j = 0.0
correct = 0.0
half_correct = 0.0
all_correct = 0.0
cutoff = 0.0
for s_full, s_disamb in sents:
c += 1
if c % 100 == 0:
print c, '/', len(sentences)
s_name = s_full.replace(
LFG_DIR + 'parses/' + PARSES_VERSION + '/', '')
s_prob = LFG_DIR + 'stochastic/' + parse_dir + wf + '/' + s_name[:
-3] + PROB_SUFFIX
data = None
try:
data = read_prolog_file(s_prob, True)
except IOError:
#print 'IOError'
continue
if not data.text:
#print 'Oh noes!'
continue
text = data.text
solutions = data.num_solutions
#print '*', len(data.opts), data.num_solutions
weighted_constrs = {}
for opts in data.opts:
weight = 0
for o in opts:
if o in data.choices_weights:
weight -= data.choices_weights[o]
os.system('mkdir ' + tex_dir)
os.system('cp /home/kasia/Dokumenty/Robota/LFG/parses/avm.sty ' + tex_dir)
processed = 0
for sent_name in sentences:
sent_count += 1
if sent_name in skip_sets[sents_dir]:
#print 'SKIP:', sent_name
continue
print '-----', sent_count, '/', num_sents, sent_name
path = '/home/kasia/Dokumenty/Robota/LFG/parses/%s/' % sents_dir + sent_name + file_suff #dla disamb trzeba dis
try:
open(path, 'r')
except:
print 'CAN\'T OPEN:', path
continue
prolog_data = read_prolog_file(path, max_solutions=20000)
if prolog_data is None:
continue
num_solutions, text = prolog_data.num_solutions, prolog_data.text
#num_solutions, text, constraint_list, phis, other_list, choices_list, nodes, root_id, opts, all_lines = read_prolog_file(path)
sentences_solutions[sent_name] = num_solutions
if (num_solutions < 1):
continue
processed += 1
if (num_solutions > 20000):
continue
num_consistent = 0.0
count = 0
#====
skladnica_path = '/home/kasia/Dokumenty/Robota/Świgrowe/zrobione%s/' % skladnica_v + \
sent_name.replace('Skladnica--FULL', 'NKJP_1M') \
MAX_SOLUTIONS = 8
sentences = []
with open (sents_file, 'r') as f:
for l in f.readlines():
if not (l.startswith('#')):
s_name = re.match(r'.*(Skladnica.*-s).*', l).group(1)
sentences.append(s_name)
i = 0
for s_name in sentences[3917:3930]:
i += 1
if (i % 200 == 0):
print i
full_forest = sents_dir + s_name + '.pl'
solutions = 0
with open(full_forest, 'r') as forest:
l = forest.readline()
while l:
if l.startswith('\t\'statistics'):
solutions = int(l[15:l.find(' solutions')])
break
l = forest.readline()
if solutions > 7 and solutions <= MAX_SOLUTIONS:
print full_forest
prolog_data = read_prolog_file(full_forest, quiet=True, max_solutions=MAX_SOLUTIONS)
print prolog_data.text
for n_id, node in prolog_data.nodes.items():
print node.short()
print prolog_data.nodes
print 'Checking solutions sums'
for ranked_forest in rfs:
i += 1
if (i % 250 == 0):
print i, '/', n
j = ranked_forest.find('-s-')
s_name = ranked_forest[:(j + 2)]
if not s_name in solutions_sum:
#print ranked_forest
continue
ranked_forests[s_name].append(ranked_forest)
#TYLKO NA RAZIE TU PRZERYWAMY, ZEBY BYLO SZYBCIEJ
#continue
prolog_data = read_prolog_file(ranked_dir + ranked_forest,
quiet=True,
max_solutions=1000000)
if prolog_data is None:
continue
#print ranked_forest, len(prolog_data.opts)
for o in prolog_data.opts:
try:
fs, eqs_dict = build_fs([
c for c, l_no in prolog_data.constraint_list
if c.cond.intersection(o)
])
except:
if not s_name in exc_forests:
print 'EXCEPTION'
print o
exc_forests[s_name] = set()
exc_forests = dict()
print 'Checking solutions sums'
for ranked_forest in rfs:
i += 1
if (i % 250 == 0):
print i, '/', n
j = ranked_forest.find('-s-')
s_name = ranked_forest[:(j + 2)]
if not s_name in solutions_sum:
#print ranked_forest
continue
ranked_forests[s_name].append(ranked_forest)
#TYLKO NA RAZIE TU PRZERYWAMY, ZEBY BYLO SZYBCIEJ
#continue
prolog_data = read_prolog_file(ranked_dir + ranked_forest, quiet=True, max_solutions=1000000)
if prolog_data is None:
continue
#print ranked_forest, len(prolog_data.opts)
for o in prolog_data.opts:
try:
fs, eqs_dict = build_fs([c for c, l_no in prolog_data.constraint_list if c.cond.intersection(o)])
except:
if not s_name in exc_forests:
print 'EXCEPTION'
print o
exc_forests[s_name] = set()
raise
exceptions[s_name] = traceback.format_exc()
exc_forests[s_name].add(ranked_forest)
solutions_sum[s_name] += len(prolog_data.opts)
#avg_first_rank = 0.0
c = 0
j = 0.0
correct = 0.0
half_correct = 0.0
all_correct = 0.0
cutoff = 0.0
for s_full, s_disamb in sents:
c += 1
if c % 100 == 0:
print c, '/', len(sentences)
s_name = s_full.replace(LFG_DIR + 'parses/' + PARSES_VERSION + '/', '')
s_prob = LFG_DIR + 'stochastic/' + parse_dir + wf + '/' + s_name[:-3] + PROB_SUFFIX
data = None
try:
data = read_prolog_file(s_prob, True)
except IOError:
#print 'IOError'
continue
if not data.text:
#print 'Oh noes!'
continue
text = data.text
solutions = data.num_solutions
#print '*', len(data.opts), data.num_solutions
weighted_constrs = {}
for opts in data.opts:
weight = 0
for o in opts:
if o in data.choices_weights:
weight -= data.choices_weights[o]
os.system('mkdir ' + tex_dir)
os.system('cp /home/kasia/Dokumenty/Robota/LFG/parses/avm.sty ' + tex_dir)
processed = 0
for sent_name in sentences:
sent_count += 1
if sent_name in skip_sets[sents_dir]:
#print 'SKIP:', sent_name
continue
print '-----', sent_count, '/', num_sents, sent_name
path = '/home/kasia/Dokumenty/Robota/LFG/parses/%s/' % sents_dir + sent_name + file_suff #dla disamb trzeba dis
try:
open(path, 'r')
except:
print 'CAN\'T OPEN:', path
continue
prolog_data = read_prolog_file(path, max_solutions=20000)
if prolog_data is None:
continue
num_solutions, text = prolog_data.num_solutions, prolog_data.text
#num_solutions, text, constraint_list, phis, other_list, choices_list, nodes, root_id, opts, all_lines = read_prolog_file(path)
sentences_solutions[sent_name] = num_solutions
if (num_solutions < 1):
continue
processed += 1
if (num_solutions > 20000):
continue
num_consistent = 0.0
count = 0
#====
skladnica_path = '/home/kasia/Dokumenty/Robota/Świgrowe/zrobione%s/' % skladnica_v + \
sent_name.replace('Skladnica--FULL', 'NKJP_1M') \
