parsing/statistics.py at master · cdg720/parsing · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
from bllipparser import Tree
from numpy import mean, median
from subprocess import Popen, PIPE, call
import gzip
import sys

def compute_dist(gold, trees):
	dist = [0.,] * 50
	for g, t in zip(gold, trees):
		argmax = 0
		score = evaluate(g, t[0])
		for i in xrange(1, len(t)):
			x = evaluate(g, t[i])
			if x > score:
				argmax = i
				score = x
		dist[argmax] += 1
	return [x / len(trees) for x in dist]

def create_separate_files(trees, gold):
	ind = 0
	# for ts, g in zip(trees, gold):
	# 	ind += 1
	# 	f = gzip.open('/home/dc65/Documents/research/self-training/code/tmp/dev/' + str(ind) + '/pred.gz', 'wb')
	# 	h = gzip.open('/home/dc65/Documents/research/self-training/code/tmp/dev/' + str(ind) + '/gold.gz', 'wb')
	# 	for t in ts:
	# 		f.write(str(t) + '\n')
	# 		h.write(str(g) + '\n')
	# 	f.close()
	# 	h.close()

def evaluate(sent1, sent2): # sent1: gold?
	score = 0
	f = open('/home/dc65/Documents/research/self-training/code/tmp/a', 'w')
	f.write(str(sent1) + '\n')
	f.flush()
	f.close()
	f = open('/home/dc65/Documents/research/self-training/code/tmp/b', 'w')
	f.write(str(sent2) + '\n')
	f.flush()
	f.close()
	out, err = None, None
	p = Popen(['/home/dc65/Documents/tools/bllip_parser/bllip-parser-master/SParseval/src/sparseval', '-v', '-h', '/home/dc65/Documents/tools/bllip_parser/bllip-parser-master/SParseval/headInfo.txt', '-p', '/home/dc65/Documents/tools/bllip_parser/bllip-parser-master/SParseval/SPEECHPAR.prm', '/home/dc65/Documents/research/self-training/code/tmp/a', '/home/dc65/Documents/research/self-training/code/tmp/b'], stdout=PIPE, stderr=PIPE)
	out, err = p.communicate()
	call(['rm', '/home/dc65/Documents/research/self-training/code/tmp/a'])
	call(['rm', '/home/dc65/Documents/research/self-training/code/tmp/b'])
	try:
		return float(out.split('\n')[13].split()[-1])
	except:
		print sent1
		print sent2
		print out
		sys.exit(0)

def read_gold(path):
	if path.endswith('.gz'):
		f = gzip.open(sys.argv[2], 'rb')
	else:
		f = open(sys.argv[2])
	gold = [Tree(x) for x in f.read().splitlines()]
	return gold

def read_nbest(path):
	if path.endswith('.gz'):
		f = gzip.open(path, 'rb')
	else:
		f = open(path, 'r')
	tmp = f.read().splitlines()
	x = tmp[::2]
	y = tmp[1::2]

	trees, scores, tmp1, tmp2 = [], [], [], []
	prev = 1
	for tree, score in zip(x, y):
		score = float(score)
		if score > prev: # new
			if tmp1:
				trees.append(tmp1)
				scores.append(tmp2)
				tmp1, tmp2 = [Tree(tree)], [score]
			prev = score
		else:
			prev = score
			tmp1.append(Tree(tree))
			tmp2.append(score)
	if tmp1:
		trees.append(tmp1)
		scores.append(tmp2)
	return trees, scores

def main():
	if len(sys.argv) != 3:
		print 'usage: python statistics.py 50best gold'
		sys.exit(0)

	trees, scores = read_nbest(sys.argv[1])
	gold = read_gold(sys.argv[2])

	# create_separate_files(trees, gold)
	# best_scores = []
	# bins = [0,] * 10
	# for score in scores:
	# 	best_scores.append(score[0])
	# 	bins[int(score[0] * 10)] += 1

	#print best_scores
	# print mean(best_scores)
	# print sum(best_scores)
	# print median(best_scores)
	# print max(best_scores)
	# print min(best_scores)
	# x = [float(x) / len(best_scores) for x in bins]
	# print '\t'.join([str(xx * 10) + '-' + str((xx+1) * 10) for xx in xrange(10)])
	# print '\t'.join(['%.4f' for xx in xrange(10)]) % (tuple(x))
	# empirical_dist = compute_dist(gold, trees)
	# print 'empirical distribution:'
	# print empirical_dist

	# x = [[] for x in xrange(50)]
	# for ss in scores:
	# 	for i in xrange(len(ss)):
	# 		x[i].append(ss[i])
	# predicted_dist = [mean(xx) for xx in x]
	# print 'rerakning parser distribution:'
	# print predicted_dist

#main()