|
3 | 3 | namespace Keboola\Test\Backend\Snowflake; |
4 | 4 |
|
5 | 5 | use Keboola\Datatype\Definition\Snowflake; |
6 | | -use Keboola\StorageApi\ClientException; |
| 6 | +use Keboola\StorageApi\Options\TableImport\DeduplicationStrategy; |
7 | 7 | use Keboola\StorageApi\Workspaces; |
8 | 8 | use Keboola\TableBackendUtils\Column\ColumnCollection; |
9 | 9 | use Keboola\TableBackendUtils\Column\ColumnInterface; |
@@ -109,7 +109,249 @@ public function testUnloadFromWSToTypedTable(): void |
109 | 109 | $this->unloadAndAssert($workspace['id'], $tableId); |
110 | 110 | } |
111 | 111 |
|
112 | | - public function testUnloadFromWSToTypedTableCTAS(): void |
| 112 | + public function testUnloadFromWSToTypedTableWithoutDeduplication(): void |
| 113 | + { |
| 114 | + // create workspace and source table in workspace |
| 115 | + $workspace = $this->initTestWorkspace(); |
| 116 | + |
| 117 | + $tableId = 'Languages3'; |
| 118 | + |
| 119 | + $backend = WorkspaceBackendFactory::createWorkspaceBackend($workspace); |
| 120 | + $backend->dropTableIfExists($tableId); |
| 121 | + |
| 122 | + $connection = $workspace['connection']; |
| 123 | + |
| 124 | + $db = $backend->getDb(); |
| 125 | + |
| 126 | + $quotedTableId = sprintf( |
| 127 | + '"%s"."%s"', |
| 128 | + $connection['schema'], |
| 129 | + $tableId, |
| 130 | + ); |
| 131 | + // exact length is required |
| 132 | + $sql = sprintf( |
| 133 | + ' |
| 134 | + CREATE TABLE %s ( |
| 135 | + "id" INT NOT NULL, |
| 136 | + "name" VARCHAR(16777216), |
| 137 | + PRIMARY KEY ("id") |
| 138 | + ) |
| 139 | + ', |
| 140 | + $quotedTableId, |
| 141 | + ); |
| 142 | + $db->query($sql); |
| 143 | + $db->query(sprintf("INSERT INTO %s VALUES (1, 'john');", $quotedTableId)); |
| 144 | + $db->query(sprintf("INSERT INTO %s VALUES (2, 'does');", $quotedTableId)); |
| 145 | + |
| 146 | + $this->unloadAndAssert($workspace['id'], $tableId); |
| 147 | + |
| 148 | + // test unload from workspace with _timestamp column exist |
| 149 | + $db->query(sprintf('DROP TABLE %s', $quotedTableId)); |
| 150 | + $sql = sprintf( |
| 151 | + ' |
| 152 | + CREATE TABLE %s ( |
| 153 | + "id" INT NOT NULL, |
| 154 | + "name" VARCHAR(16777216), |
| 155 | + PRIMARY KEY ("id") |
| 156 | + ) |
| 157 | + ', |
| 158 | + $quotedTableId, |
| 159 | + ); |
| 160 | + $db->query($sql); |
| 161 | + $db->query(sprintf("INSERT INTO %s VALUES (1, 'john');", $quotedTableId)); |
| 162 | + $db->query(sprintf("INSERT INTO %s VALUES (2, 'does');", $quotedTableId)); |
| 163 | + // snowflake does not enforce primary key so this is allowed |
| 164 | + // normally such value would be deduplicated in storage |
| 165 | + // but ctas-om does not deduplicate values |
| 166 | + $db->query(sprintf("INSERT INTO %s VALUES (2, 'doesToo');", $quotedTableId)); |
| 167 | + |
| 168 | + $runId = $this->_client->generateRunId(); |
| 169 | + $this->_client->setRunId($runId); |
| 170 | + $this->initEvents($this->_client); |
| 171 | + |
| 172 | + // do full load |
| 173 | + $this->_client->writeTableAsyncDirect($this->tableId, [ |
| 174 | + 'dataWorkspaceId' => $workspace['id'], |
| 175 | + 'dataTableName' => $tableId, |
| 176 | + 'deduplicationStrategy' => DeduplicationStrategy::INSERT, |
| 177 | + ]); |
| 178 | + $eventAssertCallback = function ($events) { |
| 179 | + $this->assertCount(1, $events); |
| 180 | + $this->assertArrayHasKey('performance', $events[0]); |
| 181 | + $this->assertIsArray($events[0]['performance']); |
