def main(args, classify):
init.argcheck(args, 4, 4, 'Identify errors in parser output',
'<gold> <test> <prefix_for_output_files>')
# Output setup
out_dict = {
'out': sys.stdout,
'err': sys.stderr,
'gold_trees': sys.stdout,
'test_trees': sys.stdout,
'error_counts': sys.stdout
}
prefix = args[3]
out_dict['out'] = open(prefix + '.out', 'w')
out_dict['err'] = open(prefix + '.log', 'w')
out_dict['gold_trees'] = open(prefix + '.gold_trees', 'w')
out_dict['test_trees'] = open(prefix + '.test_trees', 'w')
out_dict['error_counts'] = open(prefix + '.error_counts', 'w')
out_dict['init_errors'] = open(prefix + '.init_errors', 'w')
init.header(args, out_dict.values())
# Classification
print >> out_dict['out'], "Printing tree transformations"
print >> out_dict['err'], "Printing tree transformations"
gold_in = open(args[1])
test_in = sys.stdin if args[2] == '-' else open(args[2])
sent_no = 0
error_counts = defaultdict(lambda: [])
while True:
sent_no += 1
gold_text = gold_in.readline()
test_text = test_in.readline()
if gold_text == '' and test_text == '':
print >> out_dict['err'], "End of both input files"
break
elif gold_text == '':
print >> out_dict['err'], "End of gold input"
break
elif test_text == '':
print >> out_dict['err'], "End of test input"
break
print >> out_dict['out'], "Sentence {}:".format(sent_no)
print >> out_dict['err'], "Sentence {}:".format(sent_no)
print >> out_dict['init_errors'], "Sentence {}:".format(sent_no)
compare(gold_text.strip(), test_text.strip(), out_dict, error_counts,
classify)
print >> out_dict['init_errors'], "\n"
# Results
counts_to_print = []
for error in error_counts:
if error == 'UNSET init':
continue
counts_to_print.append(
(len(error_counts[error]), sum(error_counts[error]), error))
counts_to_print.sort(reverse=True)
for error in counts_to_print:
print >> out_dict['error_counts'], "{} {} {}".format(*error)
def main(args, classify):
init.argcheck(args, 4, 4, 'Identify errors in parser output', '<gold> <test> <prefix_for_output_files>')
# Output setup
out_dict = {
'out': sys.stdout,
'err': sys.stderr,
'gold_trees': sys.stdout,
'test_trees': sys.stdout,
'error_counts': sys.stdout
}
prefix = args[3]
out_dict['out'] = open(prefix + '.out', 'w')
out_dict['err'] = open(prefix + '.log', 'w')
out_dict['gold_trees'] = open(prefix + '.gold_trees', 'w')
out_dict['test_trees'] = open(prefix + '.test_trees', 'w')
out_dict['error_counts'] = open(prefix + '.error_counts', 'w')
out_dict['init_errors'] = open(prefix + '.init_errors', 'w')
init.header(args, out_dict.values())
# Classification
print >> out_dict['out'], "Printing tree transformations"
print >> out_dict['err'], "Printing tree transformations"
gold_in = open(args[1])
test_in = sys.stdin if args[2] == '-' else open(args[2])
sent_no = 0
error_counts = defaultdict(lambda: [])
while True:
sent_no += 1
gold_text = gold_in.readline()
test_text = test_in.readline()
if gold_text == '' and test_text == '':
print >> out_dict['err'], "End of both input files"
break
elif gold_text == '':
print >> out_dict['err'], "End of gold input"
break
elif test_text == '':
print >> out_dict['err'], "End of test input"
break
print >> out_dict['out'], "Sentence {}:".format(sent_no)
print >> out_dict['err'], "Sentence {}:".format(sent_no)
print >> out_dict['init_errors'], "Sentence {}:".format(sent_no)
compare(gold_text.strip(), test_text.strip(), out_dict, error_counts, classify)
print >> out_dict['init_errors'], "\n"
# Results
counts_to_print = []
for error in error_counts:
if error == 'UNSET init':
continue
counts_to_print.append((len(error_counts[error]), sum(error_counts[error]), error))
counts_to_print.sort(reverse=True)
for error in counts_to_print:
print >> out_dict['error_counts'], "{} {} {}".format(*error)
def main():
