Method computeBulkStringStats

pj_datastore/src/chunk.cpp:377–413

}

void TopicChunkBuilder::computeBulkStringStats(std::size_t col_index, std::size_t first_row, std::size_t count) {
  if (count == 0) {
    return;
  }

  auto& cs = stats_.column_stats[col_index];
  const auto& col = columns_[col_index];
  const bool has_validity = col.hasNulls();

  std::optional<std::string_view> last_valid;

  if (first_row > 0 && cs.run_count > 0) {
    for (std::size_t j = first_row; j > 0; --j) {
      if (!has_validity || col.isValid(j - 1)) {
        last_valid = col.readString(j - 1);
        break;
      }
    }
  }

  for (std::size_t i = 0; i < count; ++i) {
    const std::size_t row = first_row + i;
    if (has_validity && !col.isValid(row)) {
      cs.null_count++;
      continue;
    }

    std::string_view current = col.readString(row);
    if (!last_valid.has_value()) {
      cs.run_count = 1;
    } else if (current != *last_valid) {
      cs.is_constant = false;
      cs.run_count++;
    }
    last_valid = current;
  }
}

bool TopicChunkBuilder::isFull() const noexcept {
  return rowCount() >= max_rows_;