | + Importer::main, |
| 91 | + ) |
| 92 | + runCustomPackage( |
| 93 | + AssetImportCfg( |
| 94 | + assetsFile = "$testDirectory${File.separator}current-file-transformed.csv", |
| 95 | + assetsUpsertSemantic = "upsert", |
| 96 | + assetsDeltaSemantic = "full", |
| 97 | + assetsFailOnErrors = true, |
| 98 | + assetsPreviousFilePrefix = PREVIOUS_FILES_PREFIX, |
| 99 | + ), |
| 100 | + com.atlan.pkg.aim.Importer::main, |
| 101 | + ) |
| 102 | + } |
| 103 | + |
| 104 | + override fun teardown() { |
| 105 | + removeConnection(conn1, conn1Type) |
| 106 | + } |
| 107 | + |
| 108 | + @Test(groups = ["rab.ctcol.create"]) |
| 109 | + fun tableHasOnlyTopLevelColumns() { |
| 110 | + // The table_columns relationship must only contain the 5 top-level columns; |
| 111 | + // nested child columns have their table reference cleared by ColumnXformer. |
| 112 | + val c1 = Connection.findByName(client, conn1, conn1Type, connectionAttrs)[0]!! |
| 113 | + val request = |
| 114 | + Table |
| 115 | + .select(client) |
| 116 | + .where(Table.CONNECTION_QUALIFIED_NAME.eq(c1.qualifiedName)) |
| 117 | + .includesOnResults(tableAttrs) |
| 118 | + .includeOnRelations(Asset.NAME) |
| 119 | + .toRequest() |
| 120 | + val response = retrySearchUntil(request, 1) |
| 121 | + val tbl = response.assets[0] as Table |
| 122 | + assertEquals("TEST_TBL", tbl.name) |
| 123 | + assertEquals(5, tbl.columnCount) |
| 124 | + val topLevelNames = tbl.columns.map { it.name }.toSet() |
| 125 | + assertEquals(setOf("PLAIN_COL", "STRUCT_COL", "ARRAY_COL", "MAP_COL", "NESTED_STRUCT_COL"), topLevelNames) |
| 126 | + } |
| 127 | + |
| 128 | + @Test(groups = ["rab.ctcol.create"]) |
| 129 | + fun structColChildColumns() { |
| 130 | + // STRUCT<city:STRING, zip:INT> → two direct children at depth 1 |
| 131 | + val c1 = Connection.findByName(client, conn1, conn1Type, connectionAttrs)[0]!! |
| 132 | + val structColQN = "${c1.qualifiedName}/TEST_DB/TEST_SCHEMA/TEST_TBL/STRUCT_COL" |
| 133 | + val request = |
| 134 | + Column |
| 135 | + .select(client) |
| 136 | + .where(Column.PARENT_COLUMN_QUALIFIED_NAME.eq(structColQN)) |
| 137 | + .includesOnResults(columnAttrs) |
| 138 | + .toRequest() |
| 139 | + val response = retrySearchUntil(request, 2) |
| 140 | + val found = response.assets |
| 141 | + assertEquals(2, found.size) |
| 142 | + assertEquals(setOf("city", "zip"), found.map { it.name }.toSet()) |
| 143 | + found.forEach { asset -> |
| 144 | + val col = asset as Column |
| 145 | + assertEquals(structColQN, col.parentColumnQualifiedName) |
| 146 | + assertEquals("STRUCT_COL", col.parentColumnName) |
| 147 | + assertEquals(1, col.columnDepthLevel) |
| 148 | + assertNotNull(col.columnHierarchy) |
| 149 | + assertFalse(col.columnHierarchy.isEmpty()) |
| 150 | + assertTrue(col.tableName.isNullOrEmpty()) |
| 151 | + assertTrue(col.tableQualifiedName.isNullOrEmpty()) |
| 152 | + when (col.name) { |
| 153 | + "city" -> { |
