[
https://issues.apache.org/jira/browse/BEAM-2994?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Reuven Lax updated BEAM-2994:
-----------------------------
Fix Version/s: (was: 2.2.0)
2.3.0
> Refactor TikaIO
> ---------------
>
> Key: BEAM-2994
> URL: https://issues.apache.org/jira/browse/BEAM-2994
> Project: Beam
> Issue Type: Task
> Components: sdk-java-extensions
> Affects Versions: 2.2.0
> Reporter: Sergey Beryozkin
> Assignee: Sergey Beryozkin
> Fix For: 2.3.0
>
>
> TikaIO is currently implemented as a BoundedSource and asynchronous
> BoundedReader returning individual document's text chunks as Strings,
> eventually passed unordered (and not linked to the original documents) to the
> pipeline functions.
> It was decided in the recent beam-dev thread that initially TikaIO should
> support the cases where only a single composite bean per file, capturing the
> file content, location (or name) and metadata, should flow to the pipeline,
> and thus avoiding the need to implement TikaIO as a BoundedSource/Reader.
> Enhancing TikaIO to support the streaming of the content into the pipelines
> may be considered in the next phase, based on the specific use-cases...
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
