class Activity(db.Model):
class ActivityType(enum.Enum):
created = "Created"
changed_name = "Changed name"
changed_description = "Changed Description"
added_version = "Added version"
started_log = "Log started"
__tablename__ = "activities"
id: str = db.Column(GUID, primary_key=True, default=generate_uuid)
user_id: str = db.Column(GUID, db.ForeignKey("users.id"))
user: User = db.relationship("User", backref=__tablename__)
dataset_id: str = db.Column(GUID, db.ForeignKey("datasets.id"))
dataset: Dataset = db.relationship("Dataset", backref=__tablename__)
# We would want the type of change and the comments associated
type: ActivityType = db.Column(db.Enum(ActivityType), nullable=False)
timestamp: datetime.datetime = db.Column(db.DateTime,
default=datetime.datetime.utcnow)
comments: str = db.Column(db.Text)
__mapper_args__ = {"polymorphic_on": type}
class UploadSessionFile(db.Model):
__tablename__ = "upload_session_files"
id: str = db.Column(GUID, primary_key=True, default=generate_uuid)
session_id: str = db.Column(GUID, db.ForeignKey("upload_sessions.id"))
session: UploadSession = db.relationship("UploadSession",
backref=__tablename__)
# filename submitted by user
filename: str = db.Column(db.Text)
encoding: str = db.Column(db.Text)
initial_filetype: InitialFileType = db.Column(db.Enum(InitialFileType))
initial_s3_key: str = db.Column(db.Text)
converted_filetype: S3DataFile.DataFileFormat = db.Column(
db.Enum(S3DataFile.DataFileFormat))
converted_s3_key: str = db.Column(db.Text)
compressed_s3_key: str = db.Column(db.Text)
s3_bucket: str = db.Column(db.Text)
gcs_path: str = db.Column(db.Text)
generation_id: str = db.Column(db.Text)
short_summary: str = db.Column(db.Text)
long_summary: str = db.Column(db.Text)
column_types_as_json: Dict[str, str] = db.Column(db.JSON)
original_file_sha256: str = db.Column(db.Text)
original_file_md5: str = db.Column(db.Text)
data_file_id: str = db.Column(GUID, db.ForeignKey("datafiles.id"))
data_file: DataFile = db.relationship(
"DataFile",
uselist=False,
foreign_keys="UploadSessionFile.data_file_id")
class ProvenanceNode(db.Model):
__tablename__ = "provenance_nodes"
class NodeType(enum.Enum):
Dataset = "dataset"
External = "external"
Process = "process"
node_id = db.Column(GUID, primary_key=True, default=generate_uuid)
graph_id = db.Column(GUID, db.ForeignKey("provenance_graphs.graph_id"))
graph = db.relationship("ProvenanceGraph", backref=__tablename__)
datafile_id = db.Column(GUID, db.ForeignKey("datafiles.id"), nullable=True)
datafile = db.relationship("DataFile", backref=__tablename__)
label = db.Column(db.Text)
type = db.Column(db.Enum(NodeType))
class ConversionCache(db.Model):
__tablename__ = "conversion_cache"
id: str = db.Column(GUID, primary_key=True, default=generate_uuid)
dataset_version_id: str = db.Column(GUID,
db.ForeignKey("dataset_versions.id"))
datafile_name: str = db.Column(db.String(80))
format: str = db.Column(db.String(80))
status: str = db.Column(db.Text)
task_id: str = db.Column(db.Text)
urls_as_json: str = db.Column(db.Text)
state: ConversionEntryState = db.Column(db.Enum(ConversionEntryState))
class Folder(Entry):
# Enum Folder types
# TODO: Could be a good idea to transform these enums into Classes. So they can have different behaviors if needed
class FolderType(enum.Enum):
home = "home"
trash = "trash"
folder = "folder"
__tablename__ = "folders"
# TODO: Instead of using a string 'entry.id', can we use Entry.id?
id: str = db.Column(GUID, db.ForeignKey("entries.id"), primary_key=True)
folder_type: FolderType = db.Column(db.Enum(FolderType))
# TODO: This should be a set, not a list.
entries: List[Entry] = db.relationship(
"Entry", secondary=folder_entry_association_table, backref="parents")
__mapper_args__ = {"polymorphic_identity": "Folder"}
def __repr__(self):
return "Folder name: {} and id: {}".format(self.name, self.id)
class DatasetVersion(Entry):
class DatasetVersionState(enum.Enum):
approved = "Approved"
deprecated = "Deprecated"
deleted = "Deleted"
# TODO: Missing the permaname of the DatasetVersion
__tablename__ = "dataset_versions"
id: str = db.Column(GUID, db.ForeignKey("entries.id"), primary_key=True)
dataset_id: str = db.Column(GUID, db.ForeignKey("datasets.id"))
dataset: Dataset = db.relationship(
"Dataset",
foreign_keys=[dataset_id],
backref=db.backref(__tablename__),
single_parent=True,
cascade="all, delete-orphan",
)
# Filled out by the server
version: int = db.Column(db.Integer)
# State of the version
state: DatasetVersionState = db.Column(
db.Enum(DatasetVersionState), default=DatasetVersionState.approved)
# Reason for the state of the version. Should be empty if approved
reason_state: str = db.Column(db.Text)
changes_description: str = db.Column(db.Text)
__table_args__ = (UniqueConstraint("dataset_id", "version"), )
# TODO: See how to manage the status (persist.py)
__mapper_args__ = {"polymorphic_identity": "DatasetVersion"}
class S3DataFile(DataFile):
# IMPORTANT: Need to sync with frontend for each changes
class DataFileFormat(enum.Enum):
Raw = "Raw"
HDF5 = "HDF5"
Columnar = "Columnar"
format: DataFileFormat = db.Column(db.Enum(DataFileFormat))
encoding = db.Column(db.Text)
s3_bucket: str = db.Column(db.Text)
s3_key: str = db.Column(db.Text)
compressed_s3_key: str = db.Column(db.Text)
short_summary: str = db.Column(db.Text)
long_summary: str = db.Column(db.Text)
column_types_as_json: Dict[str, str] = db.Column(db.JSON)
original_file_sha256: str = db.Column(db.Text)
original_file_md5: str = db.Column(db.Text)
__mapper_args__ = {"polymorphic_identity": "s3"}
@property
def underlying_file_id(self):
return None