def compare(gold_text, test_text, out_dict, error_counts, language='english'):
""" Compares two trees in text form.
This checks for empty trees and mismatched numbers
of words.
"""
gold_text = gold_text.strip()
test_text = test_text.strip()
if len(gold_text) == 0:
mprint("No gold tree", out_dict, ['out', 'err'])
return
elif len(test_text) == 0:
mprint("Not parsed", out_dict, ['out', 'err'])
return
gold_tree = read_tree(gold_text, out_dict, 'gold')
test_tree = read_tree(test_text, out_dict, 'test')
if gold_tree is None or test_tree is None:
mprint("Not parsed, but had output", out_dict,
['out', 'err', 'init_errors'])
return
mprint(
render_tree.text_coloured_errors(test_tree, gold_tree).strip(),
out_dict, 'init_errors')
gold_words = gold_tree.word_yield()
test_words = test_tree.word_yield()
if len(test_words.split()) != len(gold_words.split()):
mprint("Sentence lengths do not match...", out_dict, ['out', 'err'])
mprint("Gold: " + gold_words.__repr__(), out_dict, ['out', 'err'])
mprint("Test: " + test_words.__repr__(), out_dict, ['out', 'err'])
return
return compare_trees(gold_tree, test_tree, out_dict, error_counts,
language)
def compare(gold_text, test_text, out_dict, error_counts, language='english'):
""" Compares two trees in text form.
This checks for empty trees and mismatched numbers
of words.
"""
gold_text = gold_text.strip()
test_text = test_text.strip()
if len(gold_text) == 0:
mprint("No gold tree", out_dict, ['out', 'err'])
return
elif len(test_text) == 0:
mprint("Not parsed", out_dict, ['out', 'err'])
return
gold_tree = read_tree(gold_text, out_dict, 'gold')
test_tree = read_tree(test_text, out_dict, 'test')
if gold_tree is None or test_tree is None:
mprint("Not parsed, but had output", out_dict, ['out', 'err', 'init_errors'])
return
mprint(render_tree.text_coloured_errors(test_tree, gold_tree).strip(), out_dict, 'init_errors')
gold_words = gold_tree.word_yield()
test_words = test_tree.word_yield()
if len(test_words.split()) != len(gold_words.split()):
mprint("Sentence lengths do not match...", out_dict, ['out', 'err'])
mprint("Gold: " + gold_words.__repr__(), out_dict, ['out', 'err'])
mprint("Test: " + test_words.__repr__(), out_dict, ['out', 'err'])
return
return compare_trees(gold_tree, test_tree, out_dict, error_counts, language)
def compare_trees(gold_tree,
test_tree,
out_dict,
error_counts,
language='english'):
""" Compares two trees. """
init_errors = test_tree.get_errors(gold_tree)
error_count = len(init_errors)
mprint("%d Initial errors" % error_count, out_dict, 'out')
iters, path = greedy_search(gold_tree, test_tree, language)
mprint("%d on fringe, %d iterations" % iters, out_dict, 'out')
if path is not None:
mprint(test_tree.__repr__(), out_dict, 'test_trees')
mprint(gold_tree.__repr__(), out_dict, 'gold_trees')
for tree in path[1:]:
mprint(
str(tree[2]) + " Error:" + tree[1]['classified_type'],
out_dict, 'out')
if len(path) > 1:
for tree in path:
mprint("Step:" + tree[1]['classified_type'], out_dict, 'out')
error_counts[tree[1]['classified_type']].append(tree[2])
mprint(tree[1].__repr__(), out_dict, 'out')
mprint(
render_tree.text_coloured_errors(tree[0],
gold=gold_tree).strip(),
out_dict, 'out')
else:
mprint("no path found", out_dict, 'out')
mprint("", out_dict, ['out', 'err'])
def compare_trees(gold_tree, test_tree, out_dict, error_counts, language='english'):
""" Compares two trees. """
init_errors = test_tree.get_errors(gold_tree)
error_count = len(init_errors)
mprint("%d Initial errors" % error_count, out_dict, 'out')
iters, path = greedy_search(gold_tree, test_tree, language)
mprint("%d on fringe, %d iterations" % iters, out_dict, 'out')
if path is not None:
mprint(test_tree.__repr__(), out_dict, 'test_trees')
mprint(gold_tree.__repr__(), out_dict, 'gold_trees')
for tree in path[1:]:
mprint(str(tree[2]) + " Error:" + tree[1]['classified_type'], out_dict, 'out')
if len(path) > 1:
for tree in path:
mprint("Step:" + tree[1]['classified_type'], out_dict, 'out')
error_counts[tree[1]['classified_type']].append(tree[2])
mprint(tree[1].__repr__(), out_dict, 'out')
mprint(render_tree.text_coloured_errors(tree[0], gold=gold_tree).strip(), out_dict, 'out')
else:
mprint("no path found", out_dict, 'out')
mprint("", out_dict, ['out', 'err'])
