def _process(self, contributor):
# get member if we can find them
he = dat.HasErrors(pipeline_stage=self.__class__.__name__ + '.data')
if 'name' in contributor and 'first_name' in contributor:
name = contributor['name']
if ';' in name:
msg = f'Bad symbol in name {name!r}'
he.addError(msg)
logd.error(msg)
fn = contributor['first_name']
ln = contributor['last_name']
if ' ' in fn:
fn, mn = fn.split(' ', 1)
contributor['middle_name'] = mn
contributor['first_name'] = fn
if ' ' in ln:
msg = f'Malformed last_name {ln!r}'
he.addError(msg)
logd.error(msg)
ln = ln.replace(' ', '-')
failover = f'{fn}-{ln}'
member = self.member(fn, ln)
if member is not None:
userid = OntId('https://api.blackfynn.io/users/' + member.id)
contributor['blackfynn_user_id'] = userid
else:
member = None
failover = 'no-orcid-no-name'
log.warning(f'No name!' + lj(contributor))
orcid = None
if 'contributor_orcid_id' in contributor:
orcid = contributor['contributor_orcid_id']
if type(orcid) == str and 'orcid.org' in orcid:
orcid = OrcidId(orcid) # FIXME reloading from json
if isinstance(orcid, OrcidId):
s = orcid
else: # it's not an orcid or its a bad orcid
orcid = None
if orcid is None:
if member is not None:
s = userid
else:
log.debug(lj(contributor))
s = OntId(self.dsid + '/contributors/' + failover)
contributor['id'] = s
he.embedErrors(contributor)
def transformed(self):
try:
return self._transformer.data
except (exc.FileTypeError, exc.NoDataError, exc.BadDataError) as e:
# these errors mostly happen after __init__ now
# since they are properly pipelined
data = {}
he = dat.HasErrors(pipeline_stage=self.__class__.__name__ + '.transformer')
logd.exception(e) # FIXME isn't this were we should accumulate errors?
he.addError(e, path=self.path)
he.embedErrors(data)
return data
def _transformer(self):
try:
return self.data_transformer_class(self.path, schema_version=self.schema_version)
except (exc.FileTypeError, exc.NoDataError, exc.BadDataError) as e:
class NoData: # FIXME
data = {}
t = f'No data for {self.path}'
he = dat.HasErrors(pipeline_stage=self.__class__.__name__ + '._transformer')
logd.exception(e) # FIXME isn't this were we should accumulate errors?
he.addError(e, path=self.path)
he.embedErrors(NoData.data)
return NoData
def _process(self, contributor):
# get member if we can find them
he = dat.HasErrors(pipeline_stage=self.__class__.__name__ + '.data')
if 'contributor_name' in contributor and 'first_name' in contributor:
name = contributor['contributor_name']
if ';' in name:
msg = f'Bad symbol in name {name!r}'
he.addError(msg)
logd.error(msg)
fn = contributor['first_name']
ln = contributor['last_name']
if ' ' in fn:
fn, mn = fn.split(' ', 1)
mn, _mn = mn.rstrip('.'), mn
if mn != _mn:
he.addError(f'Middle initials don\'t need periods :) {name!r}',
logfunc=logd.error)
contributor['middle_name'] = mn
contributor['first_name'] = fn
if ' ' in ln:
msg = f'Malformed last_name {ln!r}'
he.addError(msg)
logd.error(msg)
ln = ln.replace(' ', '-')
failover = f'{fn}-{ln}'
member = self.member(fn, ln)
if member is not None:
userid = OntId('https://api.blackfynn.io/users/' + member.id)
contributor['blackfynn_user_id'] = userid
else:
member = None
failover = 'no-orcid-no-name'
log.warning(f'No name!' + lj(contributor))
orcid = None
if 'contributor_orcid_id' in contributor:
orcid = contributor['contributor_orcid_id']
if type(orcid) == str and 'orcid.org' in orcid:
orcid = idlib.Orcid(orcid) # FIXME reloading from json
if isinstance(orcid, idlib.Orcid):
s = orcid
else: # it's not an orcid or its a bad orcid
orcid = None
if orcid is None:
if member is not None:
s = userid
else:
log.debug(lj(contributor))
s = OntId(self.dsid + '/contributors/' + failover)
contributor['id'] = s
he.embedErrors(contributor)
# lifting + adding
if 'contributor_affiliation' in contributor:
ca = contributor['contributor_affiliation']
maybe_ror = self.lifters.affiliations(ca)
if maybe_ror is not None:
contributor['affiliation'] = maybe_ror