| 182 | + $this->assertArrayHasKey('importDecomposed', $events[0]['performance']); |
| 183 | + $this->assertCount(1, $events[0]['performance']['importDecomposed']); |
| 184 | + $this->assertArrayHasKey('name', $events[0]['performance']['importDecomposed'][0]); |
| 185 | + // there is only one stage in full load |
| 186 | + $this->assertSame('ctasLoad', $events[0]['performance']['importDecomposed'][0]['name']); |
| 187 | + |
| 188 | + // set event as last event |
| 189 | + $this->lastEventId = $events[0]['uuid']; |
| 190 | + }; |
| 191 | + $query = new EventsQueryBuilder(); |
| 192 | + $query->setEvent('storage.tableImportDone') |
| 193 | + ->setTokenId($this->tokenId) |
| 194 | + ->setRunId($runId); |
| 195 | + $this->assertEventWithRetries($this->_client, $eventAssertCallback, $query); |
| 196 | + |
| 197 | + $expectedFullLoad = [ |
| 198 | + [ |
| 199 | + [ |
| 200 | + 'columnName' => 'id', |
| 201 | + 'value' => '1', |
| 202 | + 'isTruncated' => false, |
| 203 | + ], |
| 204 | + [ |
| 205 | + 'columnName' => 'name', |
| 206 | + 'value' => 'john', |
| 207 | + 'isTruncated' => false, |
| 208 | + ], |
| 209 | + ], |
| 210 | + [ |
| 211 | + [ |
| 212 | + 'columnName' => 'id', |
| 213 | + 'value' => '2', |
| 214 | + 'isTruncated' => false, |
| 215 | + ], |
| 216 | + [ |
| 217 | + 'columnName' => 'name', |
| 218 | + 'value' => 'does', |
| 219 | + 'isTruncated' => false, |
| 220 | + ], |
| 221 | + ], |
| 222 | + [ |
| 223 | + [ |
| 224 | + 'columnName' => 'id', |
| 225 | + 'value' => '2', |
| 226 | + 'isTruncated' => false, |
| 227 | + ], |
| 228 | + [ |
| 229 | + 'columnName' => 'name', |
| 230 | + 'value' => 'doesToo', |
| 231 | + 'isTruncated' => false, |
| 232 | + ], |
| 233 | + ], |
| 234 | + ]; |
| 235 | + |
| 236 | + /** @var array $data */ |
| 237 | + $data = $this->_client->getTableDataPreview($this->tableId, [ |
| 238 | + 'format' => 'json', |
| 239 | + ]); |
| 240 | + self::assertEquals($expectedFullLoad, $data['rows']); |
| 241 | + |
| 242 | + // do incremental load |
| 243 | + // incremental load also does not deduplicate values |
| 244 | + // this will cause that all values are present twice |
| 245 | + $this->_client->writeTableAsyncDirect($this->tableId, [ |
| 246 | + 'dataWorkspaceId' => $workspace['id'], |
| 247 | + 'dataTableName' => $tableId, |
| 248 | + 'incremental' => true, |
| 249 | + 'deduplicationStrategy' => DeduplicationStrategy::INSERT->value, |
| 250 | + ]); |
| 251 | + |
| 252 | + $eventAssertCallback = function ($events) { |
| 253 | + $this->assertCount(1, $events); |
| 254 | + $this->assertArrayHasKey('performance', $events[0]); |
| 255 | + $this->assertIsArray($events[0]['performance']); |
| 256 | + $this->assertArrayHasKey('importDecomposed', $events[0]['performance']); |
| 257 | + $this->assertCount(1, $events[0]['performance']['importDecomposed']); |
| 258 | + $this->assertArrayHasKey('name', $events[0]['performance']['importDecomposed'][0]); |
| 259 | + // there is only one stage in incremental load |
| 260 | + $this->assertSame('insertIntoTargetFromStaging', $events[0]['performance']['importDecomposed'][0]['name']); |
| 261 | + }; |
| 262 | + $query = new EventsQueryBuilder(); |
| 263 | + $query->setEvent('storage.tableImportDone') |
| 264 | + ->setTokenId($this->tokenId) |
| 265 | + ->setRunId($runId); |
| 266 | + $this->assertEventWithRetries($this->_client, $eventAssertCallback, $query); |
| 267 | + |
