added a count command

[gardenia]

This adds basic support for the count command:

https://hadoop.apache.org/docs/stable/hadoop-project-dist/hadoop-common/FileSystemShell.html#count

Currently no command-line options supported (other than -h) but I find this adds value to my own use-cases as is.

[colinmarc]

Hi @gardenia, thanks for the contribution. This command feels pretty un-unixy, which makes me think it's maybe not a good fit for this binary, but I'm open to discussing it. Are there usecases not covered by ls [-R] | wc?

[gardenia]

Hi @colinmarc ,

The main benefits as I see it are:

• the count command provides the convenience of reporting
  Number dirs
  Number files
  Number bytes

For a particular path in a single command invocation

• It uses an efficient single RPC call provided by the hdfs namenode (getContentSummary). So on a large cluster (large number of files/dirs) a ls -R may be many round trips and pulling back big listings just to compute the same information whereas this count implementation is just a single RPC

• The "count" command is standard within hdfs and therefore is something users of the original Java Hadoop command line may be familiar with and have got used to relying upon (as I myself had)

https://hadoop.apache.org/docs/stable/hadoop-project-dist/hadoop-common/FileSystemShell.html#count