Commits
rjzamora authored and Wes McKinney committed 997226a9263
ARROW-1983: [C++][Parquet] Add AppendRowGroups and WriteMetaDataFile methods @wesm @jorisvandenbossche - This is a rough RFC for one possible solution to the write phase of [ARROW-1983](https://issues.apache.org/jira/browse/ARROW-1983). Summary: I added new `AppendRowGroups` and `WriteMetaDataFile` methods to the `FileMetData` class (along with python wrappers). I also added a `get_row_group` method to `FileMetaData::FileMetaDataImpl` to make the append function easier to write (not sure if this is really necessary). Feel free to recommend changes and/or complete rewrites. I realize that an ideal solution might be to modify the existing `ParquetWriter`/`write_metadata` funtionality to accept an existing `FileMetaData` object as input, so I will wait to add tests until I know this approach seems reasonable. Author: rjzamora <rzamora@nvidia.com> Closes #4405 from rjzamora/metadata-append and squashes the following commits: b4ffc9feb <rjzamora> fixing clang-format style issues 5999609cd <rjzamora> improve pytesting and remove use of shared_ptr in API 4085e0efa <rjzamora> adding WriteMetaDataFile to FileMetaData API 1973b61ee <rjzamora> easy code-review fixes (typos and renaming) 3cc44231d <rjzamora> more pytest linting 873ba25fb <rjzamora> fixing a few failed checks b3dd4ddb1 <rjzamora> sync with upstream master 9a01fcbbe <rjzamora> testing read_metadata and lint d0f86298e <rjzamora> using write_table in test 695e9a56c <rjzamora> fixing some linting and warnings 2544da17e <rjzamora> add python test for multi-dataset metadata b7eb85977 <rjzamora> adding test for AppendRowGroups 7a4c76cfb <rjzamora> using existing WriteFileMetaData function instead of new WriteMetaDataFile ffd343530 <rjzamora> minor cleanup 1486f8da0 <rjzamora> add WriteMetaDataFile function 54f3ea00a <rjzamora> adding method to append row-group metadata to an existing FileMetadata object