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.

Usage

from taskhuddler import TaskGraph

# All queries here will result in API calls
graph = TaskGraph('M5hSue6oRSu_klunMRHolg')
for task in graph.tasks():
    # These two are equivalent. Task() object knows about some features of a task
    print(task.json['status']['taskId'])
    print(task.taskid)  
# All tasks will be cached locally when TaskGraph is called
# But this means data may get stale.
cached_graph = TaskGraph('M5hSue6oRSu_klunMRHolg', caching=True)
for task in cached_graph.tasks():
    print(task.taskid)
cached_graph.refresh_task_cache()

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

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

Plans

• Let TaskGraph be used as a context manager
• Reduce the per-query limit for cached graphs so the initial response is quicker
• Make better use of Task classes for extra checks.
• Allow filters for task lists, so that things like this work: `graph.tasks(filter=lambda name: task['metadata']['name'] == name)