You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@arrow.apache.org by "Wes McKinney (JIRA)" <ji...@apache.org> on 2019/08/05 21:25:00 UTC
[jira] [Commented] (ARROW-6133) [Java][Python] Schema Missing
Exception in ArrowStreamReader
[ https://issues.apache.org/jira/browse/ARROW-6133?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16900419#comment-16900419 ]
Wes McKinney commented on ARROW-6133:
-------------------------------------
The output of {{pyarrow.serialize}} is not an IPC stream compatible with ArrowStreamReader. You need to use the Python stream writer class, see http://arrow.apache.org/docs/python/generated/pyarrow.RecordBatchStreamWriter.html#pyarrow.RecordBatchStreamWriter
> [Java][Python] Schema Missing Exception in ArrowStreamReader
> ------------------------------------------------------------
>
> Key: ARROW-6133
> URL: https://issues.apache.org/jira/browse/ARROW-6133
> Project: Apache Arrow
> Issue Type: Bug
> Components: Java
> Affects Versions: 0.14.1
> Reporter: Boris V.Kuznetsov
> Priority: Major
>
> Hello
> My colleague and I are trying to pass Arrow thru Kafka. He uses a PyArrow, I'm using Scala Java API.
> Here's the Transmitter code:
> {code:python}
> import pyarrow as pa
> def record_batch_to_bytes(df):
> batch = pa.RecordBatch.from_pandas(df)
> ser_ = pa.serialize(batch)
> return bytes(ser_.to_buffer())
> {code}
>
> My colleague is able to read this stream with the Python API:
> {code:python}
> def bytes_to_batch_record(bytes_):
> batch = pa.deserialize(bytes_)
> print(batch.schema)
> {code}
> On the Receiver side, I use the following from Java API:
> {code:scala}
> def deserialize(din: Chunk[BArr]): Chunk[ArrowStreamReader] =
> for {
> arr <- din
> stream = new ByteArrayInputStream(arr)
> } yield new ArrowStreamReader(stream, allocator)
>
> reader = deserialize(arr)
> schema = reader.map(r => r.getVectorSchemaRoot.getSchema)
> empty = reader.map(r => r.loadNextBatch)
> {code}
>
> Which fails with exception on both lines 2 and 3 in the last snippet:
> {code:bash}
> Fiber failed.
> An unchecked error was produced.
> java.io.IOException: Unexpected end of input. Missing schema.
> at org.apache.arrow.vector.ipc.ArrowStreamReader.readSchema(ArrowStreamReader.java:135)
> at org.apache.arrow.vector.ipc.ArrowReader.initialize(ArrowReader.java:178)
> at org.apache.arrow.vector.ipc.ArrowReader.ensureInitialized(ArrowReader.java:169)
> at org.apache.arrow.vector.ipc.ArrowReader.getVectorSchemaRoot(ArrowReader.java:62)
> at nettest.ArrowSpec.$anonfun$testConsumeArrow$7(Arrow.scala:96)
> at zio.Chunk$Arr.map(Chunk.scala:722)
> {code}
>
> The full Scala code is [here|https://github.com/Clover-Group/zio-tsp/blob/46e34c7c060bf4061067922077bbe05ea4b9f301/src/test/scala/Arrow.scala#L95]
>
> How do I resolve that ? We both are using Arrow 0.14.1 and my colleague has no issues with PyArrow API.
> Thank you!
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)