# TODO, shift to a uniform style of module documentation, then just skip all of this!
desc = __doc__.split("\n")
arg_info = desc[1]
further_desc = "\n".join(desc[2:])
desc = desc[0]
init.argcheck(sys.argv, 3, 7, desc, arg_info, further_desc)
out = open(sys.argv[1] + ".table", "w")
log = open(sys.argv[1] + ".table.log", "w")
init.header(sys.argv, log)
data = get_data(sys.argv[2])
mapping = {}
if len(sys.argv) > 3:
mapping = get_mapping(sys.argv[3])
system_order_file = None
if len(sys.argv) > 5:
system_order_file = sys.argv[5]
error_order_file = None
if len(sys.argv) > 4:
error_order_file = sys.argv[4]
system_order, error_order, extra_info = get_order(data, system_order_file, error_order_file)
print >> log, "System order:", system_order
print >> log, "Error order:", error_order
print_top(error_order, extra_info, out)
print_data(system_order, error_order, data, mapping, extra_info, out)
print_bottom(error_order, extra_info, out)
if len(sys.argv) < 6:
system_out = open(sys.argv[1] + ".table.system_order", "w")
print_system_order(system_order, extra_info, system_out)
system_out.close()
if len(sys.argv) < 5:
error_out = open(sys.argv[1] + ".table.error_order", "w")
print_error_order(error_order, error_out)
error_out.close()
if len(sys.argv) < 4:
mapping_out = open(sys.argv[1] + ".table.name_mapping", "w")
print_mapping(mapping, data, mapping_out)
mapping_out.close()
out.close()
log.close()
return multifile_process(path, call)
if __name__ == '__main__':
formats = {
'bart': read_bart,
'cherrypicker': read_cherrypicker,
'conll': read_conll,
'ims': read_ims,
### 'opennlp': read_opennlp,
'reconcile': read_reconcile,
### 'relaxcor': read_relaxcor,
'stanford_xml': read_stanford_xml,
'stanford': read_stanford,
'uiuc': read_uiuc
}
init.argcheck(sys.argv, 5, 6, "Translate a system output into the CoNLL format", "<prefix> <[{}]> <dir | file> <gold dir>".format(','.join(formats.keys())))
out = open(sys.argv[1] + '.out', 'w')
log = open(sys.argv[1] + '.log', 'w')
init.header(sys.argv, log)
auto_src = sys.argv[3]
gold_src = sys.argv[4]
if sys.argv[2] not in formats:
print "Invalid format. Valid options are:"
print '\n'.join(formats.keys())
sys.exit(1)
auto, gold = formats[sys.argv[2]](auto_src, gold_src)
for doc in auto:
for part in auto[doc]:
#!/usr/bin/env python # -*- coding: utf-8 -*- # vim: set ts=2 sw=2 noet: import sys from nlp_util import coreference_reading, coreference_rendering, coreference, init if __name__ == '__main__': init.argcheck(sys.argv, 4, 5, "Print coreference resolution errors", "<prefix> <gold_dir> <test> [resolve span errors first? T | F]") auto = coreference_reading.read_conll_coref_system_output(sys.argv[3]) gold = coreference_reading.read_conll_matching_files(auto, sys.argv[2]) out_cluster_errors = open(sys.argv[1] + '.cluster_errors', 'w') out_cluster_context = open(sys.argv[1] + '.cluster_context', 'w') out_cluster_missing = open(sys.argv[1] + '.cluster_missing', 'w') out_cluster_extra = open(sys.argv[1] + '.cluster_extra', 'w') out_mention_list = open(sys.argv[1] + '.mention_list', 'w') out_mention_text = open(sys.argv[1] + '.mention_text', 'w') out_files = [out_cluster_errors, out_cluster_context, out_cluster_missing, out_cluster_extra, out_mention_list, out_mention_text] init.header(sys.argv, out_files) for function, outfile in [ (coreference_rendering.print_mention_text, out_mention_text), (coreference_rendering.print_mention_list, out_mention_list), (coreference_rendering.print_cluster_errors, out_cluster_errors),
coreference_rendering.print_conll_style_part(out['error: extra entity'], text, auto_mentions_extra_entity, doc_name, part_name)
