class BlindingSet(db.Model): id = db.Column(db.Integer, primary_key=True) name = db.Column(db.String(100), nullable=False) study_id = db.Column(db.Integer, db.ForeignKey(Study.id), nullable=False) study = db.relationship(Study, backref=db.backref("blinding_sets")) def __repr__(self): return self.name def __lt__(self, other): return self.name < other.name def get_blind_ids(self, unblind_id, user): result = [] for bt in self.blinding_types: blind_id = bt.get_blind_id(unblind_id, user) if blind_id: result.append(blind_id) return result def get_unblind_id(self, blind_id): for bt in self.blinding_types: unblind_id = bt.get_unblind_id(blind_id) if unblind_id: return unblind_id
class ParticipantIdentifierSource(db.Model): __tablename__ = 'participant_identifier_source' id = db.Column(db.Integer, primary_key=True) linked_minimum_patient_identifier_source_id = db.Column( db.Integer, db.ForeignKey("participant_identifier_source.id"), nullable=True) type = db.Column(db.String(100), nullable=False) study_id = db.Column(db.Integer, db.ForeignKey(Study.id), nullable=True) study = db.relationship( Study, backref=db.backref("participant_identifier_sources")) __mapper_args__ = { 'polymorphic_identity': 'participant_identifier_source', 'polymorphic_on': type, } last_updated_datetime = db.Column(db.DateTime, nullable=False, default=datetime.utcnow) last_updated_by_user_id = db.Column(db.Integer, db.ForeignKey(User.id)) last_updated_by_user = db.relationship(User) identifiers = db.relationship( "ParticipantIdentifier", secondary=participant_identifiers__participant_identifier_sources, back_populates="sources", collection_class=set) def __str__(self): return f"{self.type.name}: {self.identifier}"
class Blinding(db.Model): id = db.Column(db.Integer, primary_key=True) unblind_id = db.Column(db.String(100), nullable=False) blinding_type_id = db.Column(db.Integer, db.ForeignKey(BlindingType.id), nullable=False) blinding_type = db.relationship(BlindingType, backref=db.backref("blindings")) pseudo_random_id_id = db.Column(db.Integer, db.ForeignKey(PseudoRandomId.id), nullable=False) pseudo_random_id = db.relationship(PseudoRandomId) last_updated_datetime = db.Column(db.DateTime, nullable=False, default=datetime.utcnow) last_updated_by_user_id = db.Column(db.Integer, db.ForeignKey(User.id)) last_updated_by_user = db.relationship(User) def __repr__(self): return '; '.join([ self.blinding_type.name, self.unblind_id, self.pseudo_random_id.full_code, ]) def __lt__(self, other): return self.name < other.name
class DemographicsRequestColumn(db.Model): id = db.Column(db.Integer, primary_key=True) demographics_request_id = db.Column(db.Integer, db.ForeignKey(DemographicsRequest.id)) demographics_request = db.relationship( DemographicsRequest, foreign_keys=[demographics_request_id], backref=db.backref("columns")) name = db.Column(db.String(500)) last_updated_datetime = db.Column(db.DateTime, nullable=False, default=datetime.utcnow) last_updated_by_user_id = db.Column(db.Integer, db.ForeignKey(User.id)) last_updated_by_user = db.relationship( User, foreign_keys=[last_updated_by_user_id]) def __lt__(self, other): return self.name < other.name def __repr__(self): fields = '; '.join([ f'{key}="{value}"' for key, value in self.__dict__.items() if key[0] != '_' ]) return f'[{type(self).__name__}({fields})]'
