Follow github instructions https://github.com/d6t/d6tpipe#installation

First-time setup and registration

  1. Sign up at https://pipe.databolt.tech/gui/user-signup/
  2. Get API token at http://pipe.databolt.tech/gui/api-access
  3. Set up d6tpipe to use token

Using command line

d6tpipe profiles token set --token <your-token>

Or using python

import d6tpipe
# cloud repo API token
api = d6tpipe.api.APIClient()
api.setToken('<your-token>') # DONT SHARE YOUR TOKEN! Do not save in code, just run it once

Run this ONCE. You do NOT to have to run this every time you use d6tpipe

Advanced: create profile with custom filerepo path

import d6tpipe
api = d6tpipe.api.APIClient(profile='profile-name')
api.setToken('<your-token>') # DONT SHARE YOUR TOKEN! Do not save in code, just run it once

See Config and Connect for details.

Where are files saved?

See Config

Pull files from remote to local

Using the command line tool is the fastest way to pull files.

# list data pipes you have access to
d6tpipe list remote

# pull files from a data pipe
d6tpipe pull --pipe intro-stat-learning --preview
d6tpipe pull --pipe intro-stat-learning
d6tpipe --help

To pull files using python, you connect to a data pipe. A data pipe lets you manage remote and local data files.

import d6tpipe
api = d6tpipe.api.APIClient()

# list data pipes you have access to

# pull files from a data pipe
pipe = d6tpipe.Pipe(api, 'intro-stat-learning') # connect to a data pipe
pipe.pull_preview() # preview files and size
pipe.pull() # download all data with just one command

# advanced: show location and credentials

See Pull Files for details.

Access and read local files

Remote files that are pulled get stored in a central local file directory.

# show local files
pipe.dirpath # where files are stored

# read a file into pandas
import pandas as pd
df = pd.read_csv(pipe.dirpath/'Advertising.csv')

See Read Files for details.

Process files

You now have a lot of powerful functions to easily manage all your files from a central location across multiple projects.

# use schema to quickly load data
df = pd.read_csv(pipe.dirpath / 'Advertising.csv', **pipe.schema['pandas'])

# read multiple files into dask
import dask.dataframe as dd
files = pipe.filepaths(include='Advertising*.csv')
ddf = dd.read_csv(files, **pipe.schema['dask'])

# open most recent CSV
df = pd.read_csv(pipe.files(sortby='mod')[-1])

# save data to local files

See Process Files for details.

Advanced Topics

This covers pushing files and creating your own remote file storage and data pipes.

Write Local Files and Push to Remote

You can easily save new files to the pipe. You can also push files from local to remote if you have write access or manage your own pipes.

# create some new data
import sklearn.preprocessing
df_scaled = df.apply(lambda x: sklearn.preprocessing.scale(x))

# conveniently save files in a central repo
df_scaled.to_csv(pipe.dirpath/'Advertising-scaled.csv') # pipe.dirpath points to local pipe folder

# alternatively, import another folder

# list files in local directory

# upload files - just one command!
pipe.push_preview() # preview files and size
pipe.push() # execute

See Push for details.

Alternatively you can use the command line tool.

d6tpipe push --pipe intro-stat-learning --preview
d6tpipe push --pipe intro-stat-learning
d6tpipe --help

Register and administer pipes

You can register your own pipes that point to your own remote data storage. d6tpipe has managed remotes which makes it very easy for you to set up and manage professional remote data file storage.

import d6tpipe
api = d6tpipe.api.APIClient()

# managed remote file stores can be created quickly with just one command
d6tpipe.upsert_pipe(api, {'name': 'your-pipe'})

See Pipes for details. For creating self-hosted remotes, see Advanced Pipes.

Share pipes

After you’ve registered a pipe, you can give others access to the remote data. By default only you have access so to share it with others you have to grant them access.

import d6tpipe
api = d6tpipe.api.APIClient()

# give another user access
settings = {"username":"another-user","role":"read"} # read, write, admin
d6tpipe.upsert_permissions(api, 'your-pipe', settings)

# make data repo public
settings = {"username":"public","role":"read"}
d6tpipe.upsert_permissions(api, 'your-pipe', settings)

See Permissions for details.