def update(self, session: Session, title: str, project_ids: ProjectIDListT,
provider_user: Any, contact_user: Any, citation: str,
abstract: str, description: str, creator_users: List[Any],
associate_users: List[Any], creator_orgs: List[Any],
associate_orgs: List[Any]):
project_ids.sort()
# TODO: projects update using given list
assert project_ids == self.project_ids, "Cannot update composing projects yet"
# Redo sanity check & aggregation as underlying projects might have changed
self._add_composing_projects(session, project_ids)
coll_id = self._collection.id
# Simple fields update
self._collection.title = title
self._collection.citation = citation
self._collection.abstract = abstract
self._collection.description = description
# Copy provider user id
if provider_user is not None:
self._collection.provider_user_id = provider_user.id
# Copy contact user id
if contact_user is not None:
self._collection.contact_user_id = contact_user.id
# Dispatch members by role
by_role = {
COLLECTION_ROLE_DATA_CREATOR: creator_users,
COLLECTION_ROLE_ASSOCIATED_PERSON: associate_users
}
# Remove all to avoid diff-ing
session.query(CollectionUserRole). \
filter(CollectionUserRole.collection_id == coll_id).delete()
# Add all
for a_role, a_user_list in by_role.items():
for a_user in a_user_list:
session.add(
CollectionUserRole(collection_id=coll_id,
user_id=a_user.id,
role=a_role))
# Dispatch orgs by role
by_role_org = {
COLLECTION_ROLE_DATA_CREATOR: creator_orgs,
COLLECTION_ROLE_ASSOCIATED_PERSON: associate_orgs
}
# Remove all to avoid diff-ing
session.query(CollectionOrgaRole). \
filter(CollectionOrgaRole.collection_id == coll_id).delete()
# Add all
for a_role, an_org_list in by_role_org.items():
for an_org in an_org_list:
session.add(
CollectionOrgaRole(collection_id=coll_id,
organisation=an_org,
role=a_role))
session.commit()
def create_job(cls, session: Session, user_id: UserIDT, job_type: str,
args: Dict) -> Job:
job = Job()
job.state = DBJobStateEnum.Pending
job.progress_msg = cls.PENDING_MESSAGE
job.creation_date = job.updated_on = datetime.now()
job.type = job_type
job.owner_id = user_id
job.params = json_dumps(args)
job.inside = job.reply = json_dumps({})
job.messages = json_dumps([])
session.add(job)
session.commit()
return job
def delete(session: Session, coll_id: CollectionIDT):
"""
Completely remove the collection.
Being just a set of project references, the pointed-at projects are not impacted.
"""
# Remove links first
session.query(CollectionProject). \
filter(CollectionProject.collection_id == coll_id).delete()
session.query(CollectionUserRole). \
filter(CollectionUserRole.collection_id == coll_id).delete()
session.query(CollectionOrgaRole). \
filter(CollectionOrgaRole.collection_id == coll_id).delete()
# Remove collection
session.query(Collection). \
filter(Collection.id == coll_id).delete()
session.commit()
def get_acquisitions(cls, session: Session,
sample: Sample) -> List[Acquisition]:
""" Get acquisitions for the sample """
qry: Query = session.query(Acquisition)
qry = qry.join(Sample)
qry = qry.filter(Sample.sampleid == sample.sampleid)
return qry.all()
def _add_composing_projects(self, session: Session, project_ids: ProjectIDListT):
"""
Add the given projects into DB, doing sanity checks.
