A higher level wrapper around taskcluster-client.py with the aim of having a more Pythonic interface to taskcluster.

Currently aiming to get easier, read-only features available.

Installation

pip install taskhuddler

For further data analysis, add the optional Pandas dependency

pip install taskhuddler[pandas]

Examining a Task Group

from taskhuddler import TaskGraph

# All tasks will be cached in memory when TaskGraph is called
# But this means data may get stale.
graph = TaskGraph('M5hSue6oRSu_klunMRHolg')
for task in graph.tasks():
    print(task.taskId)
# Fetch the set of tasks again.
graph.fetch_tasks()

# On-disk caching:
os.environ['TC_CACHE_DIR'] = '/tmp/cache/'
# All future TaskGraph calls from now on will write the
# json to TC_CACHE_DIR and will avoid a call to taskcluster


# Are all the tasks in the 'completed' state?
print(graph.completed)

if graph.completed:
    started = graph.earliest_start_time
    finished = graph.latest_finished_time
    print("Graph took {} to run".format(finished-started))

Examining Tasks

TaskDefinition, TaskStatus and Task are all available to work with tasks. A Task is a wrapper around a TaskDefinition and a TaskStatus.

The Task class populates both a TaskStatus and TaskDefinition, each of which can be used by themselves

from taskhuddler import Task, TaskDefinition, TaskStatus
from dataclasses import asdict

mytask = Task.from_task_id('M5hSue6oRSu_klunMRHolg')
print(task.status.state)

print(asdict(task))

my_task_def = TaskDefinition.from_task_id('M5hSue6oRSu_klunMRHolg')
my_task_def = TaskDefinition.from_dict(mytask.task)

my_task_status = TaskStatus.from_task_id('M5hSue6oRSu_klunMRHolg')
my_task_status = TaskStatus.from_dict(mytask.status)

Pandas

With pip install taskhuddler[pandas] the to_datetime method becomes available, returning a Pandas DataFrame with task and run data:

from taskhuddler import TaskGraph

graph = TaskGraph('M5hSue6oRSu_klunMRHolg')

df = graph.to_dataframe()

Plans

• Reduce the per-query limit for cached graphs so the initial response is quicker