def get_template(self, obj, return_type='full'): task_worker = None if return_type == 'full': template = TemplateSerializer(instance=obj.project.template, many=False).data else: template = \ TemplateSerializer(instance=obj.project.template, many=False, fields=('id', 'items')).data data = obj.data if 'task_worker' in self.context: task_worker = self.context['task_worker'] for item in template['items']: aux_attrib = item['aux_attributes'] if 'src' in aux_attrib: aux_attrib['src'] = get_template_string(aux_attrib['src'], data)[0] if 'question' in aux_attrib: return_value, has_variables = get_template_string(aux_attrib['question']['value'], data) aux_attrib['question']['value'] = return_value aux_attrib['question']['is_static'] = not has_variables if 'options' in aux_attrib: if obj.project.is_review and 'task_workers' in obj.data: aux_attrib['options'] = [] display_labels = ['Top one', 'Bottom one'] sorted_task_workers = sorted(obj.data['task_workers'], key=itemgetter('task_worker')) # TODO change this to id for index, tw in enumerate(sorted_task_workers): aux_attrib['options'].append( { "value": tw['task_worker'], "display_value": display_labels[index], "data_source": [], "position": index + 1 } ) for option in aux_attrib['options']: option['value'] = get_template_string(option['value'], data)[0] if item['type'] == 'iframe': from django.conf import settings from hashids import Hashids identifier = Hashids(salt=settings.SECRET_KEY, min_length=settings.ID_HASH_MIN_LENGTH) if hasattr(task_worker, 'id'): item['identifier'] = identifier.encode(task_worker.id, task_worker.task.id, item['id']) else: item['identifier'] = 'READ_ONLY' if item['role'] == 'input' and task_worker is not None: for result in task_worker.results.all(): if item['type'] == 'checkbox' and result.template_item_id == item['id']: item['aux_attributes']['options'] = result.result # might need to loop through options elif item['type'] == 'file_upload' and result.template_item_id == item['id']: item['answer'] = { "name": result.attachment.name } elif result.template_item_id == item['id']: item['answer'] = result.result template['items'] = sorted(template['items'], key=lambda k: k['position']) return template
def create_review(project, template_data, parent_review_project=None): project_name = 'Peer Review for ' + project.name review_project = models.Project.objects.create( name=project_name, owner=project.owner, parent=project, is_prototype=False, min_rating=1.99, post_mturk=False, timeout=project.timeout, is_review=True, deleted_at=timezone.now()) if parent_review_project is not None: review_project.price = parent_review_project.price template_serializer = TemplateSerializer(data=template_data) if template_serializer.is_valid(): review_template = template_serializer.create(with_defaults=False, is_review=True, owner=project.owner) review_project.template = review_template else: raise ValidationError(template_serializer.errors) review_project.group_id = review_project.id review_project.save() return review_project
def create(self, with_defaults=True, **kwargs): template_initial = self.validated_data.pop('template') if 'template' in self.validated_data else None template_items = template_initial['items'] if template_initial else [] template = { "name": 't_' + generate_random_id(), "items": template_items } template_serializer = TemplateSerializer(data=template) self.validated_data.pop('post_mturk') project = models.Project.objects.create(owner=kwargs['owner'], post_mturk=True, amount_due=0, **self.validated_data) if template_serializer.is_valid(): project_template = template_serializer.create(with_defaults=with_defaults, is_review=False, owner=kwargs['owner']) project.template = project_template else: raise ValidationError(template_serializer.errors) project.group_id = project.id if not with_defaults: project.status = models.Project.STATUS_IN_PROGRESS project.published_at = timezone.now() # self.instance = project # if not project.is_paid: # self.pay(self.instance.price * self.instance.repetition) # self.create_task(project.id) project.save() self.create_review(project=project, template_data=template) models.BoomerangLog.objects.create(object_id=project.group_id, min_rating=project.min_rating, rating_updated_at=project.rating_updated_at, reason='DEFAULT') return project
def create(self, **kwargs): project = models.Project.objects.create(deleted=False, owner=kwargs["owner"].requester) template = {"name": "t_" + generate_random_id()} template_serializer = TemplateSerializer(data=template) template = None if template_serializer.is_valid(): template = template_serializer.create(with_default=True, owner=kwargs["owner"]) else: raise ValidationError(template_serializer.errors) models.ProjectTemplate.objects.get_or_create(project=project, template=template) return project
def create(self, **kwargs): project = models.Project.objects.create( deleted=False, owner=kwargs['owner'].requester) template = {"name": 't_' + generate_random_id()} template_serializer = TemplateSerializer(data=template) template = None if template_serializer.is_valid(): template = template_serializer.create(with_default=True, owner=kwargs['owner']) else: raise ValidationError(template_serializer.errors) models.ProjectTemplate.objects.get_or_create(project=project, template=template) return project
def get_task_template(self, obj, return_type='full'): template = None if return_type == 'full': template = TemplateSerializer(instance=obj.module.template, many=True).data[0] else: template = \ TemplateSerializer(instance=obj.module.template, many=True, fields=('id', 'template_items')).data[0] data = json.loads(obj.data) for item in template['template_items']: if item['data_source'] is not None and item['data_source'] in data: item['values'] = data[item['data_source']] template['template_items'] = sorted(template['template_items'], key=lambda k: k['position']) return template
def create_revision(instance): models.Project.objects.filter(group_id=instance.group_id).update( status=models.Project.STATUS_PAUSED) template = TemplateSerializer.create_revision( instance=instance.template) batch_files = copy.copy(instance.batch_files.all()) tasks = copy.copy(instance.tasks.all()) mturk_update_status.delay({ 'id': instance.id, 'status': models.Project.STATUS_PAUSED }) instance.pk = None instance.template = template instance.status = models.Project.STATUS_DRAFT instance.is_prototype = False instance.is_paid = False instance.save() for f in batch_files: models.ProjectBatchFile.objects.create(project=instance, batch_file=f) for t in tasks: t.pk = None t.project = instance TaskSerializer.bulk_create(data=tasks) return instance