class Study(db.Model): id = db.Column(db.Integer(), primary_key=True) name = db.Column(db.String(100)) date_created = db.Column(db.DateTime, nullable=False, default=datetime.utcnow) allow_duplicate_study_number = db.Column(db.Boolean, nullable=False, default=False) allow_empty_study_number = db.Column(db.Boolean, nullable=False, default=False) study_number_format = db.Column(db.String(50)) study_number_name = db.Column(db.String(100)) field_group_id = db.Column(db.Integer(), db.ForeignKey(FieldGroup.id)) field_group = db.relationship(FieldGroup, backref=db.backref("study")) owners = db.relationship( User, secondary=studies_owners, backref=db.backref("owned_studies", lazy="dynamic"), ) collaborators = db.relationship( User, secondary=studies_collaborators, backref=db.backref("collaborator_studies", lazy="dynamic"), ) def __str__(self): return self.name @property def upload_count(self): return len([u for u in self.uploads if not u.deleted]) def upload_count_for_user(self, user): return len( [u for u in self.uploads if not u.deleted and u.uploader == user]) @property def outstanding_upload_count(self): return len( [u for u in self.uploads if not u.deleted and not u.completed]) def get_study_number_name(self): return self.study_number_name or 'Study Number'
class ApiKey(db.Model): id = db.Column(db.Integer, primary_key=True) key = db.Column(GUID, nullable=False, default=uuid.uuid4) user_id = db.Column(db.Integer, db.ForeignKey(User.id), nullable=False) user = db.relationship(User, backref=db.backref("api_key", uselist=False)) def __repr__(self): return f'API Key for User {self.user.full_name}'
class User(BaseUser): __table_args__ = {'extend_existing': True} studies = db.relationship("Study", secondary=users_studies, backref=db.backref("users", lazy="joined")) def __str__(self): return 'user'
class LabelPack(db.Model): id = db.Column(db.Integer, primary_key=True) type = db.Column(db.String(100), nullable=False) study_id = db.Column(db.Integer, db.ForeignKey(Study.id)) study = db.relationship(Study, backref=db.backref("label_packs")) __mapper_args__ = { "polymorphic_identity": "Pack", "polymorphic_on": type, } def user_defined_participant_id(self): return False def allow_batch_printing(self): return True @property def name(self): return re.sub('([a-z])([A-Z])', r'\1 \2', self.__class__.__name__) def print(self, count): for _ in range(count): current_app.logger.info( f'Printing label for study {self.study.name}') self._do_print() db.session.commit() time.sleep(current_app.config['PRINTING_SET_SLEEP']) def save_participant_id(self, participant_id): pit = ParticipantIdentifierType.get_study_participant_id() pi = ParticipantIdentifier.query.filter_by( participant_identifier_type_id=pit.id, identifier=participant_id, ).one_or_none() if pi is None: db.session.add( ParticipantIdentifier( participant_identifier_type_id=pit.id, identifier=participant_id, last_updated_by_user_id=current_user.id, )) def _do_print(self, participant_id=None): pass def __repr__(self): return self.name def __lt__(self, other): return self.name < other.name
class StudyParticipant(db.Model): id = db.Column(db.Integer, primary_key=True) study_id = db.Column(db.Integer, db.ForeignKey(Study.id), nullable=False) study = db.relationship(Study, backref=db.backref("participants")) last_updated_datetime = db.Column(db.DateTime, nullable=False, default=datetime.utcnow) last_updated_by_user_id = db.Column(db.Integer, db.ForeignKey(User.id)) last_updated_by_user = db.relationship(User)
class UploadData(db.Model): id = db.Column(db.Integer(), primary_key=True) upload_id = db.Column(db.Integer(), db.ForeignKey(Upload.id)) upload = db.relationship(Upload, backref=db.backref("data")) field_id = db.Column(db.Integer(), db.ForeignKey(Field.id)) field = db.relationship(Field) value = db.Column(db.String) def __repr__(self): items = ("%s = %r" % (k, v) for k, v in self.__dict__.items()) return "<%s: {%s}>" % (self.__class__.__name__, ', '.join(items))
class Upload(db.Model): id = db.Column(db.Integer(), primary_key=True) study_id = db.Column(db.Integer(), db.ForeignKey(Study.id)) study_number = db.Column(db.String(20)) uploader_id = db.Column(db.Integer(), db.ForeignKey(User.id)) date_created = db.Column(db.DateTime, nullable=False, default=datetime.utcnow) study = db.relationship(Study, backref=db.backref("uploads")) uploader = db.relationship(User) completed = db.Column(db.Boolean, default=0) deleted = db.Column(db.Boolean, default=0)
