def test_select_rows(single_item_profile):
p = itsdb.ItsdbProfile(single_item_profile)
# assert list(itsdb.select_rows(None, p.read_table('item'))) == [['0', 'The dog barks.']]
assert list(
itsdb.select_rows(['i-id', 'i-input'],
p.read_table('item'))) == [['0', 'The dog barks.']]
assert list(
itsdb.select_rows(['item:i-id', 'parse:parse-id'],
p.join('item', 'parse'))) == [['0', '0']]
def rows(p):
if p.exists('p-result'):
return p.select('p-result', ('i-id', 'p-id', 'mrs'))
elif p.exists('result') and p.exists('parse'):
return itsdb.select_rows(
('parse:i-id', 'result:result-id', 'result:mrs'),
p.join('parse', 'result'))
else:
raise Exception('Invalid profile: ' + str(p.root))
def _select(projection, tables, condition, ts, mode, cast):
table = _select_from(tables, None, ts)
table = _select_projection(projection, table, ts)
table = _select_where(condition, table, ts)
# finally select the relevant columns from the joined table
if projection == '*':
if len(tables) == 1:
projection = [f.name for f in ts.relations[tables[0]]]
else:
projection = []
for t in tables:
projection.extend(t + ':' + f.name
for f in ts.relations[t])
return itsdb.select_rows(projection, table, mode=mode, cast=cast)
def select(args):
"""
Select data from [incr tsdb()] profiles.
"""
in_profile = _prepare_input_profile(args['PROFILE'],
filters=args['--filter'],
applicators=args['--apply'])
if args['--join']:
tbl1, tbl2 = map(str.strip, args['--join'].split(','))
rows = in_profile.join(tbl1, tbl2, key_filter=True)
# Adding : is just for robustness. We need something like
# :table:col@table@col, but may have gotten table:col@table@col
if not args['DATASPEC'].startswith(':'):
args['DATASPEC'] = ':' + args['DATASPEC']
table, cols = itsdb.get_data_specifier(args['DATASPEC'])
else:
table, cols = itsdb.get_data_specifier(args['DATASPEC'])
rows = in_profile.read_table(table, key_filter=True)
for row in itsdb.select_rows(cols, rows, mode='row'):
print(row)
def pred_strings(prof):
# need to join parse and result to get i-id to mrs mapping
rows = prof.join('parse', 'result')
cols = ('parse:i-id', 'result:result-id', 'result:mrs')
for i_id, r_id, mrs in itsdb.select_rows(cols, rows):
if r_id != '0':
continue
mrs = simplemrs.loads_one(mrs)
preds = []
valency = {}
nmz_locs = set()
# pre-scan
for ep in mrs.eps():
if ep.pred.short_form() == 'nominalization' and ep.cfrom != -1:
nmz_locs.add(ep.lnk)
valency[ep.nodeid] = extract_valency(ep)
# extract and simplify predicates
for ep in mrs.eps():
normpred = ep.pred.short_form()
pred = ep.pred.string
# skip omitted preds
if normpred == 'nominalization' or normpred.endswith('unknown'):
continue
# combine named with CARG value
if normpred == 'named':
pred = 'nmd_"{}"'.format(str(ep.carg or ''))
# normalize verbs
if ep.pred.pos == 'v':
# mark if nominalized
if ep.lnk in nmz_locs:
pred = 'nmz_' + pred
# add argument info
pred += '@' + valency[ep.nodeid]
preds.append(pred)
yield (int(i_id), ' '.join(preds))