Python Pachyderm

Official Python Pachyderm client. Created by kalugny (formerly kalugny/pypachy.)

See the API docs. Most of values are auto-generated from protobufs. It's generally easier to rely on the higher-level classes if they provide the functionality you need:

• PfsClient
• PpsClient

Installation

pip install python-pachyderm

Usage and options

All of the PFS functions used in pachctl are supported (almost) as-is.

There are some helper functions that help make things more pythonic:

• commit which is a context manager wrapper for start_commit and finish_commit
• get_files which supports getting the data from multiple files

Naming commits

All functions that accept a commit argument will accept a tuple of (repo, branch) or (repo, commit_id), a string like repo/branch or repo/commit_id and a Commit object.

e.g:

client.list_file(('my_repo', 'branch'), '/')  # tuple
client.list_file('my_repo/commit_id', '/')    # string
c = client.list_commit('my_repo')[0]          # get some commit
client.list_file(c, '/')                      # and use it directly

Basic usage example

import python_pachyderm
client = python_pachyderm.PfsClient()
client.create_repo('test')
with client.commit('test', 'master') as c:
    client.put_file_bytes(c, '/dir_a/data', b'DATA')
    client.put_file_url(c, '/dir_b/icon.png', 'http://www.pearl-guide.com/forum/images/smilies/biggrin.png')

client.get_files('test/master', '/', recursive=True)

As of version 0.1.4, there is also limited support for PPS:

pps_client = python_pachyderm.PpsClient()
pps_client.list_pipeline()
...

Contributing

This driver is co-maintained by Pachyderm and the community. If you're looking to contribute to the project, this is a fantastic place to get involved.

Getting started

To run tests, clone the repo, then run:

make init
tox