"""
qry: Query = session.query(Project).filter(Project.projid.in_(project_ids))
qry = qry.join(Sample, Project.all_samples).options(contains_eager(Project.all_samples))
db_projects = qry.all()
assert len(db_projects) == len(project_ids)
# Loop on projects, adding them and collecting aggregated data
prj_licenses: Set[LicenseEnum] = set()
samples_per_project: Dict[str, Project] = {}
problems: List[str] = []
a_db_project: Project
for a_db_project in db_projects:
self._collection.projects.append(a_db_project)
prj_licenses.add(cast(LicenseEnum, a_db_project.license))
for a_sample in a_db_project.all_samples:
sample_id = a_sample.orig_id
# Sanity check: sample orig_id must be unique in the collection
if sample_id in samples_per_project:
problems.append("Sample with orig_id %s is in both '%s'(#%d) and '%s'(#%d)" %
(sample_id, samples_per_project[sample_id].title,
samples_per_project[sample_id].projid,
a_db_project.title, a_db_project.projid))
else:
samples_per_project[sample_id] = a_db_project
# Set self to most restrictive of all licenses
max_restrict = max([DataLicense.RESTRICTION[a_prj_lic] for a_prj_lic in prj_licenses])
self._collection.license = DataLicense.BY_RESTRICTION[max_restrict]
# TODO: Default creators using classification history in DB. Knowing that it's partial.
# Report (brutally) problems
assert len(problems) == 0, "\n".join(problems)
def __init__(self, session: Session, project_ids: ProjectIDListT):
self.instrument_names: List[InstrumentIDT] = []
qry: Query = session.query(Acquisition.instrument).distinct()
qry = qry.join(Sample).join(Project)
if len(project_ids) > 0:
qry = qry.filter(Project.projid.in_(project_ids))
qry = qry.order_by(Acquisition.instrument)
self.instrument_names = [nm for nm, in qry.all()
if nm] # Filter NULL & empty strings
def get_one(session: Session, coll_id: CollectionIDT) -> Optional['CollectionBO']:
"""
Find a Collection by its ID and return it, None if not found.
"""
# Find the collection
ret = session.query(Collection).get(coll_id)
if ret is None:
return None
else:
return CollectionBO(ret).enrich()
def get_all_object_ids(cls, session: Session, acquis_id: AcquisitionIDT,
classif_ids: Optional[ClassifIDListT] = None) \
-> List[int]:
qry: Query = session.query(ObjectHeader.objid)
qry = qry.join(
Acquisition,
and_(ObjectHeader.acquisid == Acquisition.acquisid,
Acquisition.acquisid == acquis_id))
if classif_ids is not None:
qry = qry.filter(ObjectHeader.classif_id.in_(classif_ids))
return [an_id for an_id in qry.all()]
def get_sums_by_taxon(cls, session: Session, acquis_id: AcquisitionIDT) \
-> Dict[ClassifIDT, int]:
res: ResultProxy = session.execute(
"SELECT o.classif_id, count(1)"
" FROM obj_head o "
" WHERE o.acquisid = :acq"
" GROUP BY o.classif_id", {"acq": acquis_id})
return {
int(classif_id): int(cnt)
for (classif_id, cnt) in res.fetchall()
}
def get_for_update(cls, session: Session, job_id: JobIDT) -> 'JobBO':
"""
Return a single JobBO. If used in a 'with' context, the session will commit on context exit.
Note: it's not only the Job which will be committed, but the _whole_ session.
"""
job = session.query(Job).get(job_id)
if job is None:
raise ValueError
job.updated_on = datetime.now()
ret = JobBO(job)
ret._session = session
return ret
def get_sums_by_taxon(cls, session: Session, acquis_id: AcquisitionIDT) \
-> Dict[ClassifIDT, int]:
sql = text("SELECT o.classif_id, count(1)"
" FROM obj_head o "
" WHERE o.acquisid = :acq "
" AND o.classif_id IS NOT NULL "
" AND o.classif_qual = 'V'"
" GROUP BY o.classif_id")
res: Result = session.execute(sql, {"acq": acquis_id})
return {
int(classif_id): int(cnt)
for (classif_id, cnt) in res.fetchall()
}
def get_preferences_per_project(session: Session, user_id: int,
project_id: int, key: str) -> Any:
"""
Get a preference, for given project and user. Keys are not standardized (for now).