| 268 | + $expectedIncrementalLoad = [ |
| 269 | + [ |
| 270 | + [ |
| 271 | + 'columnName' => 'id', |
| 272 | + 'value' => '1', |
| 273 | + 'isTruncated' => false, |
| 274 | + ], |
| 275 | + [ |
| 276 | + 'columnName' => 'name', |
| 277 | + 'value' => 'john', |
| 278 | + 'isTruncated' => false, |
| 279 | + ], |
| 280 | + ], |
| 281 | + [ |
| 282 | + [ |
| 283 | + 'columnName' => 'id', |
| 284 | + 'value' => '2', |
| 285 | + 'isTruncated' => false, |
| 286 | + ], |
| 287 | + [ |
| 288 | + 'columnName' => 'name', |
| 289 | + 'value' => 'does', |
| 290 | + 'isTruncated' => false, |
| 291 | + ], |
| 292 | + ], |
| 293 | + [ |
| 294 | + [ |
| 295 | + 'columnName' => 'id', |
| 296 | + 'value' => '2', |
| 297 | + 'isTruncated' => false, |
| 298 | + ], |
| 299 | + [ |
| 300 | + 'columnName' => 'name', |
| 301 | + 'value' => 'doesToo', |
| 302 | + 'isTruncated' => false, |
| 303 | + ], |
| 304 | + ], |
| 305 | + [ |
| 306 | + [ |
| 307 | + 'columnName' => 'id', |
| 308 | + 'value' => '1', |
| 309 | + 'isTruncated' => false, |
| 310 | + ], |
| 311 | + [ |
| 312 | + 'columnName' => 'name', |
| 313 | + 'value' => 'john', |
| 314 | + 'isTruncated' => false, |
| 315 | + ], |
| 316 | + ], |
| 317 | + [ |
| 318 | + [ |
| 319 | + 'columnName' => 'id', |
| 320 | + 'value' => '2', |
| 321 | + 'isTruncated' => false, |
| 322 | + ], |
| 323 | + [ |
| 324 | + 'columnName' => 'name', |
| 325 | + 'value' => 'does', |
| 326 | + 'isTruncated' => false, |
| 327 | + ], |
| 328 | + ], |
| 329 | + [ |
| 330 | + [ |
| 331 | + 'columnName' => 'id', |
| 332 | + 'value' => '2', |
| 333 | + 'isTruncated' => false, |
| 334 | + ], |
| 335 | + [ |
| 336 | + 'columnName' => 'name', |
| 337 | + 'value' => 'doesToo', |
| 338 | + 'isTruncated' => false, |
| 339 | + ], |
| 340 | + ], |
| 341 | + ]; |
| 342 | + |
| 343 | + /** @var array $data */ |
| 344 | + $data = $this->_client->getTableDataPreview($this->tableId, [ |
| 345 | + 'format' => 'json', |
| 346 | + ]); |
| 347 | + self::assertEquals($expectedIncrementalLoad, $data['rows']); |
| 348 | + } |
| 349 | + |
| 350 | + /** |
| 351 | + * "Copy" of until there is a "ctas-om" feature. |
| 352 | + * @see TypedTableInWorkspaceTest::testUnloadFromWSToTypedTableWithoutDeduplication |
| 353 | + */ |
| 354 | + public function testUnloadFromWSToTypedTableWithoutDeduplicationByFeature(): void |
113 | 355 | { |
114 | 356 | if (!in_array('ctas-om', $this->_client->verifyToken()['owner']['features'], true)) { |
115 | 357 | $this->markTestSkipped( |
@@ -348,16 +590,6 @@ public function testUnloadFromWSToTypedTableCTAS(): void |
348 | 590 | 'format' => 'json', |
349 | 591 | ]); |
350 | 592 | self::assertEquals($expectedIncrementalLoad, $data['rows']); |
351 | | - |
352 | | - $this->expectException(ClientException::class); |
353 | | - // create table does not support typed tables |
354 | | - // tables are created by runner and this is never called |
355 | | - $this->expectExceptionMessage('Table import error: Source destination columns mismatch. "id NUMBER (38,0) NOT NULL"->"id VARCHAR NOT NULL DEFAULT \'\''); |
356 | | - $this->_client->createTableAsyncDirect($this->getTestBucketId(self::STAGE_IN), [ |
357 | | - 'name' => 'languagesNew', |
358 | | - 'dataWorkspaceId' => $workspace['id'], |
359 | | - 'dataObject' => $tableId, |
360 | | - ]); |
361 | 593 | } |
362 | 594 |
|
363 | 595 | private function unloadAndAssert(int $id, string $tableId): void |
|
0 commit comments