def handle(self, *args, **options):
root_folder = options['newcomers_guide_path']
related_task_count = options['related_tasks']
related_service_count = options['related_services']
print(
'All task data and task/service similarity data will be deleted and reimported'
)
print('Reading tasks...')
tasks = read_task_descriptions(root_folder)
task_ids, task_descriptions = to_task_ids_and_descriptions(tasks)
print('{} tasks read, reading services...'.format(len(task_ids)))
service_ids, service_descriptions = to_service_ids_and_descriptions(
Service.objects.all())
descriptions = task_descriptions + service_descriptions
print('{} services read, computing similarities...'.format(
len(service_ids)))
cosine_doc_similarities = compute_similarities(descriptions)
print('Saving {} task similarities...'.format(
len(task_ids) * (len(task_ids) - 1)))
save_task_similarities(task_ids, cosine_doc_similarities,
related_task_count)
print('Saving {} task-service similarities...'.format(
len(task_ids) * len(service_ids)))
save_task_service_similarity_scores(task_ids, service_ids,
cosine_doc_similarities,
related_service_count)
def test_getting_description_for_task_returns_task_title_and_description(
self):
_, descriptions = to_task_ids_and_descriptions(self.data)
self.assertEqual(
descriptions[0],
self.english_task_title + ' ' + self.english_task_description)
def test_converts_task_id_to_slug(self):
self.data['taskMap'][self.task_id]['id'] = 'This is the id'
ids, _ = to_task_ids_and_descriptions(self.data)
self.assertEqual(ids[0], 'this-is-the-id')
def test_getting_ids_for_task_returns_task_id(self):
ids, _ = to_task_ids_and_descriptions(self.data)
self.assertEqual(ids[0], self.task_id)