[
https://issues.apache.org/jira/browse/SPARK-23148?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16332999#comment-16332999
]
Henry Robinson edited comment on SPARK-23148 at 1/19/18 11:25 PM:
------------------------------------------------------------------
It seems like the problem is that
{{CodecStreams.createInputStreamWithCloseResource}} can't properly handle a
{{path}} argument that's URL-encoded. We could add an overload for
{{createInputStreamWithCloseResource(Configuration, Path)}} and then pass {{new
Path(new URI(path))}} from {{CSVDataSource.readFile()}}. This has the benefit
of being a more localised change (and doesn't change the 'contract' that comes
from {{FileScanRDD}} currently having URL-encoded pathnames everywhere). A
strawman commit is
[here|https://github.com/henryr/spark/commit/b8c51418ee7d4bca18179fd863f7f4885c98c0ef].
was (Author: henryr):
It seems like the problem is that
{{CodecStreams.createInputStreamWithCloseResource}} can't properly handle a
{{path}} argument that's URL-encoded. We could add an overload for
{{createInputStreamWithCloseResource(Configuration, Path)}} and then pass {{new
Path(new URI(path))}} from {{CSVDataSource.readFile()}}. This has the benefit
of being a more localised change (and doesn't change the 'contract' that comes
from {{FileScanRDD}} currently having URL-encoded pathnames everywhere. A
strawman commit is
[here|https://github.com/henryr/spark/commit/b8c51418ee7d4bca18179fd863f7f4885c98c0ef].
> spark.read.csv with multiline=true gives FileNotFoundException if path
> contains spaces
> --------------------------------------------------------------------------------------
>
> Key: SPARK-23148
> URL: https://issues.apache.org/jira/browse/SPARK-23148
> Project: Spark
> Issue Type: Bug
> Components: SQL
> Affects Versions: 2.4.0
> Reporter: Bogdan Raducanu
> Priority: Major
>
> Repro code:
> {code:java}
> spark.range(10).write.csv("/tmp/a b c/a.csv")
> spark.read.option("multiLine", false).csv("/tmp/a b c/a.csv").count
> 10
> spark.read.option("multiLine", true).csv("/tmp/a b c/a.csv").count
> java.io.FileNotFoundException: File
> file:/tmp/a%20b%20c/a.csv/part-00000-cf84f9b2-5fe6-4f54-a130-a1737689db00-c000.csv
> does not exist
> {code}
> Trying to manually escape fails in a different place:
> {code}
> spark.read.option("multiLine", true).csv("/tmp/a%20b%20c/a.csv").count
> org.apache.spark.sql.AnalysisException: Path does not exist:
> file:/tmp/a%20b%20c/a.csv;
> at
> org.apache.spark.sql.execution.datasources.DataSource$.org$apache$spark$sql$execution$datasources$DataSource$$checkAndGlobPathIfNecessary(DataSource.scala:683)
> at
> org.apache.spark.sql.execution.datasources.DataSource$$anonfun$15.apply(DataSource.scala:387)
> at
> org.apache.spark.sql.execution.datasources.DataSource$$anonfun$15.apply(DataSource.scala:387)
> at
> scala.collection.TraversableLike$$anonfun$flatMap$1.apply(TraversableLike.scala:241)
> at
> scala.collection.TraversableLike$$anonfun$flatMap$1.apply(TraversableLike.scala:241)
> at scala.collection.immutable.List.foreach(List.scala:381)
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
