crowdflower.job

class crowdflower.job.Job(client=None, **data)

CrowdFlower Job.

Documentation for attributes can be found at https://success.crowdflower.com/hc/en-us/articles/202703435-CrowdFlower-API-Jobs-Resource-Attributes.

Parameters:
add_tag(tag)

Add tag.

cancel()

Permanently cancel a Job, stopping any incoming judgments and refunding your account for unreceived judgments.

channels

List of enabled channels for this job.

convert_test_questions()

Convert uploaded golden units to test questions.

copy(all_units=False, gold=False)

Create a new job that is a copy of this job.

Parameters:
  • all_units – If true, all of this job’s units will be copied to the new job.
  • gold – If true, only golden units will be copied to the new job.
Returns:

crowdflower.job.Job
delete()

Delete this job, removing it from CrowdFlower. Calling Job instance will be invalid after deletion and must not be used anymore.

get_judgment(judgment_id)

Get single Judgment for this Job.

get_results_report()

Download and parse JSON report containing aggregates and individual judgments as a list of Units. :returns: list of crowdflower.unit.Unit

get_worker(worker_id)

Get Worker worker_id bound to this Job.

judgment_aggregates

List of JudgmentAggregate instances of this Job.

Warning

Judgments are paged with a maximum of 100 items per page. If your job has a lot of judgments – thousands or more – this will take a very VERY long time to finish when accessed for the first time. This might produce some nasty surprises, if Job instances are inspected with inspect.getmembers() or some such.

launch(units_count, channels=('on_demand', ))

Order job with units_count at channels.

legend()

Display generated keys submitted with the form.

pause()

Temporarily stop judgments from coming in. A paused Job may resume.

ping()

Check the status/progress of Job.

resume()

Resume a Job from pause at any time.

tags

List of tags.

units

List of UnitPromise instances of this Job.

update()

Send updates made to this instance to CrowdFlower. Note that title, instructions and cml attributes must exist (either in the update or in the job already) for any changes to really persist, and so this method raises a RuntimeError, if any of them is missing.

“At minimum, your job must have a valid title, instructions, and one required CML form element to be saved successfully.” [1]

Warning

The API will happily return a “valid” response when sent only the ‘instructions’, but nothing will change on the server side without all three. The caller is responsible for providing valid CML.

[1]https://success.crowdflower.com/hc/en-us/articles/202703435-CrowdFlower-API-Jobs-Resource-Attributes (Tue Jun 30 07:31:00 UTC 2015)
Raises:RuntimeError – if title, instructions or cml is missing
upload(data, force=False)

Upload given data as JSON.

Parameters:
  • data (collections.abc.Iterable) – Iterable of JSON serializable objects
  • force (bool) – If True force adding units even if the columns do not match existing data
upload_file(file, type_=None, force=False)

Upload a file like object or open a file for reading and upload.

Caller is responsible for handling context on file like objects. Type must be provided with data as there is no information to make a guess from. If file like object provides text (unicode) data, it will be encoded to UTF-8 bytes.

If explicit type_ is not provided and the file is a string containing a filename to open, will make a guess with mimetypes.

If type information is not given and guessing did not work, will raise a ValueError.

Valid types are text/csv and application/json for .csv and .json respectively.

Parameters:
  • file (str or file) – A file like object or a filename string, contains UTF-8 encoded data
  • type (str) – Explicit type, required for file like objects
  • force (bool) – If True force adding units even if the columns do not match existing data
Raises:

ValueError – if type information isn’t provided and cannot guess