"""
current_user: User = session.query(User).get(user_id)
prefs_for_proj: UserPreferences = current_user.preferences_for_projects.filter_by(
project_id=project_id).first()
if prefs_for_proj:
all_prefs_for_proj = json.loads(prefs_for_proj.json_prefs)
else:
all_prefs_for_proj = dict()
return all_prefs_for_proj.get(key, "")
def set_preferences_per_project(session: Session, user_id: int,
project_id: int, key: str, value: Any):
"""
Set preference for a key, for given project and user. The key disappears if set to empty string.
"""
current_user: User = session.query(User).get(user_id)
prefs_for_proj: UserPreferences = current_user.preferences_for_projects.filter_by(
project_id=project_id).first()
if prefs_for_proj:
all_prefs_for_proj = json.loads(prefs_for_proj.json_prefs)
else:
prefs_for_proj = UserPreferences()
prefs_for_proj.project_id = project_id
prefs_for_proj.user_id = user_id
session.add(prefs_for_proj)
all_prefs_for_proj = dict()
all_prefs_for_proj[key] = value
if value == '':
del all_prefs_for_proj[key]
prefs_for_proj.json_prefs = json.dumps(all_prefs_for_proj)
logger.info("for %s and %d: %s", current_user.name, project_id,
prefs_for_proj.json_prefs)
session.commit()
def create(session: Session, title: str,
project_ids: ProjectIDListT) -> CollectionIDT:
"""
Create using minimum fields.
"""
# Find the collection
db_coll = Collection()
db_coll.title = title
session.add(db_coll)
session.flush() # to get the collection ID
bo_coll = CollectionBO(db_coll)
bo_coll.set_composing_projects(session, project_ids)
session.commit()
return bo_coll.id
def read_from_db(self, session: Session, prj_id: ProjectIDT):
"""
Read the project topology from DB.
"""
qry: Query = session.query(Sample)
qry = qry.join(Acquisition).join(Process).join(ObjectHeader)
qry = qry.filter(Sample.projid == prj_id)
qry = qry.with_entities(Sample.orig_id, Acquisition.orig_id,
Process.orig_id, ObjectHeader.objid)
sam_orig_id: str
acq_orig_id: str
prc_orig_id: str
objid: ObjectIDT
for sam_orig_id, acq_orig_id, prc_orig_id, objid in qry.all():
# Get/create acquisitions for this sample
objs_for_acquisition = self.add_association(
sam_orig_id, acq_orig_id)
# Store twin process
if prc_orig_id is not None:
self.acquisition_child[acq_orig_id] = prc_orig_id
# Store objects for acquisition
objs_for_acquisition.add(objid)
def __init__(self, session: Session, project_ids: ProjectIDListT):
qry: Query = session.query(Acquisition.instrument)
qry = qry.join(Sample).join(Project)
# TODO: WTF WTF just for adding a column to the select
qry = qry.add_columns(
text(Project.__table__.name + "." +
Project.__table__.c.projid.name))
# Below SQLAlchemy complains
# qry = qry.add_columns(Project.projid)
if len(project_ids) > 0:
qry = qry.filter(Project.projid.in_(project_ids))
qry = qry.distinct()
instruments_by_proj: Dict[ProjectIDT, Set[InstrumentIDT]] = {}
instrument_names = set()
for ins_name, projid in qry.all():
if ins_name:
instruments_by_proj.setdefault(projid, set()).add(ins_name)
instrument_names.add(ins_name)
else:
pass # Filter NULL & empty strings
self.by_project = instruments_by_proj
self.instrument_names = sorted(list(instrument_names))
def __init__(self, session: Session, acquisition_id: AcquisitionIDT):
super().__init__(session)
self.acquis = session.query(Acquisition).get(acquisition_id)
def __init__(self, session: Session, sample_id: SampleIDT):
super().__init__(session)
self.sample = session.query(Sample).get(sample_id)
def get_one(cls, session: Session, job_id: JobIDT) -> Optional['JobBO']:
job = session.query(Job).get(job_id)
if job is None:
return None
return JobBO(job)
def __init__(self, session: Session, process_id: ProcessIDT):
super().__init__(session)
self.process = session.query(Process).get(process_id)