[
https://issues.apache.org/jira/browse/SPARK-57354?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Calvin Qin updated SPARK-57354:
-------------------------------
Description:
Spark's file-based data sources skip hidden files and directories (names who
start with _ or . and files ending in ._COPYING _) during file listing and
partition discovery. This filter is always on, with no way to override it.
This adds a way to opt out of the filter, with a per-read data source option
`ignoredPathSegmentRegex` and a session config that default to `^[._]`. Note
that spark-specific files remain hardcoded and their listing behavior cannot be
overridden with this option.
When changed, the pattern is applied to path segments. Positive matches are
filtered out. `""` is special-cased to also list all files. The per-read option
takes precedent over the session config.
was:
Spark's file-based data sources skip hidden files and directories (names who
start with _ or . and files ending in ._COPYING _) during file listing and
partition discovery. This filter is always on, with no way to override it.
This adds a way to opt out of the filter, with a per-read data source option
`listHiddenFiles` and a session config `spark.sql.files.listHiddenFiles`
(default false)
When true, hidden files and directories are listed. False maintains default
behavior. The per-read option takes precedent over the session config.
> Add data source option to list hidden files
> -------------------------------------------
>
> Key: SPARK-57354
> URL: https://issues.apache.org/jira/browse/SPARK-57354
> Project: Spark
> Issue Type: New Feature
> Components: Spark Core
> Affects Versions: 4.2.0
> Reporter: Calvin Qin
> Assignee: Calvin Qin
> Priority: Major
> Labels: pull-request-available
> Fix For: 4.3.0
>
>
> Spark's file-based data sources skip hidden files and directories (names who
> start with _ or . and files ending in ._COPYING _) during file listing and
> partition discovery. This filter is always on, with no way to override it.
> This adds a way to opt out of the filter, with a per-read data source option
> `ignoredPathSegmentRegex` and a session config that default to `^[._]`. Note
> that spark-specific files remain hardcoded and their listing behavior cannot
> be overridden with this option.
> When changed, the pattern is applied to path segments. Positive matches are
> filtered out. `""` is special-cased to also list all files. The per-read
> option takes precedent over the session config.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
