[
https://issues.apache.org/jira/browse/SPARK-29969?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17046359#comment-17046359
]
Victor Zhang commented on SPARK-29969:
--------------------------------------
[~younggyuchun]
The description may be a bit confusing. I use beeline to connect spark thrift
server.
The parse_url function in spark depend on java.net.URI while java.net.URL in
hive.
{code:java}
spark-sql> SELECT
parse_url('http://uzzf.down.gsxzq.com/download/%E5%B8%B8%E7%94%A8%E9%98%80%E9%97%A8CAD%E5%9B%BE%E7%BA%B8%E5%A4%',
'HOST');
NULL
Time taken: 1.211 seconds, Fetched 1 row(s)
{code}
{code:java}
hive> SELECT
parse_url('http://uzzf.down.gsxzq.com/download/%E5%B8%B8%E7%94%A8%E9%98%80%E9%97%A8CAD%E5%9B%BE%E7%BA%B8%E5%A4%',
'HOST');
OK
HEADER: _c0
uzzf.down.gsxzq.com
Time taken: 0.039 seconds, Fetched: 1 row(s)
{code}
> parse_url function result in incorrect result
> ---------------------------------------------
>
> Key: SPARK-29969
> URL: https://issues.apache.org/jira/browse/SPARK-29969
> Project: Spark
> Issue Type: Bug
> Components: SQL
> Affects Versions: 2.3.1, 2.4.4
> Reporter: Victor Zhang
> Priority: Major
> Attachments: hive-result.jpg, spark-result.jpg
>
>
> In this Jira using java.net.URI instead of java.net.URL for performance
> reason.
> https://issues.apache.org/jira/browse/SPARK-16826
> However, in the case of some unconventional parameters, it can lead to
> incorrect results.
> For example, when the URL is encoded, the function cannot resolve the correct
> result.
>
> {code}
> 0: jdbc:hive2://localhost:10000> SELECT
> parse_url('http://uzzf.down.gsxzq.com/download/%E5%B8%B8%E7%94%A8%E9%98%80%E9%97%A8CAD%E5%9B%BE%E7%BA%B8%E5%A4%',
> 'HOST');
> +------------------------------------------------------------------------------------------------------------------------+--+
> |
> parse_url(http://uzzf.down.gsxzq.com/download/%E5%B8%B8%E7%94%A8%E9%98%80%E9%97%A8CAD%E5%9B%BE%E7%BA%B8%E5%A4%,
> HOST) |
> +------------------------------------------------------------------------------------------------------------------------+--+
> | NULL |
> +------------------------------------------------------------------------------------------------------------------------+--+
> 1 row selected (0.094 seconds)
>
> hive> SELECT
> parse_url('http://uzzf.down.gsxzq.com/download/%E5%B8%B8%E7%94%A8%E9%98%80%E9%97%A8CAD%E5%9B%BE%E7%BA%B8%E5%A4%',
> 'HOST');
> OK
> HEADER: _c0
> uzzf.down.gsxzq.com
> Time taken: 4.423 seconds, Fetched: 1 row(s)
> {code}
>
> Here's a similar problem.
> https://issues.apache.org/jira/browse/SPARK-23056
> Our team used the spark function to run data for months, but now we have to
> run it again.
> It's just too painful.:(:(:(
>
>
>
>
>
>
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
