Method getRowTypeForFile

bolt/connectors/hive/HiveDataSource.cpp:415–450

    return;
  }

  addSplit(std::dynamic_pointer_cast<HiveConnectorSplit>(split));
}

bolt::RowTypePtr HiveDataSource::getRowTypeForFile(
    std::shared_ptr<PaimonConnectorSplit> split) {
  std::vector<int> columnCacheBlackList;
  auto hiveSplit = split->hiveSplits[0];
  auto fileHandle =
      fileHandleFactory_->generate(hiveSplit->filePath, fsSessionConfig_)
          .second;

  dwio::common::ReaderOptions baseReaderOpts{connectorQueryCtx_->memoryPool()};

  hive::configureReaderOptions(
      baseReaderOpts,
      hiveConfig_,
      connectorQueryCtx_->sessionProperties(),
      hiveTableHandle_->dataColumns(),
      hiveSplit);

  auto& hiveConnectorSplitCacheLimit = hiveSplit->hiveConnectorSplitCacheLimit;
  auto baseFileInput = hive::createBufferedInput(
      *fileHandle,
      baseReaderOpts,
      connectorQueryCtx_.get(),
      ioStats_,
      executor_,
      false,
      columnCacheBlackList,
      nullptr);

  auto baseReader =
      dwio::common::getReaderFactory(baseReaderOpts.getFileFormat())
          ->createReader(std::move(baseFileInput), baseReaderOpts);

  auto& fileType = baseReader->rowType();

  return fileType;