def create_review(project, template_data, parent_review_project=None): project_name = 'Peer Review for ' + project.name review_project = models.Project.objects.create(name=project_name, owner=project.owner, parent=project, is_prototype=False, min_rating=1.99, post_mturk=True, timeout=project.timeout, is_review=True, deleted_at=timezone.now()) if parent_review_project is not None: review_project.price = parent_review_project.price template_serializer = TemplateSerializer(data=template_data) if template_serializer.is_valid(): review_template = template_serializer.create(with_defaults=False, is_review=True, owner=project.owner) review_project.template = review_template else: raise ValidationError(template_serializer.errors) review_project.group_id = review_project.id review_project.save() return review_project
def get_template(self, obj, return_type='full'): template = None task_worker = None if return_type == 'full': template = TemplateSerializer(instance=obj.project.templates, many=True).data[0] else: template = \ TemplateSerializer(instance=obj.project.templates, many=True, fields=('id', 'template_items')).data[0] data = obj.data if 'task_worker' in self.context: task_worker = self.context['task_worker'] for item in template['template_items']: aux_attrib = item['aux_attributes'] if 'data_source' in aux_attrib and aux_attrib['data_source'] is not None and \ aux_attrib['data_source'] in data and 'src' in aux_attrib: aux_attrib['src'] = data[aux_attrib['data_source']] if 'question' in aux_attrib and 'data_source' in aux_attrib['question'] and \ aux_attrib['question']['data_source'] is not None and \ aux_attrib['question']['data_source'] in data.keys(): aux_attrib['question']['value'] = data[aux_attrib['question']['data_source']] if 'options' in aux_attrib: for option in aux_attrib['options']: if 'data_source' in option and option['data_source'] is not None and\ option['data_source'] in data.keys(): option['value'] = data[option['data_source']] if item['type'] == 'iframe': from django.conf import settings from hashids import Hashids identifier = Hashids(salt=settings.SECRET_KEY, min_length=settings.ID_HASH_MIN_LENGTH) if hasattr(task_worker, 'id'): item['identifier'] = identifier.encode(task_worker.id, task_worker.task.id, item['id']) else: item['identifier'] = 'READ_ONLY' if item['role'] == 'input' and task_worker is not None: for result in task_worker.task_worker_results.all(): if item['type'] == 'checkbox' and result.template_item_id == item['id']: item['aux_attributes']['options'] = result.result # might need to loop through options elif result.template_item_id == item['id']: item['answer'] = result.result template['template_items'] = sorted(template['template_items'], key=lambda k: k['position']) return template
def get_task_template(self, obj, return_type='full'): template = None if return_type == 'full': template = TemplateSerializer(instance=obj.project.templates, many=True).data[0] else: template = \ TemplateSerializer(instance=obj.project.templates, many=True, fields=('id', 'template_items')).data[0] data = ast.literal_eval(obj.data) for item in template['template_items']: aux_attrib = item['aux_attributes'] if 'data_source' in aux_attrib and aux_attrib['data_source'] is not None and \ aux_attrib['data_source'] in data and 'src' in aux_attrib: aux_attrib['src'] = data[aux_attrib['data_source']] if 'question' in aux_attrib and 'data_source' in aux_attrib['question'] and \ aux_attrib['question']['data_source'] is not None and \ aux_attrib['question']['data_source'] in data: aux_attrib['question']['value'] = data[aux_attrib['question']['data_source']] if 'options' in aux_attrib: for option in aux_attrib['options']: if 'data_source' in option and option['data_source'] is not None and option['data_source'] in data: option['value'] = data[option['data_source']] template['template_items'] = sorted(template['template_items'], key=lambda k: k['position']) return template
def create_revision(instance): models.Project.objects.filter(group_id=instance.group_id).update(status=models.Project.STATUS_PAUSED) template = TemplateSerializer.create_revision(instance=instance.template) # batch_files = copy.copy(instance.batch_files.all()) tasks = copy.copy(instance.tasks.all()) mturk_update_status.delay({'id': instance.id, 'status': models.Project.STATUS_PAUSED}) instance.pk = None instance.template = template instance.status = models.Project.STATUS_DRAFT instance.is_prototype = False instance.is_paid = False instance.save() # for f in batch_files: # models.ProjectBatchFile.objects.create(project=instance, batch_file=f) for t in tasks: t.pk = None t.project = instance TaskSerializer.bulk_create(data=tasks) return instance
class ProjectSerializer(DynamicFieldsModelSerializer): total_tasks = serializers.SerializerMethodField() age = serializers.SerializerMethodField() has_comments = serializers.SerializerMethodField() available_tasks = serializers.IntegerField(read_only=True) in_progress = serializers.IntegerField(read_only=True) completed = serializers.IntegerField(read_only=True) paid_count = serializers.IntegerField(read_only=True) awaiting_review = serializers.IntegerField(read_only=True) checked_out = serializers.IntegerField(read_only=True) returned = serializers.IntegerField(read_only=True) comments = serializers.SerializerMethodField() relaunch = serializers.SerializerMethodField() requester_rating = serializers.FloatField(read_only=True, required=False) raw_rating = serializers.IntegerField(read_only=True, required=False) # owner = UserSerializer(fields=('username', 'id'), read_only=True) requester_handle = serializers.CharField(read_only=True) batch_files = BatchFileSerializer( many=True, read_only=True, fields=('id', 'name', 'size', 'column_headers', 'format', 'number_of_rows', 'first_row', 'file')) template = TemplateSerializer(many=False, required=False) name = serializers.CharField(default='Untitled Project') status = serializers.IntegerField(default=models.Project.STATUS_DRAFT) file_id = serializers.IntegerField(write_only=True, allow_null=True, required=False) num_rows = serializers.IntegerField(write_only=True, allow_null=True, required=False) deadline = serializers.DateTimeField(required=False) revisions = serializers.SerializerMethodField() hash_id = serializers.SerializerMethodField() review_price = serializers.FloatField(required=False) amount_paid = serializers.FloatField(required=False) expected_payout_amount = serializers.FloatField(required=False) has_review = serializers.BooleanField(required=False) payout_available_by = serializers.DateTimeField(required=False) last_submitted_at = serializers.DateTimeField(required=False) template_id = serializers.IntegerField(required=False, allow_null=True) class Meta: model = models.Project fields = ('id', 'name', 'description', 'status', 