def run(self, form):
provenance = None#form.cleaned_data["provenance"] #TODO: is dit correct?
job_size = form.cleaned_data["job_size"]
self.monitor.update(10, "Executing query..")
article_ids = list(SelectionSearch(form).get_article_ids())
cj = CodingJob()
cj.project = self.project
cj.name = form.cleaned_data["name"]
cj.unitschema = form.cleaned_data["unitschema"]
cj.articleschema = form.cleaned_data["articleschema"]
cj.coder = form.cleaned_data["coder"]
cj.insertuser = self.user
self.monitor.update(50, "Creating codingjobs..")
if job_size == 0:
job_size = len(article_ids)
n_batches = len(article_ids) // job_size
n_batches += 1 if len(article_ids) % job_size else 0
for i, cid in enumerate(_create_codingjob_batches(cj, article_ids, job_size)):
progress = int((i / float(n_batches)) * (100 // 2))
msg = "Creating codingjob {} of {}..".format(i+1, n_batches)
print(50 + progress)
self.monitor.update(50 + progress, msg)
if provenance:
cj = CodingJob.objects.get(id=cid)
cj.provenance = provenance
cj.save()
return "Codingjob(s) created."
def import_codingjobs(self):
old_ids, jobs = [], []
for job in self._get_dicts("codingjobs.jsonl"):
j = CodingJob(project_id=self.status.project.id, name=job['name'], archived=job['archived'],
insertuser_id=self.status.project.owner.id, coder_id=self.status.users[job['coder']],
articleset_id=self.status.setids[job['articleset']])
if job['articleschema']:
j.articleschema_id = self.status.codingschemas[job['articleschema']]
if job['unitschema']:
j.unitschema_id = self.status.codingschemas[job['unitschema']]
jobs.append(j)
old_ids.append(job['pk'])
jobs = bulk_insert_returning_ids(jobs)
return {old_id: job.id for (old_id, job) in zip(old_ids, jobs)}
def test_create_codingjob_batches(self):
a = amcattest.create_test_set(10)
cj = CodingJob()
cj.project = a.project
cj.name = "foo"
cj.unitschema = amcattest.create_test_schema(project=a.project)
cj.articleschema = amcattest.create_test_schema(isarticleschema=True,
project=a.project)
cj.coder = amcattest.create_test_user()
cj.insertuser = amcattest.create_test_user()
arts = a.articles.all().values_list("id", flat=True)
cjs = create_codingjob_batches(cj, arts, 2)
self.assertEqual(5, len(cjs))
cjs = create_codingjob_batches(cj, arts, 3)
self.assertEqual(4, len(cjs))
def test_create_codingjob_batches(self):
a = amcattest.create_test_set(10)
cj = CodingJob()
cj.project = a.project
cj.name = "foo"
cj.unitschema = amcattest.create_test_schema(project=a.project)
cj.articleschema = amcattest.create_test_schema(isarticleschema=True, project=a.project)
cj.coder = amcattest.create_test_user()
cj.insertuser = amcattest.create_test_user()
arts = a.articles.all().values_list("id", flat=True)
cjs = create_codingjob_batches(cj, arts, 2)
self.assertEqual(5, len(cjs))
cjs = create_codingjob_batches(cj, arts, 3)
self.assertEqual(4, len(cjs))
def run(self, form):
provenance = None #form.cleaned_data["provenance"] #TODO: is dit correct?
job_size = form.cleaned_data["job_size"]
self.monitor.update(10, "Executing query..")
article_ids = list(SelectionSearch(form).get_article_ids())
cj = CodingJob()
cj.project = self.project
cj.name = form.cleaned_data["name"]
cj.unitschema = form.cleaned_data["unitschema"]
cj.articleschema = form.cleaned_data["articleschema"]
cj.coder = form.cleaned_data["coder"]
cj.insertuser = self.user
self.monitor.update(50, "Creating codingjobs..")
if job_size == 0:
job_size = len(article_ids)
n_batches = len(article_ids) // job_size
n_batches += 1 if len(article_ids) % job_size else 0
for i, cid in enumerate(
_create_codingjob_batches(cj, article_ids, job_size)):
progress = int((i / float(n_batches)) * (100 // 2))
msg = "Creating codingjob {} of {}..".format(i + 1, n_batches)
print(50 + progress)
self.monitor.update(50 + progress, msg)
if provenance:
cj = CodingJob.objects.get(id=cid)
cj.provenance = provenance
cj.save()
return "Codingjob(s) created."