class Publication(AuditMixin, CommonMixin, db.Model): id = db.Column(db.Integer(), primary_key=True) pm_id = db.Column(db.Integer()) journal = db.Column(db.String(200)) published_date = db.Column(db.Date) title = db.Column(db.UnicodeText()) academics = db.relationship("Academic", secondary=academics_publications, collection_class=set, backref=db.backref("publications", lazy="joined"))
class DemographicsRequestDataMessage(db.Model): id = db.Column(db.Integer, primary_key=True) demographics_request_data_id = db.Column( db.Integer, db.ForeignKey(DemographicsRequestData.id)) demographics_request_data = db.relationship(DemographicsRequestData, backref=db.backref("messages")) type = db.Column(db.String) source = db.Column(db.String) scope = db.Column(db.String) message = db.Column(db.String) created_datetime = db.Column(db.DateTime, nullable=False, default=datetime.utcnow) @property def is_error(self): return self.type == 'error'
class BlindingType(db.Model): id = db.Column(db.Integer, primary_key=True) name = db.Column(db.String(100), nullable=False) blinding_set_id = db.Column(db.Integer, db.ForeignKey(BlindingSet.id), nullable=False) blinding_set = db.relationship(BlindingSet, backref=db.backref("blinding_types")) pseudo_random_id_provider_id = db.Column(db.Integer, db.ForeignKey( PseudoRandomIdProvider.id), nullable=False) pseudo_random_id_provider = db.relationship(PseudoRandomIdProvider) deleted = db.Column(db.Boolean, nullable=False, default=False) duplicate_number = db.Column(db.Integer, default=0, nullable=False) def __repr__(self): return self.name def __lt__(self, other): return self.name < other.name def get_blind_id(self, unblind_id, user): blinding = (Blinding.query.filter_by( blinding_type_id=self.id).filter_by(unblind_id=unblind_id).first()) if not blinding: pseudo_random_id = self.pseudo_random_id_provider.allocate_id(user) blinding = Blinding( unblind_id=unblind_id, blinding_type=self, pseudo_random_id=pseudo_random_id, last_updated_by_user=user, ) return blinding def get_unblind_id(self, blind_id): blinding = (Blinding.query.filter_by( blinding_type_id=self.id).join(PseudoRandomId).filter_by( full_code=blind_id).first()) return blinding
class UploadFile(db.Model): id = db.Column(db.Integer(), primary_key=True) upload_id = db.Column(db.Integer(), db.ForeignKey(Upload.id)) upload = db.relationship(Upload, backref=db.backref("files")) field_id = db.Column(db.Integer(), db.ForeignKey(Field.id)) field = db.relationship(Field) filename = db.Column(db.String(500)) def get_download_filename(self): if len(self.field.download_filename_format or '') == 0: return self.filename else: return self.field.download_filename_format.format( file=self) + os.path.splitext(self.filename)[-1] def filepath(self): return os.path.join( secure_filename("{}_{}".format(self.upload.study.id, self.upload.study.name)), secure_filename("{}_{}_{}".format(self.id, self.upload.study_number, self.filename)), )
class DemographicsRequestData(db.Model): id = db.Column(db.Integer, primary_key=True) demographics_request_id = db.Column(db.Integer, db.ForeignKey(DemographicsRequest.id)) row_number = db.Column(db.Integer, nullable=False) demographics_request = db.relationship(DemographicsRequest, backref=db.backref("data")) response = db.relationship("DemographicsRequestDataResponse", uselist=False, back_populates="demographics_request_data") pmi_data = db.relationship("DemographicsRequestPmiData", uselist=False, back_populates="demographics_request_data") nhs_number = db.Column(db.String) uhl_system_number = db.Column(db.String) family_name = db.Column(db.String) given_name = db.Column(db.String) gender = db.Column(db.String) dob = db.Column(db.String) postcode = db.Column(db.String) created_datetime = db.Column(db.DateTime, nullable=False, default=datetime.utcnow) processed_datetime = db.Column(db.DateTime) pmi_pre_processed_datetime = db.Column(db.DateTime) pmi_post_processed_datetime = db.Column(db.DateTime) @property def processed(self): return self.processed_datetime is not None @property def pmi_pre_processed(self): return self.pmi_pre_processed_datetime is not None @property def pmi_post_processed(self): return self.pmi_post_processed_datetime is not None @property def has_error(self): return any(m.is_error for m in self.messages) @property def has_error(self): return any(m.is_error for m in self.messages) @property def confidence(self): if self.response is None: return 0 parts = [] if self.nhs_number: parts.append(similarity(self.nhs_number, self.response.nhs_number)) if self.family_name: parts.append(similarity(self.family_name, self.response.lastname)) if self.given_name: parts.append(similarity(self.given_name, self.response.forename)) if self.postcode: parts.append(similarity(self.postcode, self.response.postcode)) return round(sum(parts) / len(parts), 2) def __repr__(self): fields = '; '.join([ f'{key}="{value}"' for key, value in self.__dict__.items() if key[0] != '_' ]) return f'[{type(self).__name__}({fields})]'
class DemographicsRequest(db.Model): id = db.Column(db.Integer, primary_key=True) created_datetime = db.Column(db.DateTime, nullable=False, default=datetime.utcnow) filename = db.Column(db.String(500)) extension = db.Column(db.String(100), nullable=False) owner_user_id = db.Column(db.Integer, db.ForeignKey(User.id)) owner = db.relationship(User, foreign_keys=[owner_user_id], backref=db.backref("demographic_requests")) submitted_datetime = db.Column(db.DateTime) deleted_datetime = db.Column(db.DateTime) paused_datetime = db.Column(db.DateTime) data_extracted_datetime = db.Column(db.DateTime) pmi_data_pre_completed_datetime = db.Column(db.DateTime) pmi_data_post_completed_datetime = db.Column(db.DateTime) lookup_completed_datetime = db.Column(db.DateTime) result_created_datetime = db.Column(db.DateTime) result_downloaded_datetime = db.Column(db.DateTime) error_datetime = db.Column(db.DateTime) error_message = db.Column(db.Text) last_updated_datetime = db.Column(db.DateTime, nullable=False, default=datetime.utcnow) last_updated_by_user_id = db.Column(db.Integer, db.ForeignKey(User.id)) last_updated_by_user = db.relationship( User, foreign_keys=[last_updated_by_user_id]) column_definition = db.relationship("DemographicsRequestColumnDefinition", uselist=False, back_populates="demographics_request") skip_pmi = db.Column(db.Boolean, default=False) __mapper_args__ = { "polymorphic_on": extension, } @property def filepath(self): return os.path.join( current_app.config["FILE_UPLOAD_DIRECTORY"], secure_filename( "{}_{}".format(self.last_updated_by_user.id, self.last_updated_by_user.full_name)), secure_filename("{}_{}".format(self.id, self.filename)), ) @property def result_filename(self): return secure_filename("{}_result_{}".format(self.id, self.filename)) @property def result_filepath(self): return os.path.join( current_app.config["FILE_UPLOAD_DIRECTORY"], secure_filename( "{}_{}".format(self.last_updated_by_user.id, self.last_updated_by_user.full_name)), self.result_filename, ) def __lt__(self, other): return self.created_datetime < other.created_datetime @property def data_extracted(self): return self.data_extracted_datetime is not None @property def paused(self): return self.paused_datetime is not None @property def columns_defined(self): return self.column_definition and self.column_definition.is_valid @property def awaiting_submission(self): return self.columns_defined and self.submitted_datetime is None @property def submitted(self): return self.submitted_datetime is not None @property def deleted(self): return self.deleted_datetime is not None @property def result_created(self): return self.result_created_datetime is not