'repetition', 'deadline', 'timeout', 'template', 'batch_files', 'deleted_at', 'created_at', 'updated_at', 'price', 'has_data_set', 'data_set_location', 'total_tasks', 'file_id', 'age', 'is_micro', 'is_prototype', 'has_review', 'task_time', 'allow_feedback', 'feedback_permissions', 'min_rating', 'has_comments', 'available_tasks', 'comments', 'num_rows', 'requester_rating', 'raw_rating', 'post_mturk', 'qualification', 'relaunch', 'group_id', 'revisions', 'hash_id', 'is_api_only', 'in_progress', 'awaiting_review', 'completed', 'review_price', 'returned', 'requester_handle', 'allow_price_per_task', 'task_price_field', 'discussion_link', 'aux_attributes', 'payout_available_by', 'paid_count', 'expected_payout_amount', 'amount_paid', 'template_id', 'checked_out', 'publish_at', 'published_at', 'template_id', 'enable_boomerang', 'last_submitted_at') read_only_fields = ('created_at', 'updated_at', 'deleted_at', 'has_comments', 'available_tasks', 'comments', 'template', 'is_api_only', 'discussion_link', 'aux_attributes', 'payout_available_by', 'paid_count', 'expected_payout_amount', 'amount_paid', 'published_at', 'last_submitted_at') validators = [ProjectValidator()] def to_representation(self, instance): data = super(ProjectSerializer, self).to_representation(instance) task_time = int(instance.task_time.total_seconds() / 60) if instance.task_time is not None else None timeout = int(instance.timeout.total_seconds() / 60) if instance.timeout is not None else None review_project = models.Project.objects.filter( parent_id=instance.group_id, is_review=True, deleted_at__isnull=True).first() if review_project is not None: review_price = review_project.price data.update({'review_price': review_price}) # data.update({'has_review': review_project is not None}) data.update({'task_time': task_time, 'timeout': timeout}) data.update({'price': instance.price}) return data def to_internal_value(self, data): if 'task_time' in data and data['task_time'] is not None: data['task_time'] = "00:{}:00".format(data['task_time']) if 'timeout' in data and data['timeout'] is not None: data['timeout'] = "00:{}:00".format(data['timeout']) return super(ProjectSerializer, self).to_internal_value(data) def create(self, with_defaults=True, **kwargs): template_initial = self.validated_data.pop( 'template') if 'template' in self.validated_data else None template_items = template_initial['items'] if template_initial else [] template = { "name": template_initial.get('name', generate_random_id()) if template_initial is not None else 't_' + generate_random_id(), "items": template_items } if 'post_mturk' in self.validated_data: self.validated_data.pop('post_mturk') template_serializer = TemplateSerializer(data=template) project = models.Project.objects.create(owner=kwargs['owner'], amount_due=0, post_mturk=False, **self.validated_data) if template_serializer.is_valid(): project_template = template_serializer.create( with_defaults=with_defaults, is_review=False, owner=kwargs['owner']) project.template = project_template else: raise ValidationError(template_serializer.errors) project.group_id = project.id # if not with_defaults: # project.status = models.Project.STATUS_IN_PROGRESS # project.published_at = timezone.now() # self.instance = project # if not project.is_paid: # self.pay(self.instance.price * self.instance.repetition) if with_defaults: self.create_task(project.id) project.save() # self.create_review(project=project, template_data=template) models.BoomerangLog.objects.create( object_id=project.group_id, min_rating=project.min_rating, rating_updated_at=project.rating_updated_at, reason='DEFAULT') return project @staticmethod def create_review(project, template_data, parent_review_project=None): project_name = 'Peer Review for ' + project.name review_project = models.Project.objects.create( name=project_name, owner=project.owner, parent=project, is_prototype=False, min_rating=1.99, post_mturk=False, timeout=project.timeout, is_review=True, deleted_at=timezone.now()) if parent_review_project is not None: review_project.price = parent_review_project.price template_serializer = TemplateSerializer(data=template_data) if template_serializer.is_valid(): review_template = template_serializer.create(with_defaults=False, is_review=True, owner=project.owner) review_project.template = review_template else: raise ValidationError(template_serializer.errors) review_project.group_id = review_project.id review_project.save() return review_project def update(self, *args, **kwargs): self.instance.name = self.validated_data.get('name', self.instance.name) self.instance.price = self.validated_data.get('price', self.instance.price) # review_project = models.Project.objects.filter(parent_id=self.instance.group_id, is_review=True).first() # has_review = self.validated_data.get('has_review', review_project.deleted_at is None) self.instance.timeout = self.validated_data.get( 'timeout', self.instance.timeout) # if review_project is not None: # review_project.price = self.validated_data.get('review_price', review_project.price) # review_project.timeout = self.instance.timeout # review_project.deleted_at = None if has_review else timezone.now() # review_project.save() self.instance.repetition = self.validated_data.get( 'repetition', self.instance.repetition) self.instance.deadline = self.validated_data.get( 'deadline', self.instance.deadline) self.instance.publish_at = self.validated_data.get( 'publish_at', self.instance.publish_at) # self.instance.post_mturk = self.validated_data.get('post_mturk', self.instance.post_mturk) self.instance.qualification = self.validated_data.get( 'qualification', self.instance.qualification) self.instance.allow_price_per_task = self.validated_data.get( 'allow_price_per_task', self.instance.allow_price_per_task) self.instance.task_price_field = self.validated_data.get( 'task_price_field', self.instance.task_price_field) self.instance.enable_boomerang = self.validated_data.get( 'enable_boomerang', self.instance.enable_boomerang) self.instance.template_id = self.validated_data.get( 'template_id', self.instance.template_id) self.instance.save() return self.instance def update_status(self, *args, **kwargs): status = self.initial_data.get('status', self.instance.status) validator = ProjectValidator() validator.set_context(self) validator.