GH-43598: [C++][Parquet] Parquet Metadata Printer supports print sort-columns

[mapleFU]

Rationale for this change

Now we have "sort-columns" support in Parquet spec, Python ( https://github.com/apache/arrow/pull/37665/files ) and C++. We can support print it in metadata Printer

What changes are included in this PR?

Add "SortingColumns" support in parquet printer

Are these changes tested?

• TODO after Add test file with sorting columns parquet-testing#56 is merged

Are there any user-facing changes?

No

• GitHub Issue: [C++][Parquet] Parquet Metadata Printer supports print sort-columns #43598

[github-actions]

⚠️ GitHub issue #43598 has been automatically assigned in GitHub to PR creator.

[mapleFU]

After this patch:

{
  "Version": "2.6",
  "CreatedBy": "parquet-cpp-arrow version 16.1.0",
  "TotalRows": "3",
  "NumberOfRowGroups": "1",
  "NumberOfRealColumns": "2",
  "NumberOfColumns": "2",
  "Columns": [
     { "Id": "0", "Name": "a", "PhysicalType": "INT64", "ConvertedType": "NONE", "LogicalType": {"Type": "None"} },
     { "Id": "1", "Name": "b", "PhysicalType": "BYTE_ARRAY", "ConvertedType": "UTF8", "LogicalType": {"Type": "String"} }
  ],
  "RowGroups": [
     {
       "Id": "0",  "TotalBytes": "166",  "TotalCompressedBytes": "174",  "SortColumns": [{"column_idx":0, "descending":1, "nulls_first": 1}, {"column_idx":1, "descending":0, "nulls_first": 0}],  "Rows": "3",
       "ColumnChunks": [
          {"Id": "0", "Values": "3", "StatsSet": "True", "Stats": {"NumNulls": "1", "Max": "2", "Min": "1" },
           "Compression": "SNAPPY", "Encodings": "PLAIN(DICT_PAGE) RLE_DICTIONARY", "UncompressedSize": "100", "CompressedSize": "104" },
          {"Id": "1", "Values": "3", "StatsSet": "True", "Stats": {"NumNulls": "0", "Max": "c", "Min": "a" },
           "Compression": "SNAPPY", "Encodings": "PLAIN(DICT_PAGE) RLE_DICTIONARY", "UncompressedSize": "66", "CompressedSize": "70" }
        ]
     }
  ]
}

[wgtmac]

Should we break a new line for each item in SortColumns? It currently looks a little bit lengthy to me.

[mapleFU]

Update the output:

{
  "FileName": "sort_columns.parquet",
  "Version": "2.6",
  "CreatedBy": "parquet-cpp-arrow version 16.1.0",
  "TotalRows": "6",
  "NumberOfRowGroups": "2",
  "NumberOfRealColumns": "2",
  "NumberOfColumns": "2",
  "Columns": [
     { "Id": "0", "Name": "a", "PhysicalType": "INT64", "ConvertedType": "NONE", "LogicalType": {"Type": "None"} },
     { "Id": "1", "Name": "b", "PhysicalType": "BYTE_ARRAY", "ConvertedType": "UTF8", "LogicalType": {"Type": "String"} }
  ],
  "RowGroups": [
     {
       "Id": "0",  "TotalBytes": "166",  "TotalCompressedBytes": "174",  "SortColumns": [
         {"column_idx":0, "descending":1, "nulls_first": 1},
         {"column_idx":1, "descending":0, "nulls_first": 0}
       ],  "Rows": "3",
       "ColumnChunks": [
          {"Id": "0", "Values": "3", "StatsSet": "True", "Stats": {"NumNulls": "1", "Max": "2", "Min": "1" },
           "Compression": "SNAPPY", "Encodings": "PLAIN(DICT_PAGE) RLE_DICTIONARY", "UncompressedSize": "100", "CompressedSize": "104" },
          {"Id": "1", "Values": "3", "StatsSet": "True", "Stats": {"NumNulls": "0", "Max": "c", "Min": "a" },
           "Compression": "SNAPPY", "Encodings": "PLAIN(DICT_PAGE) RLE_DICTIONARY", "UncompressedSize": "66", "CompressedSize": "70" }
        ]
     },
     {
       "Id": "1",  "TotalBytes": "166",  "TotalCompressedBytes": "174",  "SortColumns": [
         {"column_idx":0, "descending":1, "nulls_first": 1},
         {"column_idx":1, "descending":0, "nulls_first": 0}
       ],  "Rows": "3",
       "ColumnChunks": [
          {"Id": "0", "Values": "3", "StatsSet": "True", "Stats": {"NumNulls": "1", "Max": "2", "Min": "1" },
           "Compression": "SNAPPY", "Encodings": "PLAIN(DICT_PAGE) RLE_DICTIONARY", "UncompressedSize": "100", "CompressedSize": "104" },
          {"Id": "1", "Values": "3", "StatsSet": "True", "Stats": {"NumNulls": "0", "Max": "c", "Min": "a" },
           "Compression": "SNAPPY", "Encodings": "PLAIN(DICT_PAGE) RLE_DICTIONARY", "UncompressedSize": "66", "CompressedSize": "70" }
        ]
     }
  ]
}

[mapleFU]

@wgtmac @pitrou would you mind take a look?

[wgtmac]

+1

Thanks @mapleFU!

[pitrou]

LGTM, just 2 nits

[pitrou]

Nit

Suggested change

	auto sorting_columns = group_metadata->sorting_columns();
	const auto& sorting_columns = group_metadata->sorting_columns();

[mapleFU]

std::vector<SortingColumn> RowGroupMetaData::sorting_columns() const

Seems that this returns a vector directly, and just auto is ok here...

[pitrou]

Similar nit

Suggested change

	auto row_group_sorting_columns = group_metadata->sorting_columns();
	const auto& row_group_sorting_columns = group_metadata->sorting_columns();

[mapleFU]

ditto

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 3 benchmarking runs that have been run so far on merge-commit 03c3f8e.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details. It also includes information about 32 possible false positives for unstable benchmarks that are known to sometimes produce them.