None @property def pmi_data_pre_completed(self): return self.pmi_data_pre_completed_datetime is not None @property def pmi_data_post_completed(self): return self.pmi_data_post_completed_datetime is not None @property def result_downloaded(self): return self.result_downloaded_datetime is not None @property def lookup_completed(self): return self.lookup_completed_datetime is not None @property def in_error(self): return self.error_datetime is not None @property def requires_column_definition(self): return not self.deleted and not self.submitted and not self.in_error @property def requires_submission(self): return not self.deleted and self.awaiting_submission and not self.in_error @property def can_be_resubmitted(self): return not self.deleted and self.submitted and not self.result_created and not self.in_error @property def can_be_paused(self): return not self.deleted and self.submitted and not self.result_created and not self.paused and not self.in_error @property def can_be_downloaded(self): return not self.deleted and self.result_created and not self.in_error @property def can_be_deleted(self): return not self.deleted @property def status(self): if self.deleted: return 'Deleted' elif self.in_error: return 'Error' elif self.paused: return 'Paused' if not self.columns_defined: return 'Uploaded' elif not self.submitted: return 'Awaiting Submission' elif not self.data_extracted: return f'Extracting Data' elif not self.pmi_data_pre_completed and not self.skip_pmi: return f'Fetching PMI details {self.prepmi_count} of {self.data_count} before spine lookup' elif not self.lookup_completed: return f'Fetching Demographics {self.fetched_count} of {self.data_count}' elif not self.pmi_data_post_completed and not self.skip_pmi: return f'Fetching PMI details {self.postpmi_count} of {self.data_count} after spine lookup' elif not self.result_created: return 'Processing Demographics' elif not self.result_downloaded: return 'Ready to Download' else: return 'Downloaded' @property def data_count(self): return DemographicsRequestData.query.filter( DemographicsRequestData.demographics_request_id == self.id).count() @property def fetched_count(self): return DemographicsRequestData.query.filter( DemographicsRequestData.demographics_request_id == self.id).filter( DemographicsRequestData.processed_datetime.isnot( None)).count() @property def prepmi_count(self): return DemographicsRequestData.query.filter( DemographicsRequestData.demographics_request_id == self.id).filter( DemographicsRequestData.pmi_pre_processed_datetime.isnot( None)).count() @property def postpmi_count(self): return DemographicsRequestData.query.filter( DemographicsRequestData.demographics_request_id == self.id).filter( DemographicsRequestData.pmi_post_processed_datetime.isnot( None)).count() def get_most_likely_uhl_system_number_column_id(self): return self._get_most_likely_column_id('(uhl|\bs).*(number|no|)') def get_most_likely_nhs_number_column_id(self): return self._get_most_likely_column_id('nhs.*(number|no|)') def get_most_likely_family_name_column_id(self): return self._get_most_likely_column_id('(surname|(family|last).*name)') def get_most_likely_given_name_column_id(self): return self._get_most_likely_column_id('(first|given|fore).*name|name') def get_most_likely_gender_column_id(self): return self._get_most_likely_column_id('(gender|sex)') def get_most_likely_dob_column_id(self): return self._get_most_likely_column_id('(dob|date.*birth|birth.*date)') def get_most_likely_postcode_column_id(self): return self._get_most_likely_column_id('(post.*code)') def _get_most_likely_column_id(self, regular_expression): repat = re.compile(regular_expression, re.IGNORECASE) ids = (c.id for c in self.columns if re.search(repat, c.name)) return next(ids, 0) def set_error(self, message): self.error_datetime = datetime.utcnow() self.error_message = message