[
https://issues.apache.org/jira/browse/HIVE-16829?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16514928#comment-16514928
]
Alex S edited comment on HIVE-16829 at 6/17/18 12:57 AM:
---------------------------------------------------------
As a work around you can change a record delimiter from \n to some other value
so hive doesn't get confused by getting \n within the formatted json record.
You will also need to change record delimiter in your json files accordingly.
You can use command below in CLI to change the delimiter value:
set textinputformat.record.delimiter='<delimiter>';
Why would we need a separate RecordReader?
was (Author: alse):
{{As a work around you can change a record delimiter from \n to some other
value so hive doesn't get confused by getting \n within the formatted json
record. You will also need to change record delimiter in your json files
accordingly. }}
{{You can use command below in CLI to change the delimiter value:}}
{{set textinputformat.record.delimiter='<delimiter>';}}
{{Why would we need a separate RecordReader? }}
> A multiline/pretty-print JSON Data Record Reader
> ------------------------------------------------
>
> Key: HIVE-16829
> URL: https://issues.apache.org/jira/browse/HIVE-16829
> Project: Hive
> Issue Type: Improvement
> Components: File Formats
> Reporter: Ujjawal Nayak
> Priority: Trivial
> Original Estimate: 168h
> Remaining Estimate: 168h
>
> We already have 'org.apache.hive.hcatalog.data.JsonSerDe' which can
> serialize/deserialize a JSON Record. But it fails when it gets a JSON record
> which is in pretty-print format. And we do not have a RecordReader which can
> read this multiline JSON structure.
> We should create a RecordReader which can create a record from pretty-print
> format of JSON.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