| 154 | + assertEquals("STRING", col.dataType) |
| 155 | + assertEquals("1", col.nestedColumnOrder) |
| 156 | + } |
| 157 | + |
| 158 | + "zip" -> { |
| 159 | + assertEquals("INT", col.dataType) |
| 160 | + assertEquals("2", col.nestedColumnOrder) |
| 161 | + } |
| 162 | + } |
| 163 | + } |
| 164 | + } |
| 165 | + |
| 166 | + @Test(groups = ["rab.ctcol.create"]) |
| 167 | + fun arrayColChildColumns() { |
| 168 | + // ARRAY<STRUCT<id:INT, name:STRING>> → two children whose QN passes through /items/ synthetic node |
| 169 | + // but whose parentColumnQualifiedName points directly at ARRAY_COL |
| 170 | + val c1 = Connection.findByName(client, conn1, conn1Type, connectionAttrs)[0]!! |
| 171 | + val arrayColQN = "${c1.qualifiedName}/TEST_DB/TEST_SCHEMA/TEST_TBL/ARRAY_COL" |
| 172 | + val request = |
| 173 | + Column |
| 174 | + .select(client) |
| 175 | + .where(Column.PARENT_COLUMN_QUALIFIED_NAME.eq(arrayColQN)) |
| 176 | + .includesOnResults(columnAttrs) |
| 177 | + .toRequest() |
| 178 | + val response = retrySearchUntil(request, 2) |
| 179 | + val found = response.assets |
| 180 | + assertEquals(2, found.size) |
| 181 | + assertEquals(setOf("id", "name"), found.map { it.name }.toSet()) |
| 182 | + found.forEach { asset -> |
| 183 | + val col = asset as Column |
| 184 | + assertEquals(arrayColQN, col.parentColumnQualifiedName) |
| 185 | + assertEquals(1, col.columnDepthLevel) |
| 186 | + // QN must route through the synthetic "items" node |
| 187 | + assertTrue(col.qualifiedName.contains("/items/"), "Expected /items/ in QN: ${col.qualifiedName}") |
| 188 | + assertTrue(col.tableName.isNullOrEmpty()) |
| 189 | + } |
| 190 | + } |
| 191 | + |
| 192 | + @Test(groups = ["rab.ctcol.create"]) |
| 193 | + fun mapColChildColumns() { |
| 194 | + // MAP<STRING, STRUCT<key:STRING, value:DOUBLE>> → two children whose QN passes through /values/ synthetic node |
| 195 | + val c1 = Connection.findByName(client, conn1, conn1Type, connectionAttrs)[0]!! |
| 196 | + val mapColQN = "${c1.qualifiedName}/TEST_DB/TEST_SCHEMA/TEST_TBL/MAP_COL" |
| 197 | + val request = |
| 198 | + Column |
| 199 | + .select(client) |
| 200 | + .where(Column.PARENT_COLUMN_QUALIFIED_NAME.eq(mapColQN)) |
| 201 | + .includesOnResults(columnAttrs) |
| 202 | + .toRequest() |
| 203 | + val response = retrySearchUntil(request, 2) |
| 204 | + val found = response.assets |
| 205 | + assertEquals(2, found.size) |
| 206 | + assertEquals(setOf("key", "value"), found.map { it.name }.toSet()) |
| 207 | + found.forEach { asset -> |
| 208 | + val col = asset as Column |
| 209 | + assertEquals(mapColQN, col.parentColumnQualifiedName) |
| 210 | + assertEquals(1, col.columnDepthLevel) |
| 211 | + // QN must route through the synthetic "values" node |
| 212 | + assertTrue(col.qualifiedName.contains("/values/"), "Expected /values/ in QN: ${col.qualifiedName}") |
| 213 | + assertTrue(col.tableName.isNullOrEmpty()) |
| 214 | + } |
| 215 | + } |
| 216 | + |
