Function printStatistics

tools/src/FileStatistics.cc:27–63

#include <string>

void printStatistics(const char* filename, bool withIndex) {
  orc::ReaderOptions opts;
  std::unique_ptr<orc::Reader> reader;
  reader = orc::createReader(orc::readFile(std::string(filename), opts.getReaderMetrics()), opts);
  // print out all selected columns statistics.
  std::unique_ptr<orc::Statistics> colStats = reader->getStatistics();
  std::cout << "File " << filename << " has " << colStats->getNumberOfColumns() << " columns"
            << std::endl;
  for (uint32_t i = 0; i < colStats->getNumberOfColumns(); ++i) {
    std::cout << "*** Column " << i << " ***" << std::endl;
    std::cout << colStats->getColumnStatistics(i)->toString() << std::endl;
  }

  // test stripe statistics
  std::unique_ptr<orc::StripeStatistics> stripeStats;
  std::cout << "File " << filename << " has " << reader->getNumberOfStripes() << " stripes"
            << std::endl;
  if (reader->getNumberOfStripeStatistics() == 0) {
    std::cout << "File " << filename << " doesn't have stripe statistics" << std::endl;
  } else {
    for (unsigned int j = 0; j < reader->getNumberOfStripeStatistics(); j++) {
      stripeStats = reader->getStripeStatistics(j);
      std::cout << "*** Stripe " << j << " ***" << std::endl << std::endl;

      for (unsigned int k = 0; k < stripeStats->getNumberOfColumns(); ++k) {
        std::cout << "--- Column " << k << " ---" << std::endl;
        std::cout << stripeStats->getColumnStatistics(k)->toString() << std::endl;
        if (withIndex) {
          for (unsigned int r = 0; r < stripeStats->getNumberOfRowIndexStats(k); ++r) {
            std::cout << "--- RowIndex " << r << " ---" << std::endl;
            std::cout << stripeStats->getRowIndexStatistics(k, r)->toString() << std::endl;
          }
        }
      }
    }
  }
}

int main(int argc, char* argv[]) {
  static struct option longOptions[] = {{"help", no_argument, nullptr, 'h'},