[
https://issues.apache.org/jira/browse/ARROW-8231?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17068053#comment-17068053
]
Wes McKinney commented on ARROW-8231:
-------------------------------------
We do this in C++ -- the metadata keys are passed through unmodified
> Parse key_value_metadata from parquet FileMetaData into arrow schema metadata
> -----------------------------------------------------------------------------
>
> Key: ARROW-8231
> URL: https://issues.apache.org/jira/browse/ARROW-8231
> Project: Apache Arrow
> Issue Type: Improvement
> Components: Rust
> Reporter: Jörn Horstmann
> Priority: Minor
>
> The parquet-format FileMetaData struct contains optional key value pairs with
> additional metadata about the schema:
> [https://docs.rs/parquet-format/2.6.0/src/parquet_format/parquet_format.rs.html#3821]
> When the parquet file was generated using the java avro parquet writer, this
> for example contains the original avro schema under the `parquet.avro.schema`
> or `avro.schema` keys.
> It would be nice if this metadata was accessible through the
> `arrow::datatypes::Schema.metadata` field.
> I'm willing to implement and create a pull request for this feature.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
