spark-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Eemeli Niemelä (JIRA) <j...@apache.org>
Subject [jira] [Comment Edited] (SPARK-12928) Oracle FLOAT datatype is not properly handled when reading via JDBC
Date Thu, 28 Apr 2016 10:46:12 GMT

    [ https://issues.apache.org/jira/browse/SPARK-12928?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15260031#comment-15260031
] 

Eemeli Niemelä edited comment on SPARK-12928 at 4/28/16 10:45 AM:
------------------------------------------------------------------

With pyspark this Oracle FLOAT-error appears with exception 
{noformat}
ValueError: (ValueError(u'Could not parse datatype: decimal(38,-127)',)
{noformat}

Workaround: in Oracle query, cast FLOAT to NUMBER if precision and scale can be set so that
all data can be included. If not, convert to FLOATs with TO_CHAR and change type later, eg.
{noformat}
sql = "(SELECT TO_CHAR(float_column) AS FLOAT_COLUMN FROM some_table)"
df = sc.read.format('jdbc').options(.. table=sql)

df = df.withColumn("FLOAT_COLUMN", df["FLOAT_COLUMN"].cast("float"))
{noformat}


was (Author: eniemela):
With pyspark this Oracle FLOAT-error appears with exception 
{noformat}
ValueError: (ValueError(u'Could not parse datatype: decimal(38,-127)',)
{noformat}

Workaround: in Oracle query, cast FLOAT to NUMBER if precision and scale can be set so that
all data can be included. If not, convert to FLOATS with TO_CHAR and change type later, eg.
{noformat}
sql = "(SELECT TO_CHAR(float_column) AS FLOAT_COLUMN FROM some_table)"
df = sc.read.format('jdbc').options(.. table=sql)

df = df.withColumn("FLOAT_COLUMN", df["FLOAT_COLUMN"].cast("float"))
{noformat}

> Oracle FLOAT datatype is not properly handled when reading via JDBC
> -------------------------------------------------------------------
>
>                 Key: SPARK-12928
>                 URL: https://issues.apache.org/jira/browse/SPARK-12928
>             Project: Spark
>          Issue Type: Bug
>          Components: SQL
>    Affects Versions: 1.6.0
>         Environment: Oracle Database 11g Enterprise Edition   11.2.0.3.0      64bit Production
>            Reporter: Greg Michalopoulos
>            Priority: Minor
>   Original Estimate: 24h
>  Remaining Estimate: 24h
>
> When trying to read in a table from Oracle and saveAsParquet, an IllegalArgumentException
is thrown when a column of FLOAT datatype is encountered.
> Below is the code being run:
> {code}val jdbcDF = sqlContext.read.format("jdbc").options(
>   Map("url" -> jdbcConnectionString,
>           "dbtable" -> "(select someFloat from someTable)",
>           "fetchSize" -> fetchSize)).load()
>   jdbcDF.saveAsParquetFile(destinationDirectory + table)
> {code}
> Here is the exception:
> {code}java.lang.IllegalArgumentException: Unsupported dataType: {"type":"struct","fields":[{"name":"someFloat","type":"decimal(38,-127)","nullable":true,"metadata":{"name":"someFloat"}}]},
[1.1] failure: `TimestampType' expected but `{' found
> {code}
> From the exception it was clear that the FLOAT datatype was presenting itself as scale
-127 which appears to be the problem. 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org


Mime
View raw message