def test_event_decrease():
test_file = os.path.join(path_here, 'sofia_event_decreased.json')
sp = sofia.process_json_file(test_file)
assert len(sp.statements) == 1, sp.statements
stmt = sp.statements[0]
assert isinstance(stmt, Event), stmt
assert stmt.delta.polarity == -1, stmt.delta
assert stmt.concept.name == 'rainfall', stmt.concept
# Test extraction filter
sp = sofia.process_json_file(test_file, extract_filter={'influence'})
assert len(sp.statements) == 0
def test_influence_event_polarity():
test_file = os.path.join(path_here, 'sofia_infl_polarities.json')
sp = sofia.process_json_file(test_file)
assert len(sp.statements) == 1, sp.statements
stmt = sp.statements[0]
assert isinstance(stmt, Influence)
assert stmt.subj.delta['polarity'] == 1, stmt.subj.delta
assert stmt.obj.delta['polarity'] == -1, stmt.obj.delta
def test_event_decrease():
test_file = os.path.join(path_here, 'sofia_event_decreased.json')
sp = sofia.process_json_file(test_file)
assert len(sp.statements) == 1, sp.statements
stmt = sp.statements[0]
assert isinstance(stmt, Event), stmt
assert stmt.delta['polarity'] == -1, stmt.delta
assert stmt.concept.name == 'rainfall', stmt.concept
def test_influence_event_polarity():
test_file = os.path.join(path_here, 'sofia_infl_polarities.json')
sp = sofia.process_json_file(test_file)
assert len(sp.statements) == 1, sp.statements
stmt = sp.statements[0]
assert isinstance(stmt, Influence)
assert stmt.subj.delta.polarity == 1, stmt.subj.delta
assert stmt.obj.delta.polarity == -1, stmt.obj.delta
def test_event_decrease():
test_file = os.path.join(path_here, 'sofia_event_decreased.json')
sp = sofia.process_json_file(test_file)
assert len(sp.statements) == 1, sp.statements
stmt = sp.statements[0]
assert isinstance(stmt, Event), stmt
assert stmt.delta.polarity == -1, stmt.delta
assert stmt.concept.name == 'rainfall', stmt.concept
def test_process_json():
test_file = os.path.join(path_here, 'sofia_test.json')
sp = sofia.process_json_file(test_file)
assert len(sp.statements) == 2
assert isinstance(sp.statements[0], Influence)
assert sp.statements[0].subj.concept.name == 'rainfall'
assert sp.statements[0].obj.concept.name == 'floods'
assert len(sp.statements[0].evidence) == 1, len(sp.statements[0].evidence)
assert isinstance(sp.statements[1], Event)
assert sp.statements[1].concept.name == 'inflation'
assert isinstance(sp.statements[1].context, WorldContext)
assert sp.statements[1].context.time.text == '28, JULY, 2016'
assert sp.statements[1].context.geo_location.name == 'South Sudan'
def test_process_json():
test_file = os.path.join(path_here, 'sofia_test.json')
sp = sofia.process_json_file(test_file)
assert len(sp.statements) == 2
assert isinstance(sp.statements[0], Influence)
assert sp.statements[0].subj.concept.name == 'rainfall'
assert sp.statements[0].obj.concept.name == 'floods'
assert len(sp.statements[0].evidence) == 1, len(sp.statements[0].evidence)
assert isinstance(sp.statements[1], Event)
assert sp.statements[1].concept.name == 'inflation'
assert isinstance(sp.statements[1].context, WorldContext)
assert sp.statements[1].context.time.text == '28, JULY, 2016'
assert sp.statements[1].context.geo_location.name == 'South Sudan'
def test_compositional_grounding():
test_file = os.path.join(path_here, 'sofia_test_comp_no_causal.json')
sp = sofia.process_json_file(test_file, grounding_mode='compositional')
assert len(sp.statements) == 2
assert isinstance(sp.statements[0], Event)
assert isinstance(sp.statements[1], Event)
assert sp.statements[0].concept.db_refs['TEXT'] == 'wheat', \
sp.statements[0].concept.db_refs
assert isinstance(sp.statements[0].concept.db_refs['WM'], list)
assert isinstance(sp.statements[0].concept.db_refs['WM'][0], tuple), \
sp.statements[0].concept.db_refs['WM']
assert sp.statements[1].concept.db_refs['TEXT'] == 'cereal'
assert isinstance(sp.statements[1].concept.db_refs['WM'], list)
assert isinstance(sp.statements[1].concept.db_refs['WM'][0], tuple)
for reader in readers:
version = reader_versions[grounding][reader]
pattern = '*' if reader != 'sofia' \
else ('*_new' if grounding == 'compositional' else '*_old')
fnames = glob.glob('/Users/ben/data/dart/%s/%s/%s' %
(reader, version, pattern))
print('Found %d files for %s' % (len(fnames), reader))
for fname in tqdm.tqdm(fnames):
if reader == 'eidos':
pp = eidos.process_json_file(fname, grounding_mode=grounding)
elif reader == 'hume':
pp = hume.process_jsonld_file(fname, grounding_mode=grounding)
elif reader == 'cwms':
pp = cwms.process_ekb_file(fname, grounding_mode=grounding)
elif reader == 'sofia':
pp = sofia.process_json_file(fname, grounding_mode=grounding)
doc_id = os.path.basename(fname)[:32]
for stmt in pp.statements:
for ev in stmt.evidence:
if 'provenance' not in ev.annotations:
ev.annotations['provenance'] = [{
'document': {
'@id': doc_id
}
}]
else:
prov = ev.annotations['provenance'][0]['document']
prov['@id'] = doc_id
stmts += pp.statements
if grounding == 'compositional':
validate_grounding_format(stmts)