__call__(value={'status': status}) self.instance.status = status # mturk_update_status.delay({'id': self.instance.id, 'status': status}) self.instance.save() return self.instance @staticmethod def get_age(model): from crowdsourcing.utils import get_relative_time if model.status == models.Project.STATUS_DRAFT: return get_relative_time(model.updated_at) else: return get_relative_time(model.published_at) @staticmethod def get_total_tasks(obj): return obj.tasks.all().count() @staticmethod def get_has_comments(obj): return obj.comments.count() > 0 @staticmethod def get_comments(obj): if obj: comments = [] tasks = obj.tasks.all() for task in tasks: task_comments = task.comments.all() for task_comment in task_comments: comments.append(task_comment) serializer = TaskCommentSerializer(many=True, instance=comments, read_only=True) return serializer.data return [] @staticmethod def has_csv_linkage(items): if items.count() > 0: template_items = items.all() for item in template_items: attribs = item.aux_attributes if 'question' in attribs and 'data_source' in attribs['question'] and \ attribs['question']['data_source'] is not None: return True if 'options' in attribs and attribs['options'] is not None: for option in attribs['options']: if 'data_source' in option and option[ 'data_source'] is not None: return True return False @staticmethod def create_task(project_id): task_data = {"project": project_id, "data": {}} task_serializer = TaskSerializer(data=task_data) if task_serializer.is_valid(): task_serializer.create() else: raise ValidationError(task_serializer.errors) def fork(self, *args, **kwargs): template = self.instance.template template_items = copy.copy(template.items.all()) # batch_files = self.instance.batch_files.all() project = self.instance project.name += ' (copy)' project.status = models.Project.STATUS_DRAFT project.is_prototype = False project.parent_id = self.instance.id project.last_opened_at = None project.aux_attributes = {"sort_results_by": "worker_id"} project.amount_due = 0 project.min_rating = 3.0 project.is_paid = False project.previous_min_rating = 3.0 project.discussion_link = None project.publish_at = None template.pk = None template.save() project.template = template review_project = models.Project.objects.filter( parent_id=self.instance.group_id, is_review=True).first() for template_item in template_items: template_item.pk = None template_item.template = template template_item.save() TemplateItemSerializer.rebuild_tree(template) project.id = None project.save() project.group_id = project.id project.save() template = {"name": 't_' + generate_random_id(), "items": []} self.create_task(project.id) self.create_review(project=project, template_data=template, parent_review_project=review_project) return project @staticmethod def create_revision(instance): models.Project.objects.filter(group_id=instance.group_id).update( status=models.Project.STATUS_PAUSED) template = TemplateSerializer.create_revision( instance=instance.template) batch_files = copy.copy(instance.batch_files.all()) tasks = copy.copy(instance.tasks.all()) # mturk_update_status.delay({'id': instance.id, 'status': models.Project.STATUS_PAUSED}) instance.pk = None instance.template = template instance.status = models.Project.STATUS_DRAFT # instance.is_prototype = False instance.is_paid = False instance.save() for f in batch_files: models.ProjectBatchFile.objects.create(project=instance, batch_file=f) for t in tasks: t.pk = None t.project = instance TaskSerializer.bulk_create(data=tasks) return instance def publish(self, amount_due): self.instance.repetition = self.validated_data.get( 'repetition', self.instance.repetition) self.instance.published_at = timezone.now() review_project = models.Project.objects.filter( parent_id=self.instance.group_id, is_review=True, deleted_at__isnull=True).first() if review_project is not None and review_project.price is not None: review_project.status = models.Project.STATUS_IN_PROGRESS review_project.name = 'Peer Review for ' + self.instance.name review_project.published_at = timezone.now() review_project.save() status = models.Project.STATUS_IN_PROGRESS # if status != self.instance.status \ # and status in (models.Project.STATUS_PAUSED, models.Project.STATUS_IN_PROGRESS) and \ # self.instance.status in (models.Project.STATUS_PAUSED, models.Project.STATUS_IN_PROGRESS): # # mturk_update_status.delay({'id': self.instance.id, 'status': status}) self.instance.status = status self.instance.revised_at = timezone.now() if status == models.Project.STATUS_IN_PROGRESS and not self.instance.is_paid: self.pay(amount_due) self.instance.save() @staticmethod def get_relaunch(obj): """ Not used since we removed csv Args: obj: project instance Returns: """ previous_revision = models.Project.objects.prefetch_related('batch_files').filter(~Q(id=obj.id), group_id=obj.group_id) \ .order_by('-id').first() previous_batch_file = previous_revision.batch_files.first( ) if previous_revision else None batch_file = obj.batch_files.first() active_workers = models.TaskWorker.objects.active().filter( task__project__group_id=obj.group_id, task__exclude_at__isnull=True, status__in=[ models.TaskWorker.STATUS_IN_PROGRESS, models.TaskWorker.STATUS_SUBMITTED, models.TaskWorker.STATUS_RETURNED, models.TaskWorker.STATUS_ACCEPTED ]).count() same_file = (previous_batch_file is not None and batch_file is not None and previous_batch_file.id == batch_file.id) different_file = (previous_batch_file is not None and batch_file is None) or \ (previous_batch_file is None and batch_file is not None) if previous_revision is None or active_workers == 0: return { "is_forced": False, "ask_for_relaunch": False, "overlaps": False } elif (previous_batch_file is None and batch_file is None) or same_file: return { "is_forced": False, "ask_for_relaunch": True, "overlaps": True } elif different_file: return { "is_forced": True, "ask_for_relaunch": False, "overlaps": False } elif previous_batch_file.id != batch_file.id: return { "is_forced": False, "ask_for_relaunch": True, "overlaps": True } def pay(self, amount_due, *args, **kwargs): #Bir - Commenting out the stripe balance operation for user # requester_account = models.FinancialAccount.objects.get(owner_id=self.instance.owner_id, # type=models.FinancialAccount.TYPE_REQUESTER, # is_system=False).id # system_account = models.FinancialAccount.objects.get(is_system=True, # type=models.FinancialAccount.TYPE_ESCROW).id # transaction_data = { # 'sender': requester_account, # 'recipient': system_account, # 'amount': amount_due, # 'method': 'daemo', # 'sender_type': models.Transaction.TYPE_PROJECT_OWNER, # 'reference': 'P#' + str(self.instance.id) # } # if amount_due < 0: # transaction_data['sender'] = system_account # transaction_data['recipient'] = requester_account # transaction_data['amount'] = abs(amount_due) # # transaction_serializer = TransactionSerializer(data=transaction_data) # if transaction_serializer.is_valid(): # if amount_due != 0: # transaction_serializer.create() # self.instance.owner.stripe_customer.account_balance -= int(amount_due * 100)#Bir - Commenting out the stripe balance operation for user # self.instance.owner.stripe_customer.save()#Bir - Commenting out the stripe balance operation for user # self.instance.is_paid = True # self.instance.save() # else: # raise ValidationError('Error in payment') self.instance.is_paid = True self.instance.save() @staticmethod def get_revisions(obj): return models.Project.objects.active().filter( group_id=obj.group_id).order_by('id').values_list('id', flat=True) def reset_boomerang(self): update_project_boomerang.delay(self.instance.id) @staticmethod def get_hash_id(obj): return to_hash(obj.group_id) @staticmethod def _set_aux_attributes(project, price_data): if project.aux_attributes is None: project.aux_attributes = {} if project.price is not None: if not len(price_data): max_price = float(project.price) min_price = float(project.price) median_price = float(project.price) else: max_price = float(np.max(price_data)) min_price = float(np.min(price_data)) median_price = float(np.median(price_data)) project.aux_attributes.update({ "min_price": min_price, "max_price": max_price, "median_price": median_price }) project.save() def create_tasks(self, project_id, file_deleted): project = models.Project.objects.filter(pk=project_id).first() if project is None: return 'NOOP' previous_rev = models.Project.objects.prefetch_related('batch_files', 'tasks'). \ filter(~Q(id=project.id), group_id=project.group_id).order_by('-id').first() previous_batch_file = previous_rev.batch_files.first( ) if previous_rev else None models.Task.objects.filter(project=project).delete() if file_deleted: models.Task.objects.filter(project=project).delete() task_data = {"project_id": project_id, "data": {}} task = models.Task.objects.create(**task_data) if previous_batch_file is None and previous_rev is not None: task.group_id = previous_rev.tasks.all().first().group_id else: task.group_id = task.id task.save() self._set_aux_attributes(project, []) project.batch_files.all().delete() return 'SUCCESS' try: with transaction.atomic(): data = project.batch_files.first().parse_csv() task_obj = [] x = 0 previous_tasks = previous_rev.tasks.all().order_by( 'row_number') if previous_batch_file else [] previous_count = len(previous_tasks) for row in data: x += 1 hash_digest = hash_task(row) price = None if project.allow_price_per_task and project.task_price_field is not None: price = row.get(project.task_price_field) if not isinstance(price, (float, int, Decimal)): price = None t = models.Task(data=row, hash=hash_digest, project_id=int(project_id), row_number=x, price=price) if previous_batch_file is not None and x <= previous_count: if len( set(row.items()) ^ set(previous_tasks[x - 1].data.items())) == 0: t.group_id = previous_tasks[x - 1].group_id task_obj.append(t) models.Task.objects.bulk_create(task_obj) price_data = models.Task.objects.filter(project_id=project_id, price__isnull=False). \ values_list('price', flat=True) self._set_aux_attributes(project, price_data) models.Task.objects.filter(project_id=project_id, group_id__isnull=True) \ .update(group_id=F('id')) except Exception as e: raise e
class ProjectSerializer(DynamicFieldsModelSerializer): total_tasks = serializers.SerializerMethodField() age = serializers.SerializerMethodField() has_comments = serializers.SerializerMethodField() available_tasks = serializers.IntegerField(read_only=True) in_progress = serializers.IntegerField(read_only=True) completed = serializers.IntegerField(read_only=True) awaiting_review = serializers.IntegerField(read_only=True) comments = serializers.SerializerMethodField() relaunch = serializers.SerializerMethodField() requester_rating = serializers.FloatField(read_only=True, required=False) raw_rating = serializers.IntegerField(read_only=True, required=False) owner = UserSerializer(fields=('username', 'id'), read_only=True) batch_files = BatchFileSerializer(many=True, read_only=True, fields=('id', 'name', 'size', 'column_headers', 'format', 'number_of_rows')) template = TemplateSerializer(many=False, required=False) name = serializers.CharField(default='Untitled Project') status = serializers.IntegerField(default=models.Project.STATUS_DRAFT) file_id = serializers.IntegerField(write_only=True, allow_null=True, required=False) num_rows = serializers.IntegerField(write_only=True, allow_null=True, required=False) deadline = serializers.DateTimeField(required=False) revisions = serializers.SerializerMethodField() hash_id = serializers.SerializerMethodField() review_price = serializers.FloatField(required=False) has_review = serializers.BooleanField(required=False) class Meta: model = models.Project fields = ('id', 'name', 'owner', 'description', 'status', 'repetition', 'deadline', 'timeout', 'template', 'batch_files', 'deleted_at', 'created_at', 'updated_at', 'price', 'has_data_set', 'data_set_location', 'total_tasks', 'file_id', 'age', 'is_micro', 'is_prototype', 'has_review', 'task_time', 'allow_feedback', 'feedback_permissions', 'min_rating', 'has_comments', 'available_tasks', 'comments', 'num_rows', 'requester_rating', 'raw_rating', 'post_mturk', 'qualification', 'relaunch', 'group_id', 'revisions', 'hash_id', 'is_api_only', 'in_progress', 'awaiting_review', 'completed', 'review_price') read_only_fields = ( 'created_at', 'updated_at', 'deleted_at', 'owner', 'has_comments', 'available_tasks', 'comments', 'template', 'is_api_only') validators = [ProjectValidator()] def to_representation(self, instance): data = super(ProjectSerializer, self).to_representation(instance) task_time = int(instance.task_time.total_seconds() / 60) if instance.task_time is not None else None timeout = int(instance.timeout.total_seconds() / 60) if instance.timeout is not None else None review_project = models.Project.objects.filter(parent_id=instance.group_id, is_review=True, deleted_at__isnull=True).first() if review_project is not None: review_price = review_project.price data.update({'review_price': review_price}) data.update({'has_review': review_project is not None}) data.update({'task_time': task_time, 'timeout': timeout}) data.update({'price': instance.price}) return data def to_internal_value(self, data): if 'task_time' in data and data['task_time'] is not None: data['task_time'] = "00:{}:00".format(data['task_time']) if 'timeout' in data and data['timeout'] is not None: data['timeout'] = "00:{}:00".format(data['timeout']) return super(ProjectSerializer, self).to_internal_value(data) def create(self, with_defaults=True, **kwargs): template_initial = self.validated_data.pop('template') if 'template' in self.validated_data else None template_items = template_initial['items'] if template_initial else [] template = { "name": 't_' + generate_random_id(), "items": template_items } template_serializer = TemplateSerializer(data=template) self.validated_data.pop('post_mturk') project = models.Project.objects.create(owner=kwargs['owner'], post_mturk=True, amount_due=0, **self.validated_data) if template_serializer.is_valid(): project_template = template_serializer.create(with_defaults=with_defaults, is_review=False, owner=kwargs['owner']) project.template = project_template else: raise ValidationError(template_serializer.errors) project.group_id = project.id if not with_defaults: project.status = models.Project.STATUS_IN_PROGRESS project.published_at = timezone.now() # self.instance = project # if not project.is_paid: # self.pay(self.instance.price * self.instance.repetition) # self.create_task(project.id) project.save() self.create_review(project=project, template_data=template) models.BoomerangLog.objects.create(object_id=project.group_id, min_rating=project.min_rating, rating_updated_at=project.rating_updated_at, reason='DEFAULT') return project @staticmethod def create_review(project, template_data, parent_review_project=None): project_name = 'Peer Review for ' + project.name review_project = models.Project.objects.create(name=project_name, owner=project.owner, parent=project, is_prototype=False, min_rating=1.99, post_mturk=True, timeout=project.timeout, is_review=True, deleted_at=timezone.now()) if parent_review_project is not None: review_project.price = parent_review_project.price template_serializer = TemplateSerializer(data=template_data) if template_serializer.is_valid(): review_template = template_serializer.create(with_defaults=False, is_review=True, owner=project.owner) review_project.template = review_template else: raise ValidationError(template_serializer.errors) review_project.group_id = review_project.id review_project.save() return review_project def update(self, *args, **kwargs): self.instance.name = self.validated_data.get('name', self.instance.name) self.instance.price = self.validated_data.get('price', self.instance.price) review_project = models.Project.objects.filter(parent_id=self.instance.group_id, is_review=True).first() has_review = self.validated_data.get('has_review', review_project.deleted_at is None) self.instance.timeout = self.validated_data.get('timeout', self.instance.timeout) if review_project is not None: review_project.price = self.validated_data.get('review_price', review_project.price) review_project.timeout = self.instance.timeout review_project.deleted_at = None if has_review else timezone.now() review_project.save() self.instance.repetition = self.validated_data.get('repetition', self.instance.repetition) self.instance.deadline = self.validated_data.get('deadline', self.instance.deadline) self.instance.post_mturk = self.validated_data.get('post_mturk', self.instance.post_mturk) self.instance.qualification = self.validated_data.get('qualification', self.instance.qualification) self.instance.save() return self.instance def update_status(self, *args, **kwargs): status = self.initial_data.get('status', self.instance.status) validator = ProjectValidator() validator.set_context(self) validator.__call__(value={'status': status}) self.instance.status = status mturk_update_status.delay({'id': self.instance.id, 'status': status}) self.instance.save() return self.instance @staticmethod def get_age(model): from crowdsourcing.utils import get_relative_time if model.status == models.Project.STATUS_DRAFT: return get_relative_time(model.updated_at) else: return get_relative_time(model.published_at) @staticmethod def get_total_tasks(obj): return obj.tasks.all().count() @staticmethod def get_has_comments(obj): return obj.comments.count() > 0 @staticmethod def get_comments(obj): if obj: comments = [] tasks = obj.tasks.all() for task in tasks: task_comments = task.comments.all() for task_comment in task_comments: comments.append(task_comment) serializer = TaskCommentSerializer(many=True, instance=comments, read_only=True) return serializer.data return [] @staticmethod def has_csv_linkage(items): if items.count() > 0: template_items = items.all() for item in template_items: attribs = item.aux_attributes if 'question' in attribs and 'data_source' in attribs['question'] and \ attribs['question']['data_source'] is not None: return True if 'options' in attribs and attribs['options'] is not None: for option in attribs['options']: if 'data_source' in option and option['data_source'] is not None: return True return False @staticmethod def create_task(project_id): task_data = { "project": project_id, "data": {} } task_serializer = TaskSerializer(data=task_data) if task_serializer.is_valid(): task_serializer.create() else: raise ValidationError(task_serializer.errors) def fork(self, *args, **kwargs): template = self.instance.template template_items = copy.copy(template.items.all()) # batch_files = self.instance.batch_files.all() project = self.instance project.name += ' (copy)' project.status = models.Project.STATUS_DRAFT project.is_prototype = False project.parent_id = self.instance.id template.pk = None template.save() project.template = template review_project = models.Project.objects.filter(parent_id=self.instance.group_id, is_review=True).first() for template_item in template_items: template_item.pk = None template_item.template = template template_item.save() project.id = None project.save() project.group_id = project.id project.save() template = { "name": 't_' + generate_random_id(), "items": [] } self.create_review(project=project, template_data=template, parent_review_project=review_project) return project @staticmethod def create_revision(instance): models.Project.objects.filter(group_id=instance.group_id).update(status=models.Project.STATUS_PAUSED) template = TemplateSerializer.create_revision(instance=instance.template) # batch_files = copy.copy(instance.batch_files.all()) tasks = copy.copy(instance.tasks.all()) mturk_update_status.delay({'id': instance.id, 'status': models.Project.STATUS_PAUSED}) instance.pk = None instance.template = template instance.status = models.Project.STATUS_DRAFT instance.is_prototype = False instance.is_paid = False instance.save() # for f in batch_files: # models.ProjectBatchFile.objects.create(project=instance, batch_file=f) for t in tasks: t.pk = None t.project = instance TaskSerializer.bulk_create(data=tasks) return instance def publish(self, amount_due): self.instance.repetition = self.validated_data.get('repetition', self.instance.repetition) self.instance.published_at = timezone.now() review_project = models.Project.objects.filter(parent_id=self.instance.group_id, is_review=True, deleted_at__isnull=True).first() if review_project is not None and review_project.price is not None: review_project.status = models.Project.STATUS_IN_PROGRESS review_project.name = 'Peer Review for ' + self.instance.name