coreference_rendering.print_conll_style_part(out['error: merge'], text, auto_mentions_merge, doc_name, part_name)
coreference_rendering.print_conll_style_part(out['error: missing mention'], text, auto_mentions_missing_mention, doc_name, part_name)
coreference_rendering.print_conll_style_part(out['error: missing entity'], text, auto_mentions_missing_entity, doc_name, part_name)
coreference_rendering.print_conll_style_part(out['error: extra mention prog'], text, auto_mentions_extra_mention_prog, doc_name, part_name)
coreference_rendering.print_conll_style_part(out['error: extra entity prog'], text, auto_mentions_extra_entity_prog, doc_name, part_name)
coreference_rendering.print_conll_style_part(out['error: merge prog'], text, auto_mentions_merge_prog, doc_name, part_name)
coreference_rendering.print_conll_style_part(out['error: missing mention prog'], text, auto_mentions_missing_mention_prog, doc_name, part_name)
coreference_rendering.print_conll_style_part(out['error: missing entity prog'], text, auto_mentions_missing_entity_prog, doc_name, part_name)
return errors
if __name__ == '__main__':
# Process params
init.argcheck(sys.argv, 4, 5, "Print coreference resolution errors", "<output_prefix> <gold_dir> <test_file> [remove singletons? T | F (default is True)]")
remove_singletons = True
if len(sys.argv) == 5 and sys.argv[-1] == 'F':
remove_singletons = False
out = {
'out': open(sys.argv[1] + '.classified.detailed', 'w'),
'properties': open(sys.argv[1] + '.classified.properties', 'w'),
'short out': open(sys.argv[1] + '.classified', 'w'),
'summary': open(sys.argv[1] + '.summary', 'w'),
'system output': open(sys.argv[1] + '.system', 'w'),
'gold': open(sys.argv[1] + '.gold', 'w'),
'error: original': open(sys.argv[1] + '.corrected.none', 'w'),
'error: span mismatch': open(sys.argv[1] + '.corrected.span_errors', 'w'),
'error: split': open(sys.argv[1] + '.corrected.confused_entities', 'w'),
'error: extra mention': open(sys.argv[1] + '.corrected.extra_mention', 'w'),
'error: extra entity': open(sys.argv[1] + '.corrected.extra_entity', 'w'),
if __name__ == '__main__':
formats = {
'bart': read_bart,
'cherrypicker': read_cherrypicker,
'conll': read_conll,
'ims': read_ims,
### 'opennlp': read_opennlp,
'reconcile': read_reconcile,
### 'relaxcor': read_relaxcor,
'stanford_xml': read_stanford_xml,
'stanford': read_stanford,
'uiuc': read_uiuc
}
init.argcheck(
sys.argv, 5, 5, "Translate a system output into the CoNLL format",
"<prefix> <[{}]> <dir | file> <gold dir>".format(','.join(
formats.keys())))
out = open(sys.argv[1] + '.out', 'w')
log = open(sys.argv[1] + '.log', 'w')
init.header(sys.argv, log)
auto_src = sys.argv[3]
gold_src = sys.argv[4]
if sys.argv[2] not in formats:
print "Invalid format. Valid options are:"
print '\n'.join(formats.keys())
sys.exit(1)
auto, gold = formats[sys.argv[2]](auto_src, gold_src)
for doc in auto:
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# vim: set ts=2 sw=2 noet:
import sys
try:
from nlp_util import init, coreference_reading
except ImportError:
raise Exception("Remember to either install nlp_util or set up a symlink to the nlp_util directory")
if __name__ == '__main__':
init.argcheck(sys.argv, 3, 3, "Print conll text", "<prefix> <data>")
prefix = sys.argv[1]
data = coreference_reading.read_all(sys.argv[2])
for doc in data:
for part in data[doc]:
text = data[doc][part]['text']
filename = '__'.join(doc.split('/') + [part])
out = open(prefix + filename, 'w')
for line in text:
print >> out, ' '.join(line)
