def get_new_nodel(nl,flg): # tag model, node list, tag model(if any)
# get new nodel list by the crf tag
# return: [(nodel, diff_node)]
# todo find kbest one by one changed and record the diff,done
nodel_l=[]
i=0
while i<len(flg):
if flg[i]!=3:
new_nl=nl[:i]
if flg[i]==0:#####BBBBB######
a=node()
a.son=[nl[i],nl[i+1]]
a.head=nl[i].head
a.head_pos=nl[i].head_pos
#new_nl=nl[:i]
new_nl.append(a)
new_nl+=nl[i+2:]
#nodel_l.append(a)
i+=1
elif flg[i]==1:####EEEEE###
a=node()
a.son=[nl[i],nl[i+1]]
a.head=nl[i+1].head
a.head_pos=nl[i+1].head_pos
#new_nl=nl[:i]
new_nl.append(a)
new_nl+=nl[i+2:]
#nodel_l.append(a)
i+=1
elif flg[i]==2:############SSSS######
a=node()
a.son=[nl[i]]
a.head=nl[i].head
a.head_pos=nl[i].head_pos
#new_nl=nl[:i]
new_nl.append(a)
new_nl+=nl[i+1:]
#nodel_l.append(a)
else:
print '?'
if flg[i]==2:
# fts=get_tag_feature_final(ts,ind,type_tag)
fts=get_tag_feature_final(nl,i,1)
nodel_l.append((new_nl,i,fts))# new node list, diff son
else:
# todo,attention!!!!!!<<i-1>>
fts=get_tag_feature_final(nl,i-1,2)
nodel_l.append((new_nl,i-1,fts))# new node list, diff son
#nodel_l[-1].tag=ntag
else:
pass
#nodel_l.append(nl[i])
i+=1 ################todo,notify
return nodel_l
def multi_2_binary_trick(t,tag):#all leaf
if len(t.son)<=2:
return t
elif t.son[-1].tag=='ETC' or t.son[-1].word in end_punc:#**等,**。
tt=deepcopy(t)
tt.son=tt.son[:-1]
new_son=multi_2_binary_trick(tt,tag)
t.son=[new_son,t.son[-1]]
return t
elif t.son[0].tag==t.son[-1].tag=='PU' and\
len(t.son[0].word)==len(t.son[-1].word)==1 and\
abs(ord(t.son[0].word)-ord(t.son[-1].word))==1:#"**",<>
tt=deepcopy(t)
tt.son=tt.son[1:-1]
new_son1=multi_2_binary_trick(tt,tag)
new_son2=node(atag=tag+'*')
new_son2.son=[new_son1,t.son[-1]]
t.son=[t.son[0],new_son2]####t.son[0]!!!
return t
t=check_item_complete(t,tag)###每一项都是完整的一项了
if len(t.son)>2:
t=get_son_join(t.son,tag)
t.tag=t.tag.rstrip('*')
###check n1 cc n2
return t
def get_son_join(sonl,tag):
new_node=node(atag=tag+'*')
new_node.son=[sonl[-2],sonl[-1]]
if len(sonl)==2:
return new_node
else:
sonll=sonl[:-2]
sonll.append(new_node)
return get_son_join(sonll,tag)
def binary_right(rightson,sons,tag):
#check?
new_son=[]
new_son=[sons[0],sons[1]]
new_node=node(atag=tag+'*')
new_node.son=new_son
if len(rightson)==0:
return new_node
else:
return binary_right(rightson[1:],[new_node,rightson[0]],tag)##right!!
def binary_left(leftson,sons,tag):
#check?
new_son=[]
new_son=[sons[0],sons[1]]
new_node=node(atag=tag+'*')
new_node.son=new_son
if len(leftson)==0:
return new_node
else:
return binary_left(leftson[:-1],[leftson[-1],new_node],tag)
def get_pieces(res):
pieces=[]
tmpl=[]
for w in res:
#####
nnode=node(aleaf=True,aword=w[0],atag=w[1],apos=w[1])
if w[-1] in ['B','S']:###
if len(tmpl)!=0:
pieces.append(tmpl)
tmpl=[]
if w[-1]=='E':
tmpl.append(nnode)
pieces.append(tmpl)
tmpl=[]
elif w[-1]=='S':
tmpl.append(nnode)
pieces.append(tmpl)
tmpl=[]
else:
tmpl.append(nnode)
if len(tmpl)!=0:
pieces.append(tmpl)
return pieces