def gold_reset(self):
'''
Mark all of this job's test questions (gold data) as NOT gold.
Splitting the /jobs/:job_id/gold API call into gold_reset() and
gold_add() is not faithful to the API, but resetting gold marks
and adding them should not have the same API endpoint in the first place.
'''
params = dict(reset='true')
res = self._connection.request('/jobs/%s/gold' % self.id, method='PUT', params=params)
# reset cache
self._cache.remove(keyfunc(self, 'properties'))
self._cache.remove(keyfunc(self, 'units'))
return res
def update(self, props):
params = rails_params({'job': props})
logger.debug('Updating Job[%d]: %r', self.id, params)
res = self._connection.request('/jobs/%s' % self.id, method='PUT', params=params)
# reset cached properties
self._cache.remove(keyfunc(self, 'properties'))
return res
def upload(self, units):
headers = {'Content-Type': 'application/json'}
data = '\n'.join(json.dumps(unit) for unit in units)
logger.debug('Upload data: {}'.format(data))
res = self._connection.request('/jobs/%s/upload' % self.id, method='POST', headers=headers, data=data)
# reset cached units
self._cache.remove(keyfunc(self, 'units'))
return res
def create_job(self, props):
'''
Creates an empty job with given `props` attributes.
'''
params = rails_params({'job': props})
job_response = self.request('/jobs', method='POST', params=params)
job = Job(job_response['id'], self)
job._properties = job_response
# bust cache of job_ids
self._cache.remove(keyfunc(self, 'job_ids'))
return job
def create_job(self, props):
'''
Creates an empty job with given `props` attributes.
'''
params = rails_params({'job': props})
job_response = self.request('/jobs', method='POST', params=params)
job = Job(job_response['id'], self)
job._properties = job_response
# bust cache of job_ids
self._cache.remove(keyfunc(self, 'job_ids'))
return job
def gold_add(self, check, check_with=None):
'''
Configure the gold labels for a task.
* check: the name of the field being checked against
- Can call /jobs/{job_id}/legend to see options
- And as far as I can tell, the job.properties['gold'] field is a
hash with keys that are "check" names, and values that are "with" names.
* check_with: the name of the field containing the gold label for check
- Crowdflower calls this field "with", which is a Python keyword
- defaults to check + '_gold'
I'm not sure why convert_units would be anything but true.
'''
params = dict(check=check, convert_units='true')
if check_with is not None:
params['with'] = check_with
res = self._connection.request('/jobs/%s/gold' % self.id, method='PUT', params=params)
# reset cache
self._cache.remove(keyfunc(self, 'properties'))
self._cache.remove(keyfunc(self, 'units'))
return res
def upload(self, units):
'''
TODO: allow setting Job parameters at the same time
'''
headers = {'Content-Type': 'application/json'}
# N.b.: CF expects newline-separated JSON, not actual JSON
# e.g., this would fail with a status 500: kwargs['data'] = json.dumps(data)
data = '\n'.join(json.dumps(unit) for unit in units)
job_response = self.request('/jobs/upload', method='POST', headers=headers, data=data)
job = Job(job_response['id'], self)
job._properties = job_response
# bust cache of job_ids
self._cache.remove(keyfunc(self, 'job_ids'))
return job
def upload(self, units):
'''
TODO: allow setting Job parameters at the same time
'''
headers = {'Content-Type': 'application/json'}
# N.b.: CF expects newline-separated JSON, not actual JSON
# e.g., this would fail with a status 500: kwargs['data'] = json.dumps(data)
data = '\n'.join(json.dumps(unit) for unit in units)
job_response = self.request('/jobs/upload',
method='POST',
headers=headers,
data=data)
job = Job(job_response['id'], self)
job._properties = job_response
# bust cache of job_ids
self._cache.remove(keyfunc(self, 'job_ids'))
return job
def _cache_flush(self, func_attr):
self._cache.remove(keyfunc(self, func_attr))
def _cache_flush(self, func_attr):
self._cache.remove(keyfunc(self, func_attr))
def add_tags(self, tags):
params = rails_params({'tags': tags})
self._connection.request('/jobs/%s/tags' % self.id, method='POST', params=params)
self._cache.remove(keyfunc(self, 'tags'))
def delete_unit(self, unit_id):
response = self._connection.request('/jobs/%s/units/%s' % (self.id, unit_id), method='DELETE')
# bust cache if the request did not raise any errors
self._cache.remove(keyfunc(self, 'units'))
return response