def get_undone(kb,beam_size): #split finished and todo may be be the same tree
newkb=[]
string_d=defaultdict(int)
for nl in kb: #todo,delete the same tree
string=''.join([node.show() for node in nl[0]])
if string_d[string]==0:
newkb.append(nl)
string_d[string]+=1
kb=list(newkb)
kb=sorted(kb,key=lambda x:-x[-1]) #sorted by pcfg
kb=kb[:beam_size] #best beam_size
done=[item for item in kb if len(item[0])==1] #todo 假如是单枝的可能还要继续,done
undone=[item for item in kb if len(item[0])!=1]
return done,undone
def split_sen(fn,resf):
lines=[x.strip().decode('utf8') for x in file(fn)]#[:1]
res=[]
i=0
for line in lines:
if len(line.strip())>0:
t=read_tree(line)
tl=split_main(t)
for node in tl:
## while not node.isleaf and len(node.son)==1:
## node=node.son[0]
res.append(node.show())
res.append('')
i+=1
if i%1000==0:
print i
write_file(resf,res)
