[
https://issues.apache.org/jira/browse/HADOOP-1722?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12679865#action_12679865
]
Klaas Bosteels commented on HADOOP-1722:
----------------------------------------
Zhuweimin, presumably you're expecting the number of bytes reported by "wc -c"
to be equal to the number of bytes in your input files, but that's not what you
should be expecting really. Here's a quick outline of what happens when you run
that command:
# Since you didn't specify an InputFormat, the TextInputFormat is used which
leads to Text values corresponding to "lines" (i.e. sequences of bytes ending
with a newline character) and LongWritable keys corresponding to the offsets of
the lines in the file.
# Because you use rawbytes for the map input, Streaming will pass the keys and
values to your mapper as ~<4 byte length><raw bytes>~ byte sequences. These
byte sequences are obtained using Writable serialization (i.e. by calling the
{{write()}} method) and prepending the length to the bytes obtained in this way.
You could probably get the behavior you're after by writing a custom
[InputFormat|http://svn.apache.org/viewvc/hadoop/core/trunk/src/mapred/org/apache/hadoop/mapred/InputFormat.java?view=markup]
and
[InputWriter|http://svn.apache.org/viewvc/hadoop/core/trunk/src/contrib/streaming/src/java/org/apache/hadoop/streaming/io/InputWriter.java?view=markup],
but out of the box it's not supported at the moment as far as I know.
> Make streaming to handle non-utf8 byte array
> --------------------------------------------
>
> Key: HADOOP-1722
> URL: https://issues.apache.org/jira/browse/HADOOP-1722
> Project: Hadoop Core
> Issue Type: Improvement
> Components: contrib/streaming
> Reporter: Runping Qi
> Assignee: Klaas Bosteels
> Fix For: 0.21.0
>
> Attachments: HADOOP-1722-branch-0.18.patch,
> HADOOP-1722-branch-0.19.patch, HADOOP-1722-v2.patch, HADOOP-1722-v3.patch,
> HADOOP-1722-v4.patch, HADOOP-1722-v4.patch, HADOOP-1722-v5.patch,
> HADOOP-1722-v6.patch, HADOOP-1722.patch
>
>
> Right now, the streaming framework expects the output sof the steam process
> (mapper or reducer) are line
> oriented UTF-8 text. This limit makes it impossible to use those programs
> whose outputs may be non-UTF-8
> (international encoding, or maybe even binary data). Streaming can overcome
> this limit by introducing a simple
> encoding protocol. For example, it can allow the mapper/reducer to hexencode
> its keys/values,
> the framework decodes them in the Java side.
> This way, as long as the mapper/reducer executables follow this encoding
> protocol,
> they can output arabitary bytearray and the streaming framework can handle
> them.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
