Function TEST_P

bolt/exec/tests/TableWriteTest.cpp:1650–1702

Runs a pipeline with read + filter + project (with substr) + write.

// Runs a pipeline with read + filter + project (with substr) + write.
TEST_P(AllTableWriterTest, scanFilterProjectWrite) {
  auto filePaths = makeFilePaths(5);
  auto vectors = makeVectors(filePaths.size(), 500);
  for (int i = 0; i < filePaths.size(); i++) {
    writeToFile(filePaths[i]->path, vectors[i]);
  }

  createDuckDbTable(vectors);

  auto outputDirectory = TempDirectoryPath::create();

  auto planBuilder = PlanBuilder();
  auto project = planBuilder.tableScan(rowType_).filter("c2 <> 0").project(
      {"c0", "c1", "c3", "c5", "c2 + c3", "substr(c5, 1, 1)"});

  auto intputTypes = project.planNode()->outputType()->children();
  std::vector<std::string> tableColumnNames = {
      "c0", "c1", "c3", "c5", "c2_plus_c3", "substr_c5"};
  const auto outputType =
      ROW(std::move(tableColumnNames), std::move(intputTypes));

  auto plan = createInsertPlan(
      project,
      outputType,
      outputDirectory->path,
      partitionedBy_,
      bucketProperty_,
      compressionKind_,
      getNumWriters(),
      connector::hive::LocationHandle::TableType::kNew,
      commitStrategy_);

  assertQueryWithWriterConfigs(
      plan, filePaths, "SELECT count(*) FROM tmp WHERE c2 <> 0");

  // To test the correctness of the generated output,
  // We create a new plan that only read that file and then
  // compare that against a duckDB query that runs the whole query.
  if (partitionedBy_.size() > 0) {
    auto newOutputType = getNonPartitionsColumns(partitionedBy_, outputType);
    assertQuery(
        PlanBuilder().tableScan(newOutputType).planNode(),
        makeHiveConnectorSplits(outputDirectory),
        "SELECT c3, c5, c2 + c3, substr(c5, 1, 1) FROM tmp WHERE c2 <> 0");
    verifyTableWriterOutput(outputDirectory->path, newOutputType, false);
  } else {
    assertQuery(
        PlanBuilder().tableScan(outputType).planNode(),
        makeHiveConnectorSplits(outputDirectory),
        "SELECT c0, c1, c3, c5, c2 + c3, substr(c5, 1, 1) FROM tmp WHERE c2 <> 0");
    verifyTableWriterOutput(outputDirectory->path, outputType, false);
  }
}

TEST_P(AllTableWriterTest, renameAndReorderColumns) {
  auto filePaths = makeFilePaths(5);