| 217 | + @Test(groups = ["rab.ctcol.create"]) |
| 218 | + fun nestedStructColDepth1Children() { |
| 219 | + // STRUCT<outer:STRUCT<...>, label:STRING> → two depth-1 children directly under NESTED_STRUCT_COL |
| 220 | + val c1 = Connection.findByName(client, conn1, conn1Type, connectionAttrs)[0]!! |
| 221 | + val nestedColQN = "${c1.qualifiedName}/TEST_DB/TEST_SCHEMA/TEST_TBL/NESTED_STRUCT_COL" |
| 222 | + val request = |
| 223 | + Column |
| 224 | + .select(client) |
| 225 | + .where(Column.PARENT_COLUMN_QUALIFIED_NAME.eq(nestedColQN)) |
| 226 | + .includesOnResults(columnAttrs) |
| 227 | + .toRequest() |
| 228 | + val response = retrySearchUntil(request, 2) |
| 229 | + val found = response.assets |
| 230 | + assertEquals(2, found.size) |
| 231 | + assertEquals(setOf("outer", "label"), found.map { it.name }.toSet()) |
| 232 | + found.forEach { asset -> |
| 233 | + val col = asset as Column |
| 234 | + assertEquals(nestedColQN, col.parentColumnQualifiedName) |
| 235 | + assertEquals(1, col.columnDepthLevel) |
| 236 | + assertTrue(col.tableName.isNullOrEmpty()) |
| 237 | + when (col.name) { |
| 238 | + "outer" -> assertEquals("STRUCT", col.dataType) |
| 239 | + "label" -> assertEquals("STRING", col.dataType) |
| 240 | + } |
| 241 | + } |
| 242 | + } |
| 243 | + |
| 244 | + @Test(groups = ["rab.ctcol.create"]) |
| 245 | + fun nestedStructColDepth2Children() { |
| 246 | + // The STRUCT-typed "outer" field expands to inner:STRING, count:INT at depth 2 |
| 247 | + val c1 = Connection.findByName(client, conn1, conn1Type, connectionAttrs)[0]!! |
| 248 | + val outerColQN = "${c1.qualifiedName}/TEST_DB/TEST_SCHEMA/TEST_TBL/NESTED_STRUCT_COL/outer" |
| 249 | + val request = |
| 250 | + Column |
| 251 | + .select(client) |
| 252 | + .where(Column.PARENT_COLUMN_QUALIFIED_NAME.eq(outerColQN)) |
| 253 | + .includesOnResults(columnAttrs) |
| 254 | + .toRequest() |
| 255 | + val response = retrySearchUntil(request, 2) |
| 256 | + val found = response.assets |
| 257 | + assertEquals(2, found.size) |
| 258 | + assertEquals(setOf("inner", "count"), found.map { it.name }.toSet()) |
| 259 | + found.forEach { asset -> |
| 260 | + val col = asset as Column |
| 261 | + assertEquals(outerColQN, col.parentColumnQualifiedName) |
| 262 | + assertEquals("outer", col.parentColumnName) |
| 263 | + assertEquals(2, col.columnDepthLevel) |
| 264 | + // Hierarchy must list both NESTED_STRUCT_COL (depth 1) and outer (depth 2) |
| 265 | + assertNotNull(col.columnHierarchy) |
| 266 | + assertEquals(2, col.columnHierarchy.size) |
| 267 | + assertTrue(col.tableName.isNullOrEmpty()) |
| 268 | + } |
| 269 | + } |
| 270 | + |
| 271 | + @Test(dependsOnGroups = ["rab.ctcol.create"]) |
| 272 | + fun filesCreated() { |
| 273 | + validateFilesExist(files) |
| 274 | + } |
| 275 | + |
| 276 | + @Test(dependsOnGroups = ["rab.ctcol.create"]) |
| 277 | + fun errorFreeLog() { |
| 278 | + validateErrorFreeLog() |
| 279 | + } |
| 280 | +} |
0 commit comments