def match(s, p): # match a single pattern, return the noun phrase
#print "in func: match string with pattern"
#TODO: for future improvement, using getMentions instead
t = '0'
if p.find('_') == 0:
t = 'h'
p = ' ' + p[2:] + ' '
else:
t = 't'
p = ' ' + p[0:-2] + ' '
#print t, p
if p in s:
#print "pattern found"
if t == 't':
x = s.split(p)[1]
ps = npc.postag(x)
np = npc.extFirstNP(ps)
return np
else:
x = s.split(p)[0]
ps = npc.postag(x)
np = npc.extLastNP(ps)
return np
else:
return ""
def get_patterns(s, ins):
# preceding patterns empty room, optionally or non-optionally
# print "preceding: [nouns] - verbs - adjectives/prepositions/determiners"
# print "preceding: nouns&adjectives - adjectives/prepositions/determiners"
# print "following: verbs - noun phrases/preposition"
ps = npc.postag(s)
ip = npc.postag(ins)
ms = get_mentions(ps, ip)
#print len(ms)
patterns = []
p = 0
l = len(ps) - 1
#print len(ms)
for i in range(len(ms)):
b = ms[i]
e = b + len(ip)
if i < len(ms) - 1:
l = ms[i + 1] - 1
else:
l = len(ps) - 1
p1 = ext1rule(ps, p, b - 1)
p2 = ext2rule(ps, p, b - 1)
p3 = ext3rule(ps, e, l)
if len(p1) != 0:
patterns.append(p1 + ' _')
#print "p1", p1
if len(p2) != 0:
patterns.append(p2 + ' _')
#print "p2", p2
if len(p3) != 0:
patterns.append('_ ' + p3)
#print "p3", p3
# update p and l
p = e
return patterns