Commits
Wes McKinney authored 362e754b395
ARROW-1103: [Python] Support read_pandas (with index metadata) on directory of Parquet files Also fixes ARROW-1041, a case where the `_metadata` file contains the pandas schema metadata but the individual dataset fragments do not. Author: Wes McKinney <wes.mckinney@twosigma.com> Closes #862 from wesm/ARROW-1103 and squashes the following commits: 3f309166 [Wes McKinney] Add test for esoteric case where _metadata has pandas metadata but the individual Parquet dataset pieces don't 5985fc13 [Wes McKinney] Add experimental replace_schema_metadata functions, get basic read_pandas with metadata working on a multi-file dataset b362d60c [Wes McKinney] Initial refactor to support common metadata, read_pandas on a dataset