Function ConstructFakeMetaData

cpp/src/parquet/page_index_test.cc:274–314

Creates an FileMetaData object w/ single row group based on data in 'row_group_ranges'. It sets the offsets and sizes of the column index and offset index members of the row group. It doesn't set the member if the input value is -1.

/// 'row_group_ranges'. It sets the offsets and sizes of the column index and offset index
/// members of the row group. It doesn't set the member if the input value is -1.
std::shared_ptr<FileMetaData> ConstructFakeMetaData(
    const RowGroupRanges& row_group_ranges) {
  format::RowGroup row_group;
  for (auto& page_index_ranges : row_group_ranges) {
    format::ColumnChunk col_chunk;
    if (page_index_ranges.column_index_offset != -1) {
      col_chunk.__set_column_index_offset(page_index_ranges.column_index_offset);
    }
    if (page_index_ranges.column_index_length != -1) {
      col_chunk.__set_column_index_length(
          static_cast<int32_t>(page_index_ranges.column_index_length));
    }
    if (page_index_ranges.offset_index_offset != -1) {
      col_chunk.__set_offset_index_offset(page_index_ranges.offset_index_offset);
    }
    if (page_index_ranges.offset_index_length != -1) {
      col_chunk.__set_offset_index_length(
          static_cast<int32_t>(page_index_ranges.offset_index_length));
    }
    row_group.columns.push_back(col_chunk);
  }

  format::FileMetaData metadata;
  metadata.row_groups.push_back(row_group);

  metadata.schema.emplace_back();
  schema::NodeVector fields;
  for (size_t i = 0; i < row_group_ranges.size(); ++i) {
    fields.push_back(schema::Int64(std::to_string(i)));
    metadata.schema.emplace_back();
    fields.back()->ToParquet(&metadata.schema.back());
  }
  schema::GroupNode::Make("schema", Repetition::REPEATED, fields)
      ->ToParquet(&metadata.schema.front());

  auto sink = CreateOutputStream();
  ThriftSerializer{}.Serialize(&metadata, sink.get());
  auto buffer = sink->Finish().MoveValueUnsafe();
  uint32_t len = static_cast<uint32_t>(buffer->size());
  return FileMetaData::Make(buffer->data(), &len);
}

/// Validates that 'DeterminePageIndexRangesInRowGroup()' selects the expected file
/// offsets and sizes or returns false when the row group doesn't have a page index.