def synonyms_from_ecospold_exchange(exch):
"""
Ecospold exchanges: synonyms are Name, CAS Number, and ', '-separated contents of synonym tags.
Care must be taken not to split on ',' as some chemical names include commas
:param exch:
:return: set of synonyms (stripped)
"""
syns = set()
name = str(exch['name'])
syns.add(name)
cas = exch.get('casNumber')
if cas is not None:
syns.add(cas)
synonym_tag = find_tag(exch, 'synonym')
if len(synonym_tag) == 1:
# parse the comma-separated list
if bool(re.search('etc\.', str(synonym_tag[0]))):
syns.add(str(synonym_tag[0]).strip())
else:
for x in str(synonym_tag[0]).split(', '):
_add_syn_if(x, syns)
else:
# multiple entries- allow embedded comma-space
for syn in synonym_tag:
_add_syn_if(str(syn), syns)
return name, syns
def synonyms_from_ilcd_flow(flow):
"""
ILCD flow files have long synonym blocks at the top. They also have a CAS number and a basename.
:param flow:
:return:
"""
ns = find_ns(flow.nsmap, 'Flow')
syns = set()
name = grab_flow_name(flow, ns=ns)
syns.add(name)
uid = str(find_common(flow, 'UUID')[0]).strip()
syns.add(uid)
cas = str(find_tag(flow, 'CASNumber', ns=ns)[0]).strip()
if cas != '':
syns.add(cas)
for syn in find_common(flow, 'synonyms'):
for x in str(syn).split(';'):
if x.strip() != '' and x.strip().lower() != 'wood':
syns.add(x.strip())
return name, syns, uid
def get_ecospold_exchanges(archive=ECOSPOLD, prefix='datasets', file=ES_FILE):
E = EcospoldV2Archive(archive, prefix=prefix)
o = E.objectify(file)
return find_tag(o, 'elementaryExchange')
def _create_process(self, filename):
"""
Extract dataset object from XML file
:param filename:
:return:
"""
o = self._get_objectified_entity(filename)
rf = None # reference flow
flowlist = []
for exch in o.dataset.flowData.getchildren():
f = self._create_flow(exch)
if hasattr(exch, 'outputGroup'):
d = 'Output'
if exch.outputGroup == 0:
assert rf is None, "Multiple reference flows found!"
rf = f
elif hasattr(exch, 'inputGroup'):
d = 'Input'
else:
raise DirectionlessExchangeError
local_q = self._create_quantity(exch.get("unit"))
v = float(exch.get('meanValue')) # returns none if missing
if local_q is not f.reference_entity:
v = v / f.cf(local_q)
flowlist.append((f, d, v))
p_meta = o.dataset.metaInformation.processInformation
n = p_meta.referenceFunction.get('name')
u = self._key_to_id(n)
try_p = self[u]
if try_p is not None:
p = try_p
assert p.entity_type == 'process', "Expected process, found %s" % p.entity_type
else:
# create new process
g = p_meta.geography.get('location')
stt = {'begin': str(find_tag(p_meta, 'startDate')[0]), 'end': str(find_tag(p_meta, 'endDate')[0])}
c = p_meta.referenceFunction.get('generalComment')
cls = [p_meta.referenceFunction.get('category'), p_meta.referenceFunction.get('subCategory')]
p = LcProcess(u, Name=n, Comment=c, SpatialScope=g, TemporalScope=stt,
Classifications=cls)
p.set_external_ref(n)
if rf is None:
rx = None
else:
rx = p.add_reference(rf, 'Output')
for flow, f_dir, val in flowlist:
self._print('Exch %s [%s] (%g)' % (flow, f_dir, val))
p.add_exchange(flow, f_dir, reference=None, value=val, add_dups=True)
self.add(p)
return p