review_project.published_at = timezone.now() review_project.save() status = models.Project.STATUS_IN_PROGRESS if status != self.instance.status \ and status in (models.Project.STATUS_PAUSED, models.Project.STATUS_IN_PROGRESS) and \ self.instance.status in (models.Project.STATUS_PAUSED, models.Project.STATUS_IN_PROGRESS): mturk_update_status.delay({'id': self.instance.id, 'status': status}) self.instance.status = status # TODO rm when mturk is removed if status == models.Project.STATUS_IN_PROGRESS and not self.instance.is_paid: # self.pay(amount_due) self.instance.save() @staticmethod def get_relaunch(obj): """ Not used since we removed csv Args: obj: project instance Returns: """ previous_revision = models.Project.objects.prefetch_related('batch_files').filter(~Q(id=obj.id), group_id=obj.group_id) \ .order_by('-id').first() previous_batch_file = previous_revision.batch_files.first() if previous_revision else None batch_file = obj.batch_files.first() active_workers = models.TaskWorker.objects.active().filter(task__project__group_id=obj.group_id, task__exclude_at__isnull=True, status__in=[models.TaskWorker.STATUS_IN_PROGRESS, models.TaskWorker.STATUS_SUBMITTED, models.TaskWorker.STATUS_RETURNED, models.TaskWorker.STATUS_ACCEPTED] ).count() same_file = ( previous_batch_file is not None and batch_file is not None and previous_batch_file.id == batch_file.id ) different_file = (previous_batch_file is not None and batch_file is None) or \ (previous_batch_file is None and batch_file is not None) if previous_revision is None or active_workers == 0: return { "is_forced": False, "ask_for_relaunch": False, "overlaps": False } elif (previous_batch_file is None and batch_file is None) or same_file: return { "is_forced": False, "ask_for_relaunch": True, "overlaps": True } elif different_file: return { "is_forced": True, "ask_for_relaunch": False, "overlaps": False } elif previous_batch_file.id != batch_file.id: return { "is_forced": False, "ask_for_relaunch": True, "overlaps": True } def pay(self, amount_due, *args, **kwargs): requester_account = models.FinancialAccount.objects.get(owner_id=self.instance.owner_id, type=models.FinancialAccount.TYPE_REQUESTER, is_system=False).id system_account = models.FinancialAccount.objects.get(is_system=True, type=models.FinancialAccount.TYPE_ESCROW).id transaction_data = { 'sender': requester_account, 'recipient': system_account, 'amount': amount_due, 'method': 'daemo', 'sender_type': models.Transaction.TYPE_PROJECT_OWNER, 'reference': 'P#' + str(self.instance.id) } if amount_due < 0: transaction_data['sender'] = system_account transaction_data['recipient'] = requester_account transaction_data['amount'] = abs(amount_due) transaction_serializer = TransactionSerializer(data=transaction_data) if transaction_serializer.is_valid(): if amount_due != 0: transaction_serializer.create() self.instance.is_paid = True self.instance.save() else: raise ValidationError('Error in payment') @staticmethod def get_revisions(obj): return models.Project.objects.active().filter(group_id=obj.group_id).order_by('id').values_list('id', flat=True) def reset_boomerang(self): update_project_boomerang.delay(self.instance.id) @staticmethod def get_hash_id(obj): return to_hash(obj.group_id)
def get_project_template(obj): return TemplateSerializer(instance=obj.task.project.template).data
class ProjectSerializer(DynamicFieldsModelSerializer): deleted = serializers.BooleanField(read_only=True) templates = TemplateSerializer(many=True, required=False) total_tasks = serializers.SerializerMethodField() file_id = serializers.IntegerField(write_only=True, allow_null=True, required=False) age = serializers.SerializerMethodField() has_comments = serializers.SerializerMethodField() available_tasks = serializers.SerializerMethodField() comments = serializers.SerializerMethodField() name = serializers.CharField(default='Untitled Project') status = serializers.IntegerField(default=1) owner = RequesterSerializer(fields=('alias', ), read_only=True) batch_files = BatchFileSerializer(many=True, read_only=True, fields=( 'id', 'name', 'size', 'column_headers', 'format', 'number_of_rows', )) num_rows = serializers.IntegerField(write_only=True, allow_null=True, required=False) requester_rating = serializers.FloatField(read_only=True, required=False) raw_rating = serializers.IntegerField(read_only=True, required=False) class Meta: model = models.Project fields = ( 'id', 'name', 'owner', 'description', 'status', 'repetition', 'timeout', 'templates', 'batch_files', 'deleted', 'created_timestamp', 'last_updated', 'price', 'has_data_set', 'data_set_location', 'total_tasks', 'file_id', 'age', 'is_micro', 'is_prototype', 'task_time', 'allow_feedback', 'feedback_permissions', 'min_rating', 'has_comments', 'available_tasks', 'comments', 'num_rows', 'requester_rating', 'raw_rating', ) read_only_fields = ( 'created_timestamp', 'last_updated', 'deleted', 'owner', 'has_comments', 'available_tasks', 'comments', 'templates', ) def create(self, **kwargs): project = models.Project.objects.create( deleted=False, owner=kwargs['owner'].requester) template = {"name": 't_' + generate_random_id()} template_serializer = TemplateSerializer(data=template) template = None if template_serializer.is_valid(): template = template_serializer.create(with_default=True, owner=kwargs['owner']) else: raise ValidationError(template_serializer.errors) models.ProjectTemplate.objects.get_or_create(project=project, template=template) return project def delete(self, instance): instance.deleted = True instance.save() return instance def get_age(self, model): from crowdsourcing.utils import get_time_delta if model.status == 1: return "Saved " + get_time_delta(model.last_updated) else: return "Posted " + get_time_delta(model.published_time) def get_total_tasks(self, obj): return obj.project_tasks.all().count() def get_has_comments(self, obj): return obj.projectcomment_project.count() > 0 def get_available_tasks(self, obj): available_task_count = models.Project.objects.values('id').raw( ''' select count(*) id from ( SELECT "crowdsourcing_task"."id" FROM "crowdsourcing_task" INNER JOIN "crowdsourcing_project" ON ("crowdsourcing_task"."project_id" = "crowdsourcing_project"."id") LEFT OUTER JOIN "crowdsourcing_taskworker" ON ("crowdsourcing_task"."id" = "crowdsourcing_taskworker"."task_id" and task_status not in (4,6)) WHERE ("crowdsourcing_task"."project_id" = %s AND NOT ( ("crowdsourcing_task"."id" IN (SELECT U1."task_id" AS Col1 FROM "crowdsourcing_taskworker" U1 WHERE U1."worker_id" = %s and U1.task_status<>6)))) GROUP BY "crowdsourcing_task"."id", "crowdsourcing_project"."repetition" HAVING "crowdsourcing_project"."repetition" > (COUNT("crowdsourcing_taskworker"."id"))) available_tasks ''', params=[ obj.id, self.context['request'].user.userprofile.worker.id ])[0].id return available_task_count def get_comments(self, obj): if obj: comments = [] tasks = obj.project_tasks.all() for task in tasks: task_comments = task.taskcomment_task.all() for task_comment in task_comments: comments.append(task_comment) serializer = TaskCommentSerializer(many=True, instance=comments, read_only=True) return serializer.data return [] def update(self, *args, **kwargs): status = self.validated_data.get('status', self.instance.status) num_rows = self.validated_data.get('num_rows', 0) if self.instance.status != status and status == 2: if self.instance.templates.all()[0].template_items.count() == 0: raise ValidationError('At least one template item is required') if self.instance.batch_files.count() == 0: task_data = { "project": self.instance.id, "status": 1, "data": {} } task_serializer = TaskSerializer(data=task_data) if task_serializer.is_valid(): task_serializer.create() else: raise ValidationError(task_serializer.errors) else: batch_file = self.instance.batch_files.first() data = batch_file.parse_csv() count = 0 for row in data: if count == num_rows: break task = {'project': self.instance.id, 'data': row} task_serializer = TaskSerializer(data=task) if task_serializer.is_valid(): task_serializer.create(**kwargs) count += 1 else: raise ValidationError(task_serializer.errors) self.instance.published_time = datetime.now() status += 1 self.instance.name = self.validated_data.get('name', self.instance.name) self.instance.price = self.validated_data.get('price', self.instance.price) self.instance.repetition = self.validated_data.get( 'repetition', self.instance.repetition) self.instance.status = status self.instance.save() return self.instance def fork(self, *args, **kwargs): templates = self.instance.templates.all() categories = self.instance.categories.all() batch_files = self.instance.batch_files.all() project = self.instance project.name = project.name + ' (copy)' project.status = 1 project.is_prototype = False project.parent = models.Project.objects.get(pk=self.instance.id) project.id = None project.save() for template in templates: project_template = models.ProjectTemplate(project=project, template=template) project_template.save() for category in categories: project_category = models.ProjectCategory(project=project, category=category) project_category.save() for batch_file in batch_files: project_batch_file = models.ProjectBatchFile(project=project, batch_file=batch_file) project_batch_file.save()
class ModuleSerializer(DynamicFieldsModelSerializer): deleted = serializers.BooleanField(read_only=True) template = TemplateSerializer(many=True) total_tasks = serializers.SerializerMethodField() file_id = serializers.IntegerField(write_only=True, allow_null=True) age = serializers.SerializerMethodField() has_comments = serializers.SerializerMethodField() available_tasks = serializers.SerializerMethodField() comments = serializers.SerializerMethodField() # comments = TaskCommentSerializer(many=True, source='module_tasks__task_workers__taskcomment_task', read_only=True) class Meta: model = models.Module fields = ('id', 'name', 'owner', 'project', 'description', 'status', 'repetition', 'module_timeout', 'deleted', 'template', 'created_timestamp', 'last_updated', 'price', 'has_data_set', 'data_set_location', 'total_tasks', 'file_id', 'age', 'is_micro', 'is_prototype', 'task_time', 'allow_feedback', 'feedback_permissions', 'min_rating', 'has_comments', 'available_tasks', 'comments') read_only_fields = ('created_timestamp', 'last_updated', 'deleted', 'owner', 'has_comments', 'available_tasks', 'comments') def create(self, **kwargs): templates = self.validated_data.pop('template') project = self.validated_data.pop('project') file_id = self.validated_data.pop('file_id') csv_data = [] if file_id is not None: uploaded_file = models.RequesterInputFile.objects.get(id=file_id) csv_data = uploaded_file.parse_csv() uploaded_file.delete() # module_tasks = self.validated_data.pop('module_tasks') module = models.Module.objects.create(deleted=False, project=project, owner=kwargs['owner'].requester, **self.validated_data) ''' experiment pool, default 6 ignored ''' experimental_models.ModulePool.objects.create(module=module, pool=6) for template in templates: template_items = template.pop('template_items') t = models.Template.objects.get_or_create(owner=kwargs['owner'], **template) models.ModuleTemplate.objects.get_or_create(module=module, template=t[0]) for item in template_items: models.TemplateItem.objects.get_or_create(template=t[0], **item) if module.has_data_set: for row in csv_data: task = {'module': module.id, 'data': json.dumps(row)} task_serializer = TaskSerializer(data=task) if task_serializer.is_valid(): task_serializer.create(**kwargs) else: raise ValidationError(task_serializer.errors) else: task = {'module': module.id, 'data': "{\"type\": \"static\"}"} task_serializer = TaskSerializer(data=task) if task_serializer.is_valid(): task_serializer.create(**kwargs) else: raise ValidationError(task_serializer.errors) return module def delete(self, instance): instance.deleted = True instance.save() return instance def get_age(self, model): from crowdsourcing.utils import get_time_delta delta = get_time_delta(model.created_timestamp) return "Posted " + delta def get_total_tasks(self, obj): return obj.module_tasks.all().count() def get_has_comments(self, obj): return obj.modulecomment_module.count() > 0 def get_available_tasks(self, obj): available_task_count = models.Module.objects.values('id').raw( ''' select count(*) id from ( SELECT "crowdsourcing_task"."id" FROM "crowdsourcing_task" INNER JOIN "crowdsourcing_module" ON ("crowdsourcing_task"."module_id" = "crowdsourcing_module"."id") LEFT OUTER JOIN "crowdsourcing_taskworker" ON ("crowdsourcing_task"."id" = "crowdsourcing_taskworker"."task_id" and task_status not in (4,6)) WHERE ("crowdsourcing_task"."module_id" = %s AND NOT ( ("crowdsourcing_task"."id" IN (SELECT U1."task_id" AS Col1 FROM "crowdsourcing_taskworker" U1 WHERE U1."worker_id" = %s and U1.task_status<>6)))) GROUP BY "crowdsourcing_task"."id", "crowdsourcing_module"."repetition" HAVING "crowdsourcing_module"."repetition" > (COUNT("crowdsourcing_taskworker"."id"))) available_tasks ''', params=[ obj.id, self.context['request'].user.userprofile.worker.id ])[0].id return available_task_count def get_comments(self, obj): if obj: comments = [] tasks = obj.module_tasks.all() for task in tasks: task_comments = task.taskcomment_task.all() for task_comment in task_comments: comments.append(task_comment) serializer = TaskCommentSerializer(many=True, instance=comments, read_only=True) return serializer.data return []