def test_asType(self):
import rdflib
from idlib.formats import rdf as _bind_rdf
o = idlib.Orcid(self.ids[0])
nt = o.asType(rdflib.URIRef)
# still haven't decided if this is a good idea or not
assert str(o) != str(nt), 'string representation of streams should not match id ???'
def contributor_orcid(self, value):
# FIXME use schema
self._error_on_na(value)
value, _j = self._deatag(value)
v = value.replace(' ',
'').strip().rstrip() # ah the rando tabs in a csv
if not v:
return
if v.startswith('http:'):
v = v.replace('http:', 'https:', 1)
if not (v.startswith('ORCID:') or v.startswith('https:')):
v = v.strip()
if not len(v):
return
elif v == '0': # FIXME ? someone using strange conventions ...
return
elif len(v) != 19:
msg = f'orcid wrong length {value!r} {self._path.as_posix()!r}'
if self.addError(idlib.Orcid._id_class.OrcidLengthError(msg)):
logd.error(msg)
return
v = 'ORCID:' + v
else:
if v.startswith('https:'):
_, numeric = v.rsplit('/', 1)
elif v.startswith('ORCID:'):
_, numeric = v.rsplit(':', 1)
if not len(numeric):
return
elif len(numeric) != 19:
msg = f'orcid wrong length {value!r} {self._path.as_posix()!r}'
if self.addError(idlib.Orcid._id_class.OrcidLengthError(msg)):
logd.error(msg)
return
try:
#log.debug(f"{v} '{self.path}'")
orcid = idlib.Orcid(v) # XXX possible encapsulation issue
if not orcid.identifier.checksumValid: # FIXME should not handle this with ifs ...
# FIXME json schema can't do this ...
msg = f'orcid failed checksum {value!r} {self._path.as_posix()!r}'
if self.addError(
idlib.Orcid._id_class.OrcidChecksumError(msg)):
logd.error(msg)
return
yield orcid
except (OntId.BadCurieError,
idlib.Orcid._id_class.OrcidMalformedError) as e:
msg = f'orcid malformed {value!r} {self._path.as_posix()!r}'
if self.addError(idlib.Orcid._id_class.OrcidMalformedError(msg)):
logd.error(msg)
yield value
def __new__(cls, something):
if '10.' in something:
if 'http' in something and 'doi.org' not in something:
pass # probably a publisher uri that uses the handle
else:
return idlib.Doi(something)
if 'orcid' in something:
return idlib.Orcid(something)
if '/ror.org/' in something or something.startswith('ror:'):
return idlib.Ror(something)
if 'protocols.io' in something:
return idlib.Pio(something)
return oq.OntId(something)
return OntTerm(something) # use the better local version of OntTerm
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