tdwokflow

Unofficial Treasure Workflow API client.

Installation

pip install tdworkflow

If you want to use development version, run as follows:

pip install git+https://github.com/chezou/tdworkflow.git

Usage

import os

from tdworkflow.client import Client

apikey = os.getenv("TD_API_KEY")
client = Client(site="us", apikey=apikey)
# Or, write endpoint explicitly
# client = Client(endpoint="api-workflow.treasuredata.com", apikey=apikey)

projects = client.projects("pandas-df")

secrets = {"td.apikey": apikey, "td.apiserver": "https://api.treasuredata.com", "test": "secret-foo"}

client.set_secrets(projects[0], secrets)

client.secrets(projects[0])
# ['td.apikey', 'td.apiserver', "test"]
client.delete_secrets(projects[0], ["test", "td.apiserver"])

Upload Project from GitHub

Before executing the example code, you have to install git-python

pip install gitpython

Clone example repository with git-python and upload a digdag project.

import tempfile
import os
import shutil

import tdworkflow

from git import Git

# Download example GitHub repositoory

tempdir = tempfile.gettempdir()

git_repo = "https://github.com/treasure-data/treasure-boxes/"

shutil.rmtree(os.path.join(tempdir, "treasure-boxes"))

try:
    Git(tempdir).clone(git_repo)
    print("Clone repository succeeded")
except Exception:
    print("Repository clone failed")
    raise

# Upload specific Workflow project

apikey = os.getenv("TD_API_KEY")
site = "us"

target_box = os.path.join("integration-box", "python")
target_path = os.path.join(tempdir, "treasure-boxes", target_box)

client = tdworkflow.client.Client(site=site, apikey=apikey)
project = client.create_project("my-project", target_path)

If you want to open Treasure Workflow console on your browser, you can get the workflow URL as the following:

CONSOLE_URL = {
    "us": "https://console.treasuredata.com/app/workflows",
    "eu01": "https://console.eu01.treasuredata.com/app/workflows",
    "jp": "https://console.treasuredata.co.jp/app/workflows",
}

workflows = client.project_workflows(project)
workflows = list(filter(lambda w: w.name != "test", workflows))
if workflows:
    print(f"Project created! Open {CONSOLE_URL[site]}/{workflows[0].id}/info on your browser and click 'New Run' button.")
else:
    print("Project creation failed.")

Start workflow session

You can start a workflow session by using Client.start_attempt.

attempt = client.start_attempt(workflows[0])

# Wait attempt until finish. This may require few minutes.
attempt = client.wait_attempt(attempt)

Connect to open source digdag

Since Treasure Workflow is hosted digdag, tdworkflow is compatible with open source digdag.

Note

Open source digdag API may be different with Treasure Workflow API so that tdworkflow might not work with some API of opensource digdag.

Here is the example code to connect local digdag server.

>>> import tdworkflow
>>> import requests
>>> session = requests.Session()
>>> client = tdworkflow.client.Client(
... endpoint="localhost:65432", apikey="", _session=session, scheme="http")
>>> client.projects()
[Project(id=1, name='python-tdworkflow', revision='134fe2f9-ded3-4e7c-af8e-8a82d55d688b', archiveType='db', archiveMd5='5Lc6F6m3DtmBN4DA5MzK8A==', createdAt='2019-11-01T13:03:26Z', deletedAt=None, updatedAt='2019-11-01T13:03:26Z')]