def fragment_lcia(self,
fragment,
quantity_ref,
scenario=None,
refresh=False,
**kwargs):
if scenario is None:
scenario = '1'
lcia_q = self.get_lcia_quantity(quantity_ref)
endpoint = 'scenarios/%s/%s/%s/lciaresults' % (scenario, fragment,
lcia_q.external_ref)
lcia_r = self._archive.get_endpoint(endpoint, cache=False)
if lcia_r is None or (isinstance(lcia_r, list)
and all(i is None for i in lcia_r)):
res = LciaResult(lcia_q, scenario=scenario)
return res
res = LciaResult(lcia_q, scenario=lcia_r.pop('scenarioID'))
total = lcia_r.pop('total')
for component in lcia_r['lciaScore']:
self.add_lcia_component(res, component)
self.check_total(res.total(), total)
return res
def test_catalog_ref(self):
"""
Create an LCIA result using a catalog ref as quantity
:return:
"""
qty = CatalogRef('fictitious.origin', 'dummy_ext_ref', entity_type='quantity', Name='Test Quantity')
res = LciaResult(qty)
self.assertEqual(res.total(), 0.0)
def retrieve_lcia_scores(self, process_uuid, rf_uuid, quantities=None):
"""
This function retrieves LCIA scores from an Ecospold02 file and stores them as characterizations in
an LcFlow entity corresponding to the *first* (and presumably, only) reference intermediate flow
Only stores cfs for quantities that exist locally.
:param process_uuid:
:param rf_uuid:
:param quantities: list of quantity entities to look for (defaults to all local lcia_methods)
:return: a dict of quantity uuid to score
"""
if quantities is None:
quantities = [l for l in self.entities_by_type('quantity') if l.is_lcia_method()]
import time
start_time = time.time()
print('Loading LCIA results for %s_%s' % (process_uuid, rf_uuid))
o = self.objectify(process_uuid, rf_uuid)
self._print('%30.30s -- %5f' % ('Objectified', time.time() - start_time))
p = self._create_process_entity(o)
rf = self._grab_reference_flow(o, rf_uuid)
exch = ExchangeValue(p, rf, 'Output', value=1.0)
tags = dict()
for q in quantities:
if 'Method' in q.keys():
if q['Name'] in tags:
raise KeyError('Name collision %s' % q['Name'])
tags[q['Name']] = q
results = LciaResults(p)
for char in find_tag(o, 'flowData').getchildren():
if 'impactIndicator' in char.tag:
m = char.impactMethodName.text
c = char.impactCategoryName.text
i = char.name.text
v = float(char.get('amount'))
my_tag = ', '.join([m, c, i])
if my_tag in tags:
q = tags[my_tag]
result = LciaResult(q)
cf = Characterization(rf, q, value=v, location=p['SpatialScope'])
result.add_score(p.get_uuid(), exch, cf, p['SpatialScope'])
results[q.get_uuid()] = result
self._print('%30.30s -- %5f' % ('Impact scores collected', time.time() - start_time))
return results
def retrieve_lcia_scores(self, filename, quantities=None):
"""
This function retrieves LCIA scores from an Ecospold02 file and stores them as characterizations in
an LcFlow entity corresponding to the *first* (and presumably, only) reference intermediate flow
Only stores cfs for quantities that exist locally.
:param filename:
:param quantities: list of quantity entities to look for (defaults to self.quantities())
:return: a dict of quantity uuid to score
"""
if quantities is None:
quantities = self.quantities()
import time
start_time = time.time()
print('Loading LCIA results from %s' % filename)
o = self.objectify(filename)
self._print('%30.30s -- %5f' % ('Objectified', time.time() - start_time))
p = self._create_process_entity(o)
rf = self._grab_reference_flow(o, spold_reference_flow(filename))
exch = ExchangeValue(p, rf, 'Output', value=1.0)
tags = dict()
for q in quantities:
if 'Method' in q.keys():
if q['Name'] in tags:
raise KeyError('Name collision %s' % q['Name'])
tags[q['Name']] = q
results = LciaResults(p)
for char in find_tag(o, 'flowData')[0].getchildren():
if 'impactIndicator' in char.tag:
m = char.impactMethodName.text
c = char.impactCategoryName.text
i = char.name.text
v = float(char.get('amount'))
my_tag = ', '.join([m, c, i])
if my_tag in tags:
q = tags[my_tag]
result = LciaResult(q)
cf = Characterization(rf, q, value=v, location=p['SpatialScope'])
result.add_score(p.get_uuid(), exch, cf, p['SpatialScope'])
results[q.get_uuid()] = result
self._print('%30.30s -- %5f' % ('Impact scores collected', time.time() - start_time))
return results
def lcia(self, quantity, ref_flow=None, scenario=None, flowdb=None):
result = LciaResult(quantity, scenario)
result.add_component(self.get_uuid(), entity=self)
for ex in self.allocated_exchanges(scenario or ref_flow):
if not ex.flow.has_characterization(quantity):
if flowdb is not None:
if quantity in flowdb.known_quantities():
factor = flowdb.lookup_single_cf(ex.flow, quantity, self['SpatialScope'])
if factor is None:
ex.flow.add_characterization(quantity)
else:
ex.flow.add_characterization(factor)
factor = ex.flow.factor(quantity)
result.add_score(self.get_uuid(), ex, factor, self['SpatialScope'])
return result
def frag_flow_lcia(fragmentflows,
quantity_ref,
scenario=None,
refresh=False,
ignore_uncached=True):
"""
Recursive function to compute LCIA of a traversal record contained in a set of Fragment Flows.
:param fragmentflows:
:param quantity_ref:
:param scenario: necessary if any remote traversals are required
:param refresh: whether to refresh the LCIA CFs
:param ignore_uncached: [True] whether to allow zero scores for un-cached, un-computable fragments
:return:
"""
result = LciaResult(quantity_ref)
for ff in fragmentflows:
if ff.term.is_null:
continue
node_weight = ff.node_weight
if node_weight == 0:
continue
try:
v = ff.term.score_cache(quantity=quantity_ref,
refresh=refresh,
ignore_uncached=ignore_uncached)
except SubFragmentAggregation:
# if we were given interior fragments, recurse on them. otherwise ask remote.
if len(ff.subfragments) == 0:
v = ff.term.term_node.fragment_lcia(quantity_ref,
scenario=scenario,
refresh=refresh)
else:
v = frag_flow_lcia(ff.subfragments,
quantity_ref,
refresh=refresh)
if v.is_null:
continue
if ff.term.direction == ff.fragment.direction:
# if the directions collide (rather than complement), the term is getting run in reverse
node_weight *= -1
result.add_summary(ff.fragment.uuid, ff, node_weight, v)
return result
def lcia(self, process, ref_flow, quantity_ref, refresh=False, **kwargs):
"""
Antelope v1 doesn't support or even have any knowledge of process reference-flows. this is a somewhat
significant design flaw. well, no matter. each antelopev1 process must therefore represent an allocated single
operation process that has an unambiguous reference flow. This is a problem to solve on the server side;
for now we just ignore the ref_flow argument.
If the quantity ref is one of the ones natively known by the antelope server-- i.e. if it is a catalog ref whose
origin matches the origin of the current archive-- then it is trivially used. Otherwise, the lcia call reduces
to obtaining the inventory and computing LCIA locally.
:param process:
:param ref_flow:
:param quantity_ref:
:param refresh:
:param kwargs:
:return:
"""
lcia_q = self.get_lcia_quantity(quantity_ref)
endpoint = '%s/%s/lciaresults' % (process, lcia_q.external_ref)
lcia_r = self._archive.get_endpoint(endpoint, cache=False)
res = LciaResult(lcia_q, scenario=lcia_r.pop('scenarioID'))
total = lcia_r.pop('total')
if len(lcia_r['lciaScore']) > 1:
raise AntelopeV1Error(
'Process LCIA result contains too many components\n%s' %
process)
component = lcia_r['lciaScore'][0]
cum = component['cumulativeResult']
self.check_total(cum, total)
if 'processes/%s' % component['processID'] != process:
raise AntelopeV1Error('Reference mismatch: %s begat %s' %
(process, component['processID']))
self.add_lcia_component(res, component)
self.check_total(res.total(), total)
return res
def fg_lcia(self, process_ref, quantities=None, dist=3, scenario=None, **kwargs):
"""
:param process_ref:
:param quantities: defaults to foreground lcia quantities
:param dist: [1] how far afield to search for cfs (see CLookup.find() from flowdb)
:param scenario: (not presently used) - some day the flow-quantity database will be scenario-sensitive
:return:
"""
if self._catalog.fg is None:
print('Missing a foreground!')
return None
if not self._catalog.is_loaded(0):
self._catalog.load(0)
if not self._catalog.is_loaded(process_ref.index):
self._catalog.load(process_ref.index)
exch = self.db.filter_exch(process_ref, elem=True, **kwargs)
qs = self._prep_quantities(quantities)
results = LciaResults(process_ref.entity())
for q in qs:
q_result = LciaResult(q)
if q in self.db.known_quantities():
for x in exch:
if not x.flow.has_characterization(q):
# look locally
local = self[0][x.flow.get_uuid()]
if local is not None and local.has_characterization(q):
cf = local.factor(q)
x.flow.add_characterization(cf)
else:
cf = self.db.lookup_single_cf(x.flow, q, dist=dist, location=process_ref['SpatialScope'])
if cf is None:
x.flow.add_characterization(q)
else:
x.flow.add_characterization(cf)
self.add_to_foreground(x.flow)
fac = x.flow.factor(q)
q_result.add_score(process_ref.id, x, fac, process_ref['SpatialScope'])
results[q.get_uuid()] = q_result
return results
def lcia(self, quantity, ref_flow=None):
if not quantity.is_entity:
# only works for quantity refs-- in other words, always works
return quantity.do_lcia(self.inventory(ref_flow=ref_flow),
locale=self['SpatialScope'])
else:
result = LciaResult(quantity)
result.add_component(self.get_uuid(), entity=self)
for ex in self.inventory(ref_flow):
factor = ex.flow.factor(quantity)
result.add_score(self.get_uuid(), ex, factor,
self['SpatialScope'])
return result
def score_cache(self,
quantity=None,
ignore_uncached=False,
refresh=False,
**kwargs):
if quantity is None:
return self._score_cache
if quantity.uuid in self._score_cache and refresh is False:
return self._score_cache[quantity.uuid]
else:
try:
res = self.compute_unit_score(quantity,
refresh=refresh,
**kwargs)
except UnCachedScore:
if ignore_uncached:
res = LciaResult(quantity)
else:
raise
self._score_cache[quantity.uuid] = res
return res
def do_lcia(self, quantity, inventory, locale='GLO', refresh=False, debug=False, **kwargs):
"""
takes a quantity and an exchanges generator; returns an LciaResult for the given quantity.
For now, does NOT pre-load quantity LCIA methods. that is a catalog action. the Qdb doesn't do catalog stuff.
I am open to re-thinking it, though.
:param quantity:
:param inventory: generates exchanges
:param locale: ['GLO']
:param refresh: [False] whether to rewrite characterization factors from the database
:param debug: [False] print extra information to screen
:param kwargs: just quell_biogenic_co2 for the moment
:return: an LciaResult whose components are the flows of the exchanges
"""
q = self[quantity.link]
q_ind = self._get_q_ind(q)
_is_quiet = self._quiet
if debug:
self._quiet = False
self._print('q_ind: %d' % q_ind)
r = LciaResult(q)
for x in inventory:
if refresh or not x.flow.has_characterization(q):
try:
factor = self.convert(flow=x.flow, query_q_ind=q_ind, locale=locale, **kwargs)
except MissingCompartment:
self._print('Missing compartment %s; abandoning this exchange' % x.flow['Compartment'])
continue
except ConversionReferenceMismatch:
print('Mismatch %s' % x)
factor = None
if factor is not None:
self._print('factor %g %s' % (factor, x))
x.flow.add_characterization(q, value=factor, overwrite=refresh)
else:
self._print('factor NONE %s' % x)
x.flow.add_characterization(q)
if x.flow.cf(q) is not None:
r.add_component(x.flow.external_ref, entity=x.flow)
fac = x.flow.factor(q)
fac.set_natural_direction(self.c_mgr)
r.add_score(x.flow.external_ref, x, fac, locale)
self._quiet = _is_quiet
return r
def lcia(x, y, z):
local = self[0][y.get_uuid()]
if local is not y:
raise AmbiguousReference('!!! this should be in foreground')
return LciaResult.from_cfs(x, self.db.factors_for_flow(local, z))
def add_lcia_score(self, quantity, score, scenario=None):
res = LciaResult(quantity, scenario=scenario)
res.add_summary(self._parent.uuid, self._parent, 1.0, score)
self._score_cache.add(res)
def compute_unit_score(self, quantity_ref, **kwargs):
"""
four different ways to do this.
0- we are a subfragment-- throw exception: use subfragment traversal results contained in the FragmentFlow
1- parent is bg: ask catalog to give us bg_lcia (process or fragment)
2- get fg lcia for unobserved exchanges
If
:param quantity_ref:
:return:
"""
if self.is_subfrag:
if self.descend:
return LciaResult(
quantity_ref
) # null result for subfragments that are explicitly followed
else:
raise SubFragmentAggregation # to be caught
if self.is_bg:
if self.is_fg:
# surprisingly not inconsistent! in the current pre-ContextRefactor world, this is how we are handling
# cached-LCIA-score nodes
raise UnCachedScore('fragment: %s\nquantity: %s' %
(self._parent, quantity_ref))
elif self.is_frag:
# need bg_lcia method for FragmentRefs
# this is probably not currently supported
return self.term_node.bg_lcia(
lcia_qty=quantity_ref,
ref_flow=self.term_flow.external_ref,
**kwargs)
try:
locale = self.term_node['SpatialScope']
except KeyError:
locale = 'GLO'
try:
res = quantity_ref.do_lcia(self._unobserved_exchanges(),
locale=locale,
**kwargs)
except PrivateArchive:
if self.is_bg:
print(
'terminations.compute_unit_score UNTESTED for private bg archives!'
)
res = self.term_node.bg_lcia(
lcia_qty=quantity_ref,
ref_flow=self.term_flow.external_ref,
**kwargs)
else:
res = self.term_node.fg_lcia(
quantity_ref,
ref_flow=self.term_flow.external_ref,
**kwargs)
print(
'terminations.compute_unit_score UNTESTED for private fg archives!'
)
# res.set_scale(self.inbound_